(19)
(11)EP 2 407 932 B1

(12)EUROPEAN PATENT SPECIFICATION

(45)Mention of the grant of the patent:
27.11.2019 Bulletin 2019/48

(21)Application number: 10750606.5

(22)Date of filing:  12.03.2010
(51)International Patent Classification (IPC): 
G06T 7/00(2017.01)
G06K 9/50(2006.01)
G06K 9/46(2006.01)
G06K 9/38(2006.01)
(86)International application number:
PCT/JP2010/001770
(87)International publication number:
WO 2010/103850 (16.09.2010 Gazette  2010/37)

(54)

IMAGE IDENTIFIER EXTRACTION DEVICE

VORRICHTUNG ZUR EXTRAKTION VON BILDIDENTIFIKATOREN

DISPOSITIF D'EXTRACTION D'IDENTIFIANTS D'IMAGES


(84)Designated Contracting States:
AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO SE SI SK SM TR

(30)Priority: 13.03.2009 JP 2009061023
14.04.2009 JP 2009097862

(43)Date of publication of application:
18.01.2012 Bulletin 2012/03

(73)Proprietor: NEC Corporation
Tokyo 108-8001 (JP)

(72)Inventors:
  • IWAMOTO, Kota
    Tokyo 108-8001 (JP)
  • OAMI, Ryoma
    Tokyo 108-8001 (JP)

(74)Representative: Betten & Resch 
Patent- und Rechtsanwälte PartGmbB Maximiliansplatz 14
80333 München
80333 München (DE)


(56)References cited: : 
WO-A1-2007/148264
JP-A- 2005 031 901
JP-A- 2007 293 438
US-A- 5 436 653
JP-A- 2004 192 555
JP-A- 2006 268 825
US-A- 5 034 991
US-A1- 2006 204 103
  
  • KOTA IWAMOTO ET AL: "Response to the Call for Proposals on Video Signature Tools", 87. MPEG MEETING; 2-2-2009 - 6-2-2009; LAUSANNE; (MOTION PICTURE EXPERT GROUP OR ISO/IEC JTC1/SC29/WG11),, no. M16171, 29 January 2009 (2009-01-29), XP030044768,
  • DAIDI ZHONG ET AL: "Face Image Retrieval System Using TFV and Combination of Subimages", 28 June 2007 (2007-06-28), ADVANCES IN VISUAL INFORMATION SYSTEMS; [LECTURE NOTES IN COMPUTER SCIENCE], SPRINGER BERLIN HEIDELBERG, BERLIN, HEIDELBERG, PAGE(S) 49 - 60, XP019082591, ISBN: 978-3-540-76413-7 * abstract, §3, §4, Figure 1 *
  • CHANG M-T ET AL: "Deformed trademark retrieval based on 2D pseudo-hidden Markov model", PATTERN RECOGNITION, ELSEVIER, GB, vol. 34, no. 5, 1 May 2001 (2001-05-01), pages 953-967, XP004321316, ISSN: 0031-3203, DOI: 10.1016/S0031-3203(00)00053-4
  
Note: Within nine months from the publication of the mention of the grant of the European patent, any person may give notice to the European Patent Office of opposition to the European patent granted. Notice of opposition shall be filed in a written reasoned statement. It shall not be deemed to have been filed until the opposition fee has been paid. (Art. 99(1) European Patent Convention).


Description

TECHNICAL FIELD



[0001] The present invention relates to a system for extracting image signatures which are features for identifying (determining the identity of) images.

BACKGROUND ART



[0002] Image signatures are image features for identifying (determining the identity of) images. By comparing an image signature extracted from an image with an image signature extracted from another image, an identity scale (in general, referred to as similarity or distance) indicating a degree of the two images being identical can be calculated from a comparison result. Further, by comparing the calculated identity scale with a threshold, it is possible to determine whether or not the two images are identical. In this context, the meaning of "two images being identical" includes not only the case where the two images are identical at the level of image signals (pixel values of the pixels constituting the images), but also the case where one image is a duplicate image of the other by means of various alteration processes such as conversion of compression format of an image, conversion of size/aspect ratio of an image, adjustment of color tone of an image, various filtering processes (sharpening, smoothing, and the like) applied to an image, local processing (caption superimposition, cutout, and the like) applied to an image, and recapturing of an image. By using image signatures, as it is possible to detect duplication of an image or a moving image which is a set of images, for example, image signatures are applicable to an illegal copy detection system for images or moving images.

[0003] Patent Document 1 describes an example of an image signature. Fig. 18 is an illustration showing a method of extracting an image signature described in Patent Document 1. This image signature is a feature vector in multiple dimensions (sixteen dimensions in Fig. 18). The method includes respectively calculating average luminance from thirty two pieces of rectangle regions 244 (among them, sixteen pieces of rectangle regions are shown in Fig. 18) at predetermined positions in an image 240. and calculating differences in the average luminance between rectangle regions forming pairs (the paired rectangle regions are linked to each other with dotted lines 248 in Fig. 18), to thereby obtain a difference vector 250 in sixteen dimensions. With respect to the difference vector 250, a composite vector is generated by means of vector transformation, and a quantization index vector in sixteen dimensions, acquired by quantizing the respective dimensions of the composite vector, is used as an image signature.

[0004] Patent Document 1: Japanese Unexamined Patent Publication No. 8-500471

[0005] The document: KOTA IWAMOTO ET AL: "Response to the Call for Proposals on Video Signature Tools", 87. MPEG MEETING; 2-2-2009 - 6-2-2009; LAUSANNE; (MOTION PICTURE EXPERT GROUP OR ISO/IEC JTC1/SC29/WG11), no. M16171, 29 January 2009 (2009-01-29), XP030044768; relates to video signatures.

[0006] Further, the document DAIDI ZHONG ET AL: "Face Image Retrieval System Using TFV and Combination of Subimages", 28 June 2007 (2007-06-28), ADVANCES IN VISUAL INFORMATION SYSTEMS; [LECTURE NOTES IN COMPUTER SCIENCE], SPRINGER BERLIN HEIDELBERG, BERLIN, HEIDELBERG, PAGE(S) 49 - 60, XP019082591, ISBN: 978-3-540-76413-7; describes a face image retrieval system using TFV and combination of subimages.

SUMMARY



[0007] In an image signature formed of a feature vector in a plurality of dimensions, as the amount of information held by the feature vector is larger (redundancy is smaller) as correlation between dimensions are smaller, such an image signature has high discrimination capability which is a degree of discriminating different images. In contrast, if correlation between dimensions in a feature vector is large, as the amount of information held by the feature vector is small (redundancy is large), the discrimination capability is low. It should be noted that correlation between dimensions is a degree of similarity occurring from the features of the dimensions, and mathematically, it is a value calculated as, if occurrence of a feature of each dimension is set as a probability variable, a correlation coefficient between probability variables or a mutual information amount, for example. As such, it is desirable that an image signature formed of a feature vector in a plurality of dimensions should be designed such that correlation between dimensions is small.

[0008] Image signals (pixel values of the pixels constituting an image) have correlation between local regions in the image. Generally, as the distance between local regions is shorter, the correlation is larger. In particular, in an image in which a specific image pattern appears repeatedly (particularly, in the case where the image pattern repeatedly appears in regular cycles) (for example, an image of windows of a building arranged in a grid pattern, see Fig. 19(A)) or in an image formed of a particular texture (see Fig. 19(B)), for example, correlation between local regions of the image is large.

[First Problem]



[0009] Regarding an image signature formed of a feature vector including features extracted from a plurality of local regions of the image, as described in Patent Document 1, as the shapes of the local regions for extracting the features are the same in each dimension (in the example of Patent Document 1, rectangle regions in the same shape) with respect to an image in which correlation between local regions in the image is large, correlation between the dimensions of the extracted features is large. As such, there is a first problem that discrimination capability of the image signature (feature vector) is low. It should be noted that the shapes are identical means that the regions are identical including their size and angle (tilt or orientation).

[0010] For example, the image signature as described in Patent Document 1 has low discrimination capability with respect to an image in which a specific image pattern appears repeatedly (see Fig. 19(A)) or in an image formed of a particular texture (see Fig. 19(B)).

[Second Problem]



[0011] A second problem of the image signature described in Patent Document 1 is that as the shapes of the regions of respective dimensions for calculating features (feature vector) are in the identical rectangle shape (including size and angle), there is a blind spot on frequencies where it is impossible to detect frequency components having a cycle which is the same as the length of a side of the rectangle or which is a fraction of the integer thereof. This is because that if an average is calculated within a region for a signal component of such a particular frequency, the value becomes 0 regardless of the magnitude of the signal component, so that signals of such frequency component cannot be detected at all. More specifically, assuming that a frequency having the same cycle as the length of a side of the rectangle is f0, components of a frequency nf0 (n=1, 2, 3, ···) cannot be detected. As such, with respect to an image in which signals concentrate on direct current components and such frequency components, an average pixel value becomes the same as the direct current component, whereby there is no difference in values between regions. Consequently, the value of every feature extracted as a difference in average pixel values between regions becomes 0, so that discrimination cannot be performed (discrimination capability is significantly lowered). Practically, as it is difficult to detect not only the components of the frequency nf0 (n=1, 2, 3, ···) but also a certain nearby frequency regions, even if signals do not concentrate on the above-described particular frequency, signal components of such a frequency band cannot be used, whereby the discrimination capability is lowered. In order to alleviate this problem, it may be possible to increase the value of the frequency f0 so as to decrease the signal electricity fallen into the frequency band which is difficult to be detected. However, increasing the value of the frequency f0 means reducing the size of the region, leading to lowering of the robustness (a degree that a feature does not vary due to various alteration processes or noise) of the frequency. For example, if a region becomes smaller, the value of the feature largely varies with respect to minor positional shift, whereby the robustness of the feature is lowered. As described above, when using the identical rectangle regions, it is extremely difficult to secure the robustness while increasing the discrimination capability.

[Object of the Invention]



[0012] An object of the present invention is to provide an image signature extraction device capable of solving a problem that an image signature, extracted from an image having large correlation between local regions in the image or an image having signals which are concentrated on a particular frequency, has a lower discrimination capability which is a degree of discriminating different images.

[0013] The invention is defined in the independent claims.

[0014] According to an aspect, an image signature extraction device includes an image signature generation unit that extracts region features from respective sub-regions in an image in accordance with a plurality of pairs of sub-regions in the image, the pairs of sub-regions including at least one pair of sub-regions in which both a combination of shapes of two sub-regions of the pair and a relative position between the two sub-regions of the pair differ from those of at least one of other pairs of sub-regions, and based on the extracted region features of the respective sub-regions, generates an image signature to be used for identifying the image: and an encoding unit that encodes the image signature.

[0015] As the present invention is configured as described above, the discrimination capability, which is a degree of discriminating different images, of an image signature can be improved. In particular, this advantageous effect is significantly achieved with respect to an image in which correlation between local regions in the image is high.

[0016] Further, the present invention also provides another advantageous effect that the discrimination capability is not lowered with respect to an image having signals which are concentrated on a particular frequency.

BRIEF DESCRIPTION OF DRAWINGS



[0017] 

Fig. 1 is a block diagram showing a first embodiment of the present invention.

Fig. 1 is a block diagram showing a first embodiment of the present invention.

Fig. 2 is an illustration showing exemplary pairs of extraction regions for respective dimensions shown by each-dimension extraction information.

Fig. 3 is a block diagram showing an example of a comparison unit in the first embodiment of the present invention.

Fig. 4 is a block diagram showing another example of a comparison unit in the first embodiment of the present invention.

Fig. 5 is a flowchart showing a flow of the processing in the first embodiment of the present invention.

Fig. 6 is a block diagram showing the main part of a second embodiment of the present invention.

Fig. 7 is a flowchart showing a flow of the processing in the second embodiment of the present invention.

Fig. 8 is a block diagram showing a third embodiment of the present invention.

Fig. 9 is a table showing examples of region feature calculation methods for respective dimensions.

Fig. 10 is a flowchart showing a flow of the processing in the third embodiment of the present invention.

Fig. 11 s a block diagram showing a fourth embodiment of the present invention.

Fig. 12 is a table showing examples of comparison and quantization methods for respective dimensions.

Fig. 13 is a flowchart showing a flow of the processing in the fourth embodiment of the present invention.

Fig. 14-a is a table showing each-dimension extraction region information used in a fifth embodiment and a sixth embodiment of the present invention.

Fig. 14-b is a table showing each-dimension extraction region information used in the fifth embodiment and the sixth embodiment of the present invention.

Fig. 14-c is a table showing each-dimension extraction region information used in the fifth embodiment and the sixth embodiment of the present invention.

Fig. 14-d is a table showing each-dimension extraction region information used in the fifth embodiment and the sixth embodiment of the present invention.

Fig. 14-e is a table showing each-dimension extraction region information used in the fifth embodiment and the sixth embodiment of the present invention.

Fig. 14-f is a table showing each-dimension extraction region information used in the fifth embodiment and the sixth embodiment of the present invention.

Fig. 14-g is a table showing each-dimension extraction region information used in the fifth embodiment and the sixth embodiment of the present invention.

Fig. 14-h is a table showing each-dimension extraction region information used in the fifth embodiment and the sixth embodiment of the present invention.

Fig. 14-i is a table showing each-dimension extraction region information used in the fifth embodiment and the sixth embodiment of the present invention.

Fig. 14-j is a table showing each-dimension extraction region information used in the fifth embodiment and the sixth embodiment of the present invention.

Fig. 15-a is a table showing each-dimension region feature calculation method information used in the fifth embodiment of the present invention.

Fig. 15-b is a table showing each-dimension region feature calculation method information used in the fifth embodiment of the present invention.

Fig. 15-c is a table showing each-dimension region feature calculation method information used in the fifth embodiment of the present invention.

Fig. 15-d is a table showing each-dimension region feature calculation method information used in the fifth embodiment of the present invention.

Fig. 15-e is a table showing each-dimension region feature calculation method information used in the fifth embodiment of the present invention.

Fig. 16-a is a table showing each-dimension region feature calculation method information used in the sixth embodiment of the present invention.

Fig. 16-b is a table showing each-dimension region feature calculation method information used in the sixth embodiment of the present invention.

Fig. 16-c is a table showing each-dimension region feature calculation method information used in the sixth embodiment of the present invention.

Fig. 16-d is a table showing each-dimension region feature calculation method information used in the sixth embodiment of the present invention.

Fig. 16-c is a table showing each-dimension region feature calculation method information used in the sixth embodiment of the present invention.

Fig. 17-a is a table showing each-dimension comparison and quantization method information used in the fifth embodiment and the sixth embodiment of the present invention.

Fig. 17-b is a table showing each-dimension comparison and quantization method information used in the fifth embodiment and the sixth embodiment of the present invention.

Fig. 17-c is a table showing each-dimension comparison and quantization method information used in the fifth embodiment and the sixth embodiment of the present invention.

Fig. 17-d is a table showing each-dimension comparison and quantization method information used in the fifth embodiment and the sixth embodiment of the present invention.

Fig. 17-e is a table showing each-dimension comparison and quantization method information used in the fifth embodiment and the sixth embodiment of the present invention.

Fig. 18 is an illustration showing a method of extracting an image signature described in Patent Document 1.

Fig. 19 is an illustration showing examples of images in which correlation between local regions is large.

Fig. 20 is a block diagram showing a matching unit which performs matching between quantization index vectors.

Fig. 21 is a flowchart showing an exemplary process performed by the matching unit which performs matching between quantization index vectors.

Fig. 22 is a flowchart showing another exemplary process performed by the matching unit which performs matching between quantization index vectors.

Fig. 23 is a flowchart showing still another exemplary process performed by the matching unit which performs matching between quantization index vectors.

Fig. 24 is a table showing examples of indexes applied to 1024 pieces of blocks formed by dividing an image into 32 in a vertical direction and 32 in a horizontal direction.

Fig. 25-a is a table showing regions belonging to one type, among regions corresponding to the respective dimensions in an seventh embodiment of the present invention.

Fig. 25-b is a table showing regions belonging to one type, among regions corresponding to the respective dimensions in the seventh embodiment of the present invention.

Fig. 25-c is a table showing regions belonging to one type, among regions corresponding to the respective dimensions in the seventh embodiment of the present invention.

Fig. 25-d is a table showing regions belonging to one type, among regions corresponding to the respective dimensions in the seventh embodiment of the present invention.

Fig. 25-e is a table showing regions belonging to one type, among regions corresponding to the respective dimensions in the seventh embodiment of the present invention.

Fig. 25-f is a table showing regions belonging to one type, among regions corresponding to the respective dimensions in the seventh embodiment of the present invention.

Fig. 25-g is a table showing regions belonging to one type, among regions corresponding to the respective dimensions in the seventh embodiment of the present invention.

Fig. 26 is a table showing a relation among a region type of each dimension, a dimension number, and an index corresponding to a threshold.

Fig. 27-a is an illustration showing an example of first and second extraction regions of a dimension of a region type a.

Fig. 27-b is an illustration showing an example of first and second extraction regions of a dimension of a region type b.

Fig. 27-c is an illustration showing an example of first and second extraction regions of a dimension of a region type c.

Fig. 27-d is an illustration showing an example of first and second extraction regions of a dimension of a region type d.

Fig. 27-e is an illustration showing an example of first and second extraction regions of a dimension of a region type e.

Fig. 27-f is an illustration showing an example of first and second extraction regions of a dimension of a region type f.

Fig. 27-g is an illustration showing an example of first and second extraction regions of a dimension of a region type g.


EXEMPLARY EMBODIMENTS


[First Embodiment]


[Configuration of First Embodiment]



[0018] Next, a first embodiment of the present invention will be described in detail with reference to the drawings.

[0019] Referring to Fig. 1, an image signature extraction device according to the first embodiment of the present invention is a system for outputting, with respect to an input image, a feature vector (more specifically, a quantization index vector) formed of a plurality of dimensions, as an image signature. The image signature extraction device includes a dimension determination unit 1, an extraction region acquisition unit 2, a region feature calculation unit 3, and a comparison unit 4.

[0020] The dimension determination unit 1 determines a dimension of a feature vector to be extracted next, and supplies it to the extraction region acquisition unit 2. The dimension determination unit 1 sequentially supplies dimensions of the feature vector to be extracted, and the constituent elements after the extraction region acquisition unit 2 extract features corresponding to the supplied dimensions. For example, if a feature vector is formed of N dimensions, the dimension determination unit 1 may sequentially supply the 1st dimension to the Nth dimension to the extraction region acquisition unit 2. The dimensions may be supplied in any order if all of the dimensions of the feature vector are supplied finally. It is also possible to supply a plurality of dimensions in parallel.

[0021] To the extraction region acquisition unit 2, each-dimension extraction region information is supplied as an input, besides the dimensions supplied from the dimension determination unit 1.

[0022] The each-dimension extraction region information is information indicating a predetermined pair of a first extraction region and a second extraction region for extracting the feature of a dimension, which is associated with each dimension of a feature vector. The first and second extraction regions have the following features as prerequisites.

[Prerequisites of First and Second Extraction Regions]



[0023] Prerequisites of the first and second extraction regions are that relative positions of a pair of extraction regions are different among the dimensions, and combinations of the shapes of the pair of extraction regions are different among the dimensions.

[0024] Fig. 2 shows an exemplary pair of extraction regions, satisfying the prerequisites, for each of the dimensions indicated by the each-dimension extraction information. Different from the extraction regions for an image signature shown in Fig. 18, combinations of the shapes of the pairs of extraction regions are different among the respective dimensions. Different in shape includes the same shapes of different angles (e.g., the second extraction region of the 1st dimension and the first extraction region of the 7th dimension in Fig. 2), and similar shapes of different sizes (e.g., the second extraction region of the 1st dimension and the second extraction region of the 9th dimension in Fig. 2). It should be noted that a minimum condition is that at least one dimension, in which a pair of extraction regions have a combination of different shapes, is included in all of the dimensions of a feature vector. It is desirable that a feature vector includes a larger number of dimensions having pairs of extraction regions in (a combination of) different shapes from each other. This is because as a feature vector includes a larger number of pairs of extraction regions in (a combination of) different shapes from each other, a larger number of correlations between dimensions become smaller in the feature vector, whereby the discrimination capability becomes higher. For example, the shapes of a pair extraction regions may be different from each other in every dimensions in a feature vector.

[0025] A first extraction region and a second extraction region in a dimension are not necessarily in the same shape as shown in the 9th dimension of Fig. 2, but may be in different shapes as shown in other dimensions of Fig. 2. If the shapes of a first extraction region and a second extraction in each dimension are different, the correlation between the features extracted from the first extraction region and the second extraction region becomes smaller, whereby the discrimination capability becomes higher. As such, it is desirable. Further, in that case, as a possibility that the first extraction region and the second extraction region become blind spots for the same frequency at the same time is low, the discrimination capability becomes high.

[0026] The respective extraction regions may take any shapes. For example, any complicated shapes, such as the second extraction region of the 6th dimension in Fig. 2, are also acceptable. If extraction regions are formed of a plurality of pixels of an image, a line segment and a curved line are also acceptable as show in the 7th dimension and the 10th dimensions in Fig. 2. Further, an extraction region may consist of a plurality of discontinuous small regions, as the first extraction region of the 8th dimension, the first and second extraction regions of the 11th dimension, and the first extraction region of the 12th dimension. As described above, if a feature vector includes extraction regions in complicated shapes, correlation between the dimensions of the features extracted therefrom can be lowered, whereby the discrimination capability can be higher.

[0027] Further, it is also possible that portions of a first extraction region and a second extraction region overlap each other, as in the 5th dimension of Fig. 2. Further, either one of a pair of extraction regions may be included in the other one. As described above, by allowing overlapping of a pair of extraction regions, as a larger number of patterns (relative position, distance) can be taken for pairs of extraction regions, patterns enabling to reduce the correlation between dimensions can be increased, whereby the possibility of improving the discrimination capability is increased.

[0028] Further, portions of extraction regions may overlap each other between dimensions as the respective dimensions shown in Fig. 2. different from the extraction regions for the image signature shown in Fig. 18. If extraction regions are taken exclusively between dimensions as shown in the extraction regions for the image signature shown in Fig. 18, possible patterns of pairs of extraction regions are limited. By allowing overlapping of extraction regions between dimensions as shown in Fig. 2. it is possible to increase patterns enabling to reduce the correlation between dimensions, whereby the possibility of improving the discrimination capability is increased. However, if there are too many overlapped portions of extract regions between dimensions, correlation between dimensions becomes large so that the discrimination capability becomes low. As such, it is not desirable.

[0029] Further, between dimensions (pair of dimensions) in which the extraction regions are close to each other, it is desirable that the combinations of the shapes of the pairs of extraction regions in such dimensions (pair of dimensions) are different from each other. For example, between dimensions (pair of dimensions) in which the extraction regions are close to each other, the combinations of the shapes of the pairs of extraction regions in such dimensions (pair of dimensions) may be different from each other. In that case, between all dimensions (pairs of dimensions) in which the extraction regions are close to each other, the combinations of the shapes of the pairs of extraction regions in those dimensions may be different from each other. It is also possible that among the dimensions (pairs of dimensions) in which the extraction regions are close to each other, the combinations of the shapes of the pairs of extraction regions may be different from each other in the dimensions (pairs of dimensions) of a predetermined proportion or larger, or among the dimensions (pairs of dimensions) in which the extraction regions are close to each other, the combinations of the shapes of the pairs of extraction regions may be different from each other in at least one pair of dimensions.

[0030] It should be noted that a pair of dimensions in which the extraction regions are close to each other means a pair of dimensions in which the distance between the extraction regions (first extraction region and second extraction region) of the two dimensions, calculated by a predetermined method, is smaller than a predetermined threshold. The distance between the extraction regions (first extraction region and second extraction region) of two dimensions may be calculated as, for example, an average value of the distances between all pixels included in the extraction regions (first extraction region and second extraction region) of the respective dimensions. The distance may also be calculated as, for example, a minimum value of the distances between all pixels included in the extraction regions (first extraction region and second extraction region) of the respective dimensions. It is also possible to calculate the center-of-gravity coordinates of the extraction regions (center-of-gravity coordinate of the first extraction region and the center-of-gravity coordinate of the second extraction region) of the respective dimensions, and calculate the distances between them (four distances are calculated in total), and then define the distance between the extraction regions of the two dimensions as an average value or a minimum value thereof, for example. However, methods of calculating the distance between the extraction regions of two dimensions are not limited to these calculation methods.

[0031] With the combinations of the shapes of the pairs of extraction regions being different from each other between dimensions (pairs of dimensions) in which the extraction regions are close to each other, the following advantageous effect can be achieved. Between dimensions in which the extraction regions are close to each other, correlation between the extracted features tends to be large. If the combinations of the shapes of the pairs of extraction regions in those regions are the same, the correlation between the features is larger, whereby the redundancy in the entire feature vector becomes high, so that the discrimination capability is lowered. As such, by allowing the combinations of the shapes of the pairs of extraction regions to be different in those regions in order to offset the correlation between the extracted features becoming large due to the fact that the extraction regions are close to each other, the correlation between the features can be reduced, whereby the redundancy in the entire feature vector can be low, so that it is possible to prevent deterioration in the discrimination capability.

[0032] From an opposite point of view, between dimensions (pair of dimensions) in which the combinations of the shapes of the pairs of dimensions are the same, it is desirable that the distance between the extraction regions in those dimensions (pair of dimensions) is large. For example, between dimensions (pair of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the distance between such dimensions (pair of dimensions) may be larger than a predetermined threshold. In that case, between all dimensions (pairs of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the distances between those dimensions (pairs of dimensions) may be larger than a predetermined threshold. It is also possible that among the dimensions (pairs of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the distances between dimensions (pairs of dimensions) may be larger than a predetermined threshold in the dimensions (pairs of dimensions) of a predetermined proportion or larger, or among the dimensions (pairs of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the distance between dimensions (pairs of the dimensions) may be larger than a predetermined threshold in at least one pair of dimensions.

[0033] Further, between dimensions (pair of dimensions) in which the extraction regions are adjacent to each other, it is desirable that the combinations of the shapes of the pairs of extraction regions in such dimensions (pair of dimensions) are different from each other. It should be noted that dimensions (pair of dimensions) in which the extraction regions are adjacent means that partial peripheries of the extraction regions contact each other. For example, between dimensions (pair of dimensions) in which the extraction regions are adjacent, the combinations of the shapes of the pairs of extraction regions in such dimensions (pair of dimensions) may be different from each other. In that case, between all dimensions (pairs of dimensions) in which the extraction regions are adjacent, the combinations of the shapes of the pairs of extraction regions in those dimensions may be different from each other. It is also possible that among the dimensions (pairs of dimensions) in which the extraction regions are adjacent, the combinations of the shapes of the pairs of extraction regions may be different from each other in the dimensions (pairs of dimensions) in a predetermined proportion or larger, or among the dimensions (pairs of dimensions) in which the extraction regions are adjacent, the combinations of the shapes of the pairs of extraction regions may be different from each other in at least one pair of dimensions.

[0034] With the combinations of the shapes of the pairs of extraction regions being different from each other between the dimensions (pairs of dimensions) in which the extraction regions are adjacent to each other, the following advantageous effect can be achieved. Between dimensions in which the extraction regions are adjacent, correlation between the extracted features tends to be large. This is because the distance between the extraction regions of the two dimensions is close between the dimensions in which the extraction regions are adjacent to each other (correlation is high between close regions within an image). If the combinations of the shapes of the pairs of extraction regions in those regions are the same, the correlation between the features is larger, whereby the redundancy in the entire feature vector becomes high, so that the discrimination capability is lowered. As such, by allowing the combinations of the shapes of the pairs of extraction regions to be different in those regions in order to offset the correlation between the extracted features becoming large due to the fact that the extraction regions are adjacent to each other, the correlation between the features can be reduced, whereby the redundancy in the entire feature vector can be low, so that it is possible to prevent deterioration in the discrimination capability.

[0035] From an opposite point of view, between dimensions (pair of dimensions) in which the combinations of the shapes of the pairs of dimensions are the same, it is desirable that the extraction regions in those dimensions (pair of dimensions) are not adjacent to each other. For example, between dimensions (pair of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the extraction regions in such dimensions (pair of dimensions) may not be adjacent. In that case, between all dimensions (pairs of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the extraction regions in those dimensions (pairs of dimensions) may not be adjacent. It is also possible that among the dimensions (pairs of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the extraction regions may not be adjacent in the dimensions (pairs of dimensions) of a predetermined proportion or larger, or among the dimensions (pairs of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the extraction regions may not be adjacent in at least one pair of dimensions.

[0036] Further, between dimensions (pair of dimensions) in which the extraction regions partially overlap, it is desirable that the combinations of the shapes of the pairs of extraction regions in such dimensions (pair of dimensions) are different from each other. For example, between dimensions (pair of dimensions) in which the extraction regions partially overlap, the combinations of the shapes of the pairs of extraction regions in such dimensions (pair of dimensions) may be different from each other. In that case, between all dimensions (pairs of dimensions) in which the extraction regions partially overlap, the combinations of the shapes of the pairs of extraction regions in those dimensions may be different from each other. It is also possible that among the dimensions (pairs of dimensions) in which the extraction regions partially overlap, the combinations of the shapes of the pairs of extraction regions may be different from each other in the dimensions (pairs of dimensions) of a predetermined proportion or larger, or among the dimensions (pairs of dimensions) in which the extraction regions partially overlap, the combinations of the shapes of the pairs of extraction regions may be different from each other in at least one pair of dimensions.

[0037] With the combinations of the shapes of the pairs of extraction regions being different from each other between the dimensions (pairs of dimensions) in which the extraction regions partially overlap, the following advantageous effect can be achieved. Between dimensions in which the extraction regions partially overlap, correlation between the extracted features tends to be large. This is because between the dimensions in which the extraction regions partially overlap, the regions used for calculating the features are shared partially. If the combinations of the shapes of the pairs of extraction regions in those regions are the same, the correlation between the features is larger, whereby the redundancy in the entire feature vector becomes high, so that the discrimination capability is lowered. As such, by allowing the combinations of the shapes of the pairs of extraction regions to be different in those regions in order to offset the correlation between the extracted features becoming large due to the fact that the extraction regions partially overlap, the correlation between the features can be reduced, whereby the redundancy in the entire feature vector can be low, so that it is possible to prevent deterioration in the discrimination capability.

[0038] With the combinations of the shapes of the pairs of extraction regions being different from each other between the dimensions (pairs of dimensions) in which the extraction regions partially overlap, another advantageous effect can be achieved. With the combinations of the shapes of the pairs of extraction regions being different from each other between the dimensions (pairs of dimensions) in which the extraction regions partially overlap, there is an advantage that tolerance against falsification (particularly, malicious falsification) of an image can be enhanced. Malicious falsification to an image is conducted by changing the pixel value of a partial region of the image, for example, in an attempt to deteriorate accuracy in identity determination by the image signature (feature vector) extracted from the image. Between the dimensions in which the extraction regions are partially overlap, if the overlapped region of the image is falsified, that is, if the pixel value of the overlapped region is changed, the feature (region feature) extracted therefrom is affected, whereby it is highly likely that the feature may be different from the feature extracted from the region which was not falsified. If the combinations of the shapes of the pairs of extraction regions in those dimensions are the same, the features extracted from the extraction regions of the respective dimensions are affected in a similar manner, whereby the possibility that the features are changed at the same time would be high, so that the robustness is low. If the combinations of the shapes of the pairs of extraction regions in those dimensions are different, even if the overlapped region is falsified, the possibility that the features are affected in a similar manner so that the features are changed at the same time can be reduced, whereby the robustness can be secured. As such, it becomes difficult to falsify the overlapped region by changing the features of a plurality of dimensions sharing the overlapped region at the same time. Accordingly, with the combinations of the shapes of the pairs of extraction regions being different from each other between the dimensions (pairs of dimensions) in which the extraction regions partially overlap, tolerance against falsification (particularly, malicious falsification) in an image can be enhanced.

[0039] From an opposite point of view, between dimensions (pair of dimensions) in which the combinations of the shapes of the pairs of dimensions are the same, it is desirable that the extraction regions in those dimensions (pair of dimensions) do not overlap each other. For example, between dimensions (pair of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the extraction regions in such dimensions (pair of dimensions) may not overlap each other. In that case, between all dimensions (pairs of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the extraction regions in those dimensions (pairs of dimensions) may not overlap each other. It is also possible that among the dimensions (pairs of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the extraction regions may not overlap each other in the dimensions (pairs of dimensions) of a predetermined proportion or larger, or among the dimensions (pairs of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the extraction regions may not overlap each other in at least one pair of dimensions.

[0040] Further, it is desirable that extraction regions are taken such that the region from which features are not extracted is small (which means almost all of the screen image can be covered) when the extraction regions for all dimensions are combined. If the region from which features are not extracted is large as in the case of Fig. 18, a large portion of the information included in the image signal (pixel values of the pixels constituting the image) is not used, so that the discrimination capability is not high. By taking extraction regions such that the region from which features are not extracted is small (which means almost all of the screen image can be covered) when the extraction regions for all dimensions are combined, a larger portion of the information included in the image signal can be reflected on the features, whereby the discrimination capability can be high. Further, it is desirable that extraction features are not biased but obtained uniformly from the entire image, when the extraction regions for all dimensions are combined. However, if the probability of performing local processing such as caption superimposition on a specific region is high, it is desirable to obtain extraction regions while avoiding such a region. Further, generally, as regions around the edge of an image often do not include feature portions of the image, it is desirable to obtain extraction regions while avoiding such surrounding regions.

[0041] Furthermore, it is desirable that the size and the relative position (distance, direction) of the extraction regions follow certain distribution (uniform distribution, for example), because if the relative position (distance, direction) follows the uniform distribution, the extraction region are not biased with respect to the distance or direction whereby the extraction regions do not concentrate on a particular distance or direction, whereby a wider variety can be achieved. Further, as the correlation between regions is larger as the relative positions are close, in order to offset such an effect, it is desirable that the difference in shape is larger as the relative positions are closer.

[0042]  The each-dimension extraction region information may be in any form if a first extraction region and a second extraction region for each dimension can be uniquely specified with the information. Further, as an extraction region must always be the same region with respect to an image of any size or aspect ratio, the each-dimension extraction region information should be in a form which can obtain the same extraction region with respect to an image of any size or aspect ratio. For example, the each-region extraction region may describe the position and the shape of an extraction region with respect to an image having a predetermined size and aspect ratio (for example, an image having a horizontal width of 320 pixels x a vertical width of 240 pixels). In that case, with respect to an image input with an arbitrary size and aspect ratio, resizing the image first to have the predetermined size and aspect ratio and then specifying the extraction region in accordance with the position and the shape of the extraction region described in the each-dimension extraction region information. In contrast, it is also possible to convert the position and the shape of the extraction region described in the each-dimension extraction region information corresponding to the image of any size and aspect ratio of the input image to thereby specify the extraction region.

[0043] The information indicating each extraction region included in the each-dimension extraction region information may be information describing a set of coordinate values of all pixels constituting the extraction region with respect to an image of a predetermined size and aspect ratio (for example, an image having a horizontal width of 320 pixels x a vertical width of 240 pixels). Further, the information indicating each extraction region included in the each-dimension extraction region information may be information describing, with parameters, the position and the shape of the extraction region with respect to an image of a predetermined size and aspect ratio. For example, if the shape of the extraction region is a quadrangle, the information may describe the coordinate values of the four corners of the quadrangle. Further, if the shape of the extraction region is a circle, the information may describe the coordinate values of the center and the radius value of the circle.

[0044] Further, it is also possible to adopt a method of generating, with use of a seed of pseudo random numbers as an each-dimension extraction region information, pseudo random numbers by starting from the seed inside the extraction region acquisition unit 2 to thereby generate extraction regions in different shapes according to the random numbers (for example, the four corners of a quadrangle are determined according to the random numbers). Specifically, an each-dimension extraction region can be acquired according to the following procedure.
  1. (1) A seed of pseudo random numbers is supplied as each-dimension extraction region information.
  2. (2) A dimension n is set to be n=1
  3. (3) Pseudo random numbers are generated, and the four corners of a quadrangle of a first extraction region for the dimension n are determined.
  4. (4) Pseudo random numbers are generated, and the four corners of a quadrangle of a second extraction region for the dimension n are determined.
  5. (5) A dimension n is set to be n+1, and the procedure returns to (3).


[0045] As the extraction regions are determined based on the random numbers, the generated extraction regions are different from each other for respective dimensions. Further, if the seed of the pseudo random numbers is the same, as the same random numbers are generated each time (with respect to any images), the same extraction regions are reproduced for different images.

[0046] The extraction region acquisition unit 2 acquires, from the each-dimension extraction region information supplied as an input, information indicating the first extraction region and the second extraction region corresponding to the dimension supplied from the dimension determination unit 1, and outputs the information to the extraction region representative value calculation unit 3.

[0047] To the region feature calculation unit 3, an image which is an extraction target for an image signature is supplied as an input, besides the input from the extraction region acquisition unit 2 (information indicating the first extraction region and the second extraction region). The region feature calculation unit 3 includes a first region feature calculation unit 31 and a second region feature calculation unit 32. With use of the first region feature calculation unit 31, the region feature calculation unit 3 calculates, from the image supplied as an input, a feature of the first extraction region as a first region feature for each dimension based on the information indicating the first extraction region supplied from the extraction region acquisition unit 2, and supplies the feature to the comparison unit 4. Further, with use of the second region feature calculation unit 32, the region feature calculation unit 3 calculates, from the image supplied as an input, a feature of the second extraction region as a second region feature for each dimension based on the information indicating the second extraction region supplied from the extraction region acquisition unit 2, and supplies the feature to the comparison unit 4.

[0048] It should be noted that in order to specify the respective extraction regions with respect to the input image based on the information indicating the first extraction region and the second extraction region, the region feature calculation unit 2 resizes the image to have a predetermined size and aspect ratio in the each-dimension extraction region information, if necessary.

[0049] The region feature calculation unit 3 calculates region features of the respective extraction regions using the pixel values of a group of pixels included in the respective extraction regions. In this embodiment, a pixel value is a value of a signal held by each pixel of the image, which is a scalar quantity or a vector quantity. For example, if the image is a luminance image, a pixel value is a luminance value (scalar quantity), and if the image is a color image, a pixel value is a vector quantity indicating a color component. If the color image is an RGB image, a pixel value is a three-dimensional vector quantity of an R component, a G component, and a B component. Further, if the color image is a YCbCr image, a pixel value is a three-dimensional vector quantity of a Y component, a Cb component, and a Cr component.

[0050] For calculating the region features of the extraction regions, any methods can be used if a method of calculating the extraction regions (first extraction region and the second extraction region) of the dimension is constant (the same calculation method is used for any input images).

[0051] Further, the region feature to be calculated may be a scalar quantity or a vector quantity. For example, if a pixel value is a scalar quantity such as a luminance value, the region feature may be calculated as an average value, a median value, a mode value, a maximum value, a minimum value, or the like (each of them is a scalar quantity). Further, it is also possible to sort pixel values included in an extraction region, and obtain a pixel value at a predetermined proportional position from the top or the bottom of the distribution (sorted order) as a region feature (which is also a scalar quantity), for example. More specifically, explanation will be given for the case where P% of the percentage (e.g., P=25%) is the predetermined proportion. Pixel values (luminance values) of the total N pieces of pixels included in an extraction region are sorted in ascending order, and a set of the pixel values (luminance values) sorted in ascending order is indicated as Y(i)={Y(0), Y(1), Y(2), ..., Y(N-1)}. In this example, a pixel value at a position of P% from the bottom of the permutation sorted in ascending order is Y(floor(NP/100)) for example, so this value is obtained as a region feature of the extraction region. It should be noted that floor( ) is a function in which after the decimal point is truncated. In this example, a region feature calculated by applying this formula (Y(floor(NP/100))) with respect to the luminance value of the pixel included in the extraction region is referred to as a "percentile luminance value feature".

[0052] Further, if the pixel value is a vector quantity such as a color component, it is possible to first convert the value into a scalar quantity by means of any method, and then calculate a region feature by the above-described method. For example, if the pixel value is a three-dimensional vector quantity of RGB components, it is possible to first convert the value into a luminance value which is a scalar quantity, and then calculate a region feature by the above-described method. Further, if the pixel value is a vector quantity, it is also possible to use an average vector of the pixel values included in the extraction region as a region feature.

[0053] Further, it is also possible to perform any operation (differential operation, filtering operation) such as edge detection or template matching with respect to an extraction region, and use the operation result as a region feature. For example, it may be a two-dimensional vector quantity indicating the edge direction (gradient direction), or a scalar quantity indicating the similarity value with a template.

[0054] Further, a histogram showing the color distribution, edge direction distribution, or edge intensity distribution, included in the extraction region, may be obtained as a region feature (each of them is a vector quantity).

[0055] Further, any of the various types of features defined in ISO/IEC 15938-3 may be used, which include dominant color, color layout, scalable color, color structure, edge histogram, homogeneous texture, texture browsing, region shape, contour shape, shape 3D, parametric motion, and motion activity.

[0056] The comparison unit 4 compares the first region feature with the second region feature supplied from the region feature calculation unit 3 for each dimension, and quantizes the comparison result to output the acquired quantization index. As the comparison unit 4 outputs quantization indexes for the respective dimensions, a quantization index vector consisting of the quantization indexes of a plurality of dimensions is finally output.

[0057] The comparison unit 4 may use any methods to compare a first region feature with a second region feature and perform quantization. Also, the number of quantization indexes for each dimension is also arbitrary.

[0058] If the region features are scalar quantities (e.g., average luminance) for example, the comparison unit 4 may compare their magnitudes, and if the first region feature is larger, set the quantization index to be +1, and in other cases, sets the quantization index to be -1, so as to quantize the comparison result into binary values of quantization indexes of +1 and -1. In should be noted that regarding a dimension n, if the first region feature is Vn1 and the second region feature is Vn2, a quantization index Qn of the dimension n can be calculated by the following expression.



[0059] Fig. 3 shows a more detailed configuration diagram of the comparison unit 4 when the comparison unit 4 performs comparison and quantization based on the above Expression 1.

[0060] Referring to Fig. 3, the comparison unit 4 includes a magnitude comparison unit 41 and a quantization unit 42.

[0061] When the first region feature and the second region features are supplied, the magnitude comparison unit 41 compares the value of a first region feature with the value of the second region feature, and supplies the comparison result to the quantization unit 42. This means that the magnitude comparison unit 41 compares the magnitude of Vn1 with that of Vn2, and supplies information indicating whether the comparison result is Vn1>Vn2 or Vn1≤Vn2 to the quantization unit 42 as a magnitude comparison result.

[0062] Based on the magnitude comparison result supplied from the magnitude comparison unit 41, the quantization unit 42 performs quantization according to Expression 1, and outputs a quantization index. As such, the quantization unit 42 outputs a quantization index in such a manner that if information indicating that the comparison result is Vn1>Vn2 is supplied, a quantization index is +1, while if information indicating that the comparison result is Vn1≤Vn2 is supplied, an quantization index is -1.

[0063] It should be noted that the comparison and quantization method according to Expression 1 is hereinafter referred to as a comparison and quantization method A.

[0064] Further, if the region feature is a scalar volume (e.g., an average luminance), the comparison unit 4 may perform quantization in such a manner that if the absolute value of the difference value is smaller than or equal to a predetermined threshold, it is determined that there is no difference between the first region feature and the second region feature so that a quantization index 0 indicating no difference is set, and in other cases, the comparison unit 4 compares their magnitude and if the first region feature is larger, a quantization index +1 is set, while in other cases, a quantization index -1 is set, whereby the quantization index is in any of ternary values of +1, 0, and -1. Assuming that the first region feature of a dimension n is Vn1 and the second region feature thereof is Vn2 and a predetermined threshold is th, a quantization index Qn of the dimension n can be calculated from the following expression.



[0065] Fig. 4 shows a more detailed configuration diagram of the comparison unit 4 when the comparison unit 4 performs comparison and quantization according to Expression 2.

[0066] Referring to Fig. 4, the comparison unit 4 includes a difference value calculation unit 43 and a quantization unit 44. To the quantization unit 44. a threshold which is predetermined information indicating the boundary of quantization (quantization boundary information) is supplied beforehand as an input.

[0067] When the first region feature and the second region feature are supplied, the difference value calculation unit 43 calculates a difference value between the value of the first region feature and the value of the second region feature, and supplies the calculated difference value to the quantization unit 44. This means that the difference value calculation unit 43 calculates Vn1-Vn2, and supplies the resultant value to the quantization unit 44.

[0068] The quantization unit 44 performs quantization according to Expression 2 based on the difference value supplied from the difference value calculation unit 43 and a threshold which is information indicating the predetermined boundary of quantization (quantization boundary information) supplied as an input, and outputs a quantization index. This means that the quantization unit 42 outputs a quantization index based on the value of Vnl-Vn2 supplied from the difference value calculation unit 41 and the threshold th supplied as an index, in such a manner that if |Vn1-Vn2|>th and Vn1-Vn2>0, the quantization index is +1, if |Vn1-Vn2|>th and Vn1-Vn2≤0, the quantization value is -1, and if |Vn1-Vn2|≤th, the quantization index is 0.

[0069] The comparison and quantization method based on Expression 2 is hereinafter referred to as a comparison and quantization method B.

[0070] Although quantization is performed in ternary values based on the difference value in this example, it is possible to perform quantization in a larger number of (levels of) quantization indexes according to the magnitude of the difference value. Even in that case, the comparison unit 4 has the configuration as shown in Fig. 4, and to the quantization unit 44, a plurality of thresholds as information indicating the predetermined boundaries of quantization for respective levels (quantization boundary information) are supplied as inputs. A comparison and quantization method for quantization into four or more levels, based on this difference value and the thresholds supplied as inputs, is hereinafter referred to as a comparison and quantization method C.

[0071] As described above, by introducing a quantization index indicating there is no difference for the case where the difference between the first region feature and the second region feature is small (smaller than or equal to a predetermined threshold) so that it is determined that no difference exists, it is possible to make the feature (quantization index) in a dimension of a pair of extraction regions having a small difference between the region features more stable, that is, more robust with respect to various alteration process and noise, compared with the method according to Expression 1. As such, it is possible to output an image signature (quantization index vector) which is stable with respect to an image having less difference between local regions in whole, that is, an flat image having less variations in whole (e.g., an image of blue sky), and is robust with respect to various alteration processes and noise.

[0072] Further, if the region feature is a vector quantity, for example, the comparison unit 4 may first convert the vector quantity into a scalar quantity by means of any arbitrary method, and then perform quantization by the above-described method (this comparison and quantization method is hereinafter referred to as a comparison and quantization method D). It is also possible to calculate a difference vector, which is a difference from the vector of a second extraction region, from the vector of a first extraction region, and quantize the difference vector to thereby obtain a quantization index, for example. In that case, representative vectors (center of gravity vectors, etc) predetermined for respective quantization indexes are supplied, and are classified into quantization indexes having the largest similarity value (smallest distance) between the representative vectors and the difference vectors (this comparison and quantization method is hereinafter referred to as a comparison and quantization method E). Further, similar to the quantization of a scalar quantity according to the above-described Expression 2, if a norm of the difference vector is smaller than or equal to a predetermined threshold, it is possible to determine that there is no difference between the first region feature and the second region feature to thereby introduce a quantization index indicating no difference as a quantization index 0 which indicates no difference.

[0073] It should be noted that when matching quantization index vectors output in the present invention (when comparing a quantization index vector extracted from an image with a quantization index vector extracted from another image to determine whether or not they are identical), the number of dimensions in which quantization indexes conform (similarity value) or the number of dimensions in which quantization indexes do not conform (Hamming distance) may be calculated as an identity scale, which is compared with a threshold, whereby the identity of the images can be determined. Further, if the quantization indexes are calculated based on Expression 2 in the comparison unit 4, an identity scale (similarity value) can be calculated as follows. First, quantization index vectors of two images are compared with each other between corresponding dimensions, and the number of dimensions in which not "both quantization indexes are 0" is calculated (this value is set to be A). Next, in the dimensions in which not "both quantization indexes are 0", the number of dimensions in which the quantization indexes conform is calculated (this value is set to be B). Then, a similarity value is calculated as B/A. If A=0 (that is, if both quantization indexes are 0 in every dimensions), the similarity value is set to be a predetermined numerical value (e.g., 0.5).

[Operation of First Embodiment]



[0074] Next, operation of the image signature extraction device according to the first embodiment will be described with reference to the flowchart of Fig. 5. In the flowchart of Fig. 5, a dimension (number) of a feature vector is indicated by "n", and there are N dimensions in total from 1 to N.

[0075] First, the dimension determination unit 1 determines a dimension 1 as the first dimension (n=1) for extracting a feature vector, and supplies it to the extraction region acquisition unit 2 (step A1).

[0076] Next, the extraction region acquisition unit 2 acquires information indicating a first extraction region and a second extraction region of the dimension n from each-dimension extraction region information supplied as an input, and supplies it to the region feature calculation unit 3 (step A2).

[0077] Then, the region feature calculation unit 3 calculates a first region feature and a second region feature of the dimension n from the image supplied as an input, and supplies the features to the comparison unit 4 (step A3).

[0078] Then, the comparison unit 4 compares the first region feature with the second region feature of the dimension n, quantizes the comparison result, and outputs a quantization index (step A4).

[0079] Then, it is determined whether or not output of quantization indexes for all dimensions has been completed (that is, it is determined whether n<N is true or false) (step A5). If output of quantization indexes for all dimensions has been completed (that is, if n<N is false), the processing ends. If output of quantization indexes for all dimensions has not been completed (that is, if n<N is true), the processing proceeds to step A6. At step A6, the dimension determination unit 1 determines the next dimension for extracting a feature vector (n=n+1), and supplies it to the extraction region acquisition unit 2. Then, the processing returns to step A2.

[0080] It should be noted that although the extraction processing is performed in order from the dimension 1 to the dimension N, any order may be taken without being limited to this order. Further, it is also possible to perform the extraction processing for a plurality of dimensions in parallel, without limiting to the above processing procedure.

[Effects of First Embodiment]



[0081] Next, advantageous effects of the first embodiment of the present invention will be described.

[0082] A first advantageous effect is that the discrimination capability, which is a degree of discriminating different images, of an image signature constituted of feature vectors of a plurality of dimensions can be improved. In particular, this effect is significant with respect to an image having large correlation between local regions of the image.

[0083] This is because as the shapes of the regions for extracting the features are different among the dimensions (the shapes of the regions are variable), correlation among the dimensions can be reduced.

[0084] A second advantageous effect is that a discrimination capability will not be degraded with respect to an image in which signals concentrate on a particular frequency.

[0085] This is because as the shapes of the regions for extracting the features are different among the dimensions (the shapes of the regions are variable), even with respect to an image in which signals concentrate on a particular frequency, a case where there is no difference between features of all (many) pairs of extraction regions (dimensions) at the same time so that the discrimination capability is deteriorated, is less caused.

[Second Embodiment]


[Configuration of Second Embodiment]



[0086] Next, a second embodiment of the present invention will be described in detail with reference to the drawings.

[0087] The second embodiment of the present invention differs from the first embodiment in that the comparison unit 4 of the first embodiment shown in Fig. 1 is replaced with a comparison unit 4A shown in Fig. 6 in detail. As the components other than the comparison unit 4A are the same as those of the first embodiment, the description of those components is omitted in this embodiment.

[0088] Referring to Fig. 6, the comparison unit 4A includes a difference value calculation unit 43, a quantization boundary determination unit 45, and a quantization unit 44.

[0089] The difference value calculation unit calculates a difference value between a first region feature and a second region feature supplied from the region feature calculation unit 3 for each dimension, and supplies the difference value to the quantization boundary determination unit 45 and the quantization unit 44.

[0090] If the region features are scalar quantities (e.g., an average luminance), the difference value is a scalar quantity obtained by subtracting the second region feature from the first region feature (or vice versa), for example. If the region features are vector quantities, it is also possible to obtain a difference value of scalar quantities after converting the respective vectors into scalar quantities by means of an arbitrary method. Further, if the region features are vector quantities, it is possible to use a difference vector between the first region feature and the second region feature as a difference value (vector quantity).

[0091] When the difference values for all dimensions of the feature vector, supplied from the difference value calculation unit 43, are supplied to the quantization boundary determination unit 45, the quantization boundary determination unit 45 determines the boundary of quantization based on the distribution of the difference values of all dimensions, and supplies information regarding the determined quantization boundary to the quantization unit 44. It should be noted that distribution of the difference values of all dimensions means frequency (probability) of occurrence with respect to the difference values (or the difference vector).

[0092] Further, determination of the boundary of quantization means determination of parameters to be assigned to the quantization indexes exclusively without fail, when quantizing the difference values. If the difference value is a scalar quantity, a value range (that is, a threshold) with respect to each quantization index (quantization level) is determined, and such a value range (threshold) is supplied to the quantization unit 43 as information of the quantization boundary, for example. Alternatively, if the difference value is a vector quantity, a parameter for performing vector quantization for example, that is, a representative vector of the respective quantization indexes, for example, is determined and supplied to the quantization unit 44 as information of the quantization boundary.

[0093] If the difference value is a scalar quantity and quantization of M value (M=2, 3, ... etc.) is to be performed, the quantization boundary determination unit 45 may determine the value range (threshold) for quantization based on the distribution of the difference values of all dimensions such that the proportions of the respective quantization indexes, with respect to all dimension, become equal.

[0094] For example, as a variation of Expression 1, in the case of performing quantization in 2 values (M=2) using a constant α where the quantization index is +1 if Vn1+α>Vn2 while the quantization index is -1 if Vn1+α≤Vn, the center point of the distribution of the difference values (the point where the integral values of left and right distributions become equal) may be determined to be a threshold α of the quantization such that the proportions of quantization indexes +1 and quantization indexes -1 become equal. Similarly, when performing quantization in M values if the difference values are vector quantities, it is possible to determine regions of the vector space assigned to the respective quantization indexes or determine a representative vector (e.g., center of gravity vector)of the respective quantization indexes when performing vector quantization, such that the proportions of the respective quantization indexes with respect to all dimensions become equal, based on the distribution of the difference vectors of all dimensions. As described above, by allowing the proportions of quantization indexes with respect to all dimensions to be equal (that is, eliminating bias of quantization indexes), the entropy can be increased, so that the identification capability can be improved.

[0095] A comparison and quantization method, in which the quantization boundary determination unit 45 determines the boundary of quantization such that the proportions of the quantization indexes with respect to all dimensions become equal and the quantization unit 44 performs quantization based on the determined boundary, is hereinafter referred to as a comparison and quantization method F.

[0096] Further, if the difference value is a scalar quantity and quantization is performed in ternary values (quantization indexes are +1, 0, -1) by means of Expression 2, for example, the quantization boundary determination unit 45 may determine a threshold th used for quantization to an quantization index 0 indicating no difference (quantization index is set to be 0 if smaller than or equal to this threshold) based on the distribution of difference values of all dimension, and supply the determined threshold th to the quantization unit 44 (in the comparison unit 4 shown in Fig. 4 of the first embodiment, the threshold th has been set beforehand). For example, the quantization boundary determination unit 45 may calculate absolute values of the difference values of all dimensions, sort the calculated values, and set a point at a predetermined proportion from the top or the bottom (such a predetermined proportion is supplied as an input, for example) to be the threshold th (this comparison and quantization method is hereinafter referred to as a comparison and quantization method G). Further, it is also possible to determine the threshold th not by a predetermined proportion but by a manner by which the proportions of quantization indexes of +1, 0, -1 become close to equal (this comparison and quantization method is hereinafter referred to as a comparison and quantization method H). The comparison and quantization method H corresponds to a specific example of the comparison and quantization method F performed in accordance with Expression 2.

[0097] A more specific method of the comparison and quantization method G will be explained with an example where the predetermined percentage is P% (e.g., P = 25%). The absolute values of the difference values of all dimensions (the number of dimensions = N) are sorted in ascending order, and a set of the absolute values, sorted in ascending order, of the difference values is indicated as D(i)={D(0), D(1), D(2), ... D(N-1)}. In this example, the value at a position of P% from the bottom of the order sorted in an ascending manner is D(floor(NP/100)) for example, and the threshold th=D(floor(NP/100)). It should be noted that floor( ) is a function in which the places after the decimal point are truncated.

[0098] The method in the present embodiment can be compared with the case where the comparison unit 4 takes the configuration shown in Fig. 4, as in the first embodiment. While a predetermined threshold th is supplied as an input in the configuration shown in Fig. 4 of the first embodiment, in the above-described method of the second embodiment, the quantization boundary determination unit 45 calculates a threshold th adaptively with respect to the image, based on the distribution of the difference values of all dimensions. As described above, while the threshold th is fixed in the first embodiment, the threshold th is calculated adaptively with respect to the image in the second embodiment. As the threshold th is calculated adaptively with respect to the image, it is possible to prevent the values of the dimensions of a feature vector from being biased to particular quantization indexes (probability of appearance of particular quantization indexes is high) compared with the case where the threshold th is fixed (particularly with respect to an image having less relief), the discrimination capability can be higher.. For example, in the case of using a fixed threshold th as in the first embodiment, quantization indexes become 0 in most of the dimensions (or all of the dimensions) of a feature vector in the image of less relief. However, if an adaptive threshold th of the second embodiment is used, as the threshold is automatically adjusted to a small value with respect to an image having less relief, the case where quantization indexes become 0 in most of the dimensions of the feature vector will not be caused.

[0099] The quantization unit 44 performs quantization based on the difference values of the respective dimensions supplied from the difference value calculation unit 43 and the information of the quantization boundary supplied from the quantization boundary determination unit 45, and outputs quantization indexes.

[0100] It should be noted that the quantization unit 44 must follow the quantization method which has been expected when the quantization boundary determination unit 45 determined the quantization boundary, because there is no point if the quantization unit 44 performs quantization without taking into account the quantization boundary information output from the quantization boundary determination unit 45.

[Operation of Second Embodiment]



[0101] Next, operation of the image signature extraction device according to the second embodiment will be described with reference to the flowchart of Fig. 7. In the flowchart of Fig. 7, a dimension (number) of a feature vector is indicated by "n", and there are N dimensions in total from 1 to N.

[0102]  First, the dimension determination unit 1 determines a dimension 1 as the first dimension (n=1) for extracting a feature vector, and supplies it to the extraction region acquisition unit 2 (step B1).

[0103] Next, the extraction region acquisition unit 2 acquires information indicating a first extraction region and a second extraction region of the dimension n from each-dimension extraction region information supplied as an input, and supplies the information to the region feature representative value calculation unit 3 (step B2).

[0104] Then, the region feature representative value calculation unit 3 calculates a first region feature and a second region feature of the dimension n from the image supplied as an input, and supplies the features to the difference value calculation unit 43 (step B3).

[0105] Then, the difference value calculation unit 43 calculates a difference value between the first region feature and the second region feature of the dimension n, and supplies the difference value to the quantization boundary determination unit 45 and the quantization unit 44 (step B4).

[0106] Then, it is determined whether or not the processing up to calculation of the difference values for all dimensions has been completed (that is, it is determined whether n<N is true or false) (step B5). If the processing up to calculation of the difference values for all dimensions has been completed (that is, if n<N is false), the processing proceeds to step B7. If the processing up to calculation of the difference values for all dimensions has not been completed (that is, if n<N is true), the processing proceeds to step B6. At step B6, the dimension determination unit 1 determines the next dimension for extracting a feature vector (n=n+1), and supplies it to the extraction region acquisition unit 2. Then, the processing returns to step B2.

[0107] It should be noted that although the extraction processing is performed in order from the dimension 1 to the dimension N in this embodiment, any order may be taken without being limited to this order.

[0108] Then, when the difference values for all dimensions of the feature vector supplied from the difference value calculation unit 43 have been supplied, the quantization boundary determination unit 45 determines the boundary of quantization based on the distribution of the difference values of the all dimensions, and supplies the determined quantization boundary information to the quantization unit 44 (step B7).

[0109] Then, at step B8, dimension 1 is set (n=1) as the first dimension for performing quantization (quantization indexes are calculated).

[0110] Then, the quantization unit 44 performs quantization based on the difference value of the dimension n and the quantization boundary supplied from the quantization boundary determination unit 45, and outputs a quantization index (step B9).

[0111] Then, it is determined whether or not output of the quantization indexes for all dimensions has been completed (that is, it is determined whether n<N is true or false) (step B10). If output of the quantization indexes for all dimensions has been completed (that is, if n<N is false), the processing ends. If output of the quantization indexes for all dimensions has not been completed (that is, if n<N is true), the processing proceeds to step B11. At step B11, as a dimension of a feature vector for performing quantization, the next dimension is set (n=n+1). Then, the processing returns to step B9.

[0112] It should be noted that although the extraction processing is performed in order from the dimension 1 to the dimension N in this embodiment, any order may be taken without being limited to this order.

[Effects of Second Embodiment]



[0113] Compares with the first embodiment in which the boundary of quantization is fixed, the second embodiment is different in that the boundary of quantization is calculated adaptively (dynamically) with respect to an image. If the boundary of quantization is fixed as in the first embodiment, there is a case where the values of the dimensions of a feature vector are biased to particular quantization indexes (probability of appearance of particular quantization indexes is high) with respect to a particular image (e.g., a flat image having less relief) (entropy is low), causing a problem that the discrimination capability is deteriorated with respect to such an image. On the other hand, if the boundary of quantization is adaptively (dynamically) calculated with respect to an image as in the second embodiment, as it is possible to prevent the case where the values of the dimensions of a feature vector is biased to particular quantization indexes (probability of appearance of particular quantization indexes is high) with respect to any images, the discrimination capability can be improved.

[Third Embodiment]


[Configuration of Third Embodiment]



[0114] Next, a third embodiment of the present invention will be described in detail with reference to the drawings.

[0115] Referring to Fig. 8, the configuration of the third embodiment of the present invention is different from that of the first embodiment shown in Fig. 1 in that a region feature calculation method acquisition unit 5 is added, and the region feature calculation unit 3 is replaced with a region feature calculation unit 3A including a first region feature calculation unit 31A and a second region feature calculation unit 32A. As the other components are the same as those of the first embodiment, the explanation thereof is omitted in this embodiment. It should be noted that although a combination with the first embodiment is described in this embodiment, a combination with the second embodiment is also acceptable.

[0116] To the region feature calculation method acquisition unit 5, a dimension from the dimension determination unit 1 and each-dimension region feature calculation method information are supplied.

[0117] The each-dimension region feature calculation method information is information indicating a method of calculating a region feature in a dimension, which is associated with each dimension of a feature vector, and a prerequisite is that region feature calculation methods must be different among the dimensions. It should be noted that region feature calculation methods being different includes applying different parameters (threshold or the like) to the identical procedure.

[0118] In this embodiment, region feature calculation methods include various types of methods described in the explanation of the region feature calculation unit 3 of the first embodiment, and the parameters associated thereto.

[0119] It should be noted that the region feature calculation method for each dimension indicated by the each-dimension region feature calculation method information has a minimum condition such that at least one pair of dimensions in which the region feature calculation methods are different should be included in the all dimensions of the feature vector. It is desirable that the number of dimensions in which region feature calculation methods are different is larger, because as the number of such dimensions is larger, the number of dimensions having small correlation between them is smaller in the feature vector, whereby the discrimination capability is higher. For example, the region feature calculation methods may be different in all dimensions in the feature vector.

[0120] It should be noted that the information showing the region feature calculation method for each dimension may take any form, if the method of calculating the region feature is uniquely specified.

[0121] Fig. 9 shows examples of the region feature calculation methods for respective dimensions. As shown in Fig. 9, the region feature calculation methods are different among the dimensions. Further, as shown in the examples of Fig. 9, features of scalar quantities and vector quantities may be mixed (the 1st, 3rd, 5th, 6th, 8th, 9th, 10th, and 12th dimensions are scalar quantities and 2nd, 4th, 7th, and the 11th dimensions are vector quantities).

[0122] Further, between dimensions (pair of dimensions) in which the extraction regions are close to each other, it is desirable that the region feature calculation methods in such dimensions (pair of dimensions) are different from each other. For example, between dimensions (pair of dimensions) in which the extraction regions are close to each other, the region feature calculation methods in such dimensions (pair of dimensions) may be different from each other. In that case, between all dimensions (pairs of dimensions) in which the extraction regions are close to each other, the region feature calculation methods in those dimensions may be different from each other. It is also possible that among the dimensions (pairs of dimensions) in which the extraction regions are close to each other, the region feature calculation methods may be different from each other in the dimensions (pairs of dimensions) of a predetermined proportion or larger, or among the dimensions (pairs of dimensions) in which the extraction regions are close to each other, the region feature calculation methods may be different from each other in at least one pair of dimensions.

[0123]  It should be noted that a pair of dimensions in which the extraction regions are close to each other means a pair of dimensions in which a distance between the extraction regions (first extraction region and second extraction region) of the two dimensions, calculated by a predetermined method, is smaller than a predetermined threshold. The distance between the extraction regions (first extraction region and second extraction region) of two dimensions may be calculated as, for example, an average value of the distances between all pixels included in the extraction regions (first extraction region and second extraction region) of the respective dimensions. The distance may also be calculated as, for example, a minimum value of the distances between all pixels included in the extraction regions (first extraction region and second extraction region) of the respective dimensions. It is also possible to calculate center-of-gravity coordinates of the extraction regions (center-of-gravity coordinate of the first extraction region and the center-of-gravity coordinate of the second extraction region) of the respective dimensions, and calculate the distances between them (four distances are calculated in total), and then define the distance between the extraction regions of the two dimensions as an average value or a minimum value thereof, for example. However, methods of calculating the distance between the extraction regions of two dimensions are not limited to these calculation methods.

[0124] With the region feature calculation methods being different from each other between the dimensions (pairs of dimensions) in which the extraction regions are close to each other, the following advantageous effect can be achieved. Between dimensions in which the extraction regions are close to each other, correlation between the extracted features tends to be large. If the region feature calculation methods in those regions are the same, the correlation between the features is larger, whereby the redundancy in the entire feature vector becomes high, so that the discrimination capability is lowered. As such, by allowing the region feature calculation methods to be different in those regions in order to offset the correlation between the extracted features becoming large due to the fact that the extraction regions are close to each other, the correlation between the features can be reduced, whereby the redundancy in the entire feature vector can be low, so that it is possible to prevent deterioration in the discrimination capability.

[0125] From an opposite point of view, between dimensions (pair of dimensions) in which the region feature calculation methods are the same, it is desirable that a distance between the extraction regions in those dimensions (pair of dimensions) is large. For example, between dimensions (pair of dimensions) in which the region feature calculation methods are the same, the distance between such dimensions (pair of dimensions) may be larger than a predetermined threshold. In that case, between all dimensions (pairs of dimensions) in which the region feature calculation methods are the same, the distances between those dimensions (pairs of dimensions) may be larger than a predetermined threshold. It is also possible that among the dimensions (pairs of dimensions) in which the region feature calculation methods are the same, the distances between dimensions (pairs of dimensions) may be larger than a predetermined threshold in the dimensions of a predetermined proportion or larger, or the distance between the pairs of the extraction regions may be larger than a predetermined threshold in at least one pair of dimensions.

[0126] Further, between dimensions (pair of dimensions) in which the extraction regions are adjacent to each other, it is desirable that the region feature calculation methods in such dimensions (pair of dimensions) are different from each other. It should be noted that dimensions (pair of dimensions) in which the extraction regions are adjacent means that partial peripheries of the extraction regions contact each other. For example, between dimensions (pair of dimensions) in which the extraction regions are adjacent, the region feature calculation methods in such dimensions (pair of dimensions) may be different from each other. In that case, between all dimensions (pairs of dimensions) in which the extraction regions are adjacent, the region feature calculation methods in those dimensions may be different from each other. It is also possible that among the dimensions (pairs of dimensions) in which the extraction regions are adjacent, the region feature calculation methods may be different from each other in the dimensions (pairs of dimensions) in a predetermined proportion or larger, or among the dimensions (pairs of dimensions) in which the extraction regions are adjacent, the region feature calculation methods may be different from each other in at least one pair of dimensions.

[0127] With the region feature calculation methods being different from each other between the dimensions (pairs of dimensions) in which the extraction regions are adjacent to each other, the following advantageous effect can be achieved. Between dimensions in which the extraction regions are adjacent, correlation between the extracted features tends to be large. This is because the distance between the extraction regions of the two dimensions is close between the dimensions in which the extraction regions are adjacent (correlation is high between close regions within an image). If the region feature calculation methods in those regions are the same, the correlation between the features is larger, whereby the redundancy in the entire feature vector becomes high, so that the discrimination capability is lowered. As such, by allowing the region feature calculation methods to be different in those regions in order to offset the correlation between the extracted features becoming large due to the fact that the extraction regions are adjacent, the correlation between the features can be reduced, whereby the redundancy in the entire feature vector can be low, so that it is possible to prevent deterioration in the discrimination capability.

[0128] From an opposite point of view, between dimensions (pair of dimensions) in which the region feature calculation methods are the same, it is desirable that the extraction regions in those dimensions (pair of dimensions) are not adjacent to each other. For example, between dimensions (pair of dimensions) in which the region feature calculation methods are the same, the extraction regions in such dimensions (pair of dimensions) may not be adjacent. In that case, between all dimensions (pairs of dimensions) in which the region feature calculation methods are the same, the extraction regions in those dimensions (pairs of dimensions) may not be adjacent. It is also possible that among the dimensions (pairs of dimensions) in which the region feature calculation methods are the same, the extraction regions may not be adjacent in the dimensions (pairs of dimensions) in a predetermined proportion or larger, or the extraction regions may not be adjacent in at least one pair of dimensions.

[0129] Further, between dimensions (pair of dimensions) in which the extraction regions partially overlap, it is desirable that the region feature calculation methods in such dimensions (pair of dimensions) are different from each other. For example, between dimensions (pair of dimensions) in which the extraction regions partially overlap, the region feature calculation methods in such dimensions (pair of dimensions) may be different from each other. In that case, between all dimensions (pairs of dimensions) in which the extraction regions partially overlap, the region feature calculation methods in those dimensions may be different from each other. It is also possible that among the dimensions (pairs of dimensions) in which the extraction regions partially overlap, the region feature calculation methods may be different from each other in the dimensions (pairs of dimensions) in a predetermined proportion or larger, or among the dimensions (pairs of dimensions) in which the extraction regions partially overlap, the region feature calculation methods may be different from each other in at least one pair of dimensions.

[0130] With the region feature calculation methods being different from each other between the dimensions (pairs of dimensions) in which the extraction regions partially overlap, the following advantageous effect can be achieved. Between dimensions in which the extraction regions partially overlap, correlation between the extracted features tends to be large. This is because between the dimensions in which the extraction regions partially overlap, the regions used for calculating the features are shared partially. If the region feature calculation methods in those regions are the same, the correlation between the features is larger, whereby the redundancy in the entire feature vector becomes high, so that the discrimination capability is lowered. As such, by allowing the region feature calculation methods to be different in those regions in order to offset the correlation between the extracted features becoming large due to the fact that the extraction regions partially overlap, the correlation between the features can be reduced, whereby the redundancy in the entire feature vector can be low, so that it is possible to prevent deterioration in the discrimination capability.

[0131] With the region feature calculation methods being different from each other between the dimensions (pairs of dimensions) in which the extraction regions partially overlap, another advantageous effect can be achieved. With the region feature calculation methods being different from each other between the dimensions (pairs of dimensions) in which the extraction regions partially overlap, there is an advantage that tolerance against falsification (particularly, malicious falsification) of an image can be enhanced. Malicious falsification to an image is conducted by changing the pixel value of a partial region of the image, for example, in an attempt to deteriorate accuracy in identity determination by the image signature (feature vector) extracted from the image. Between the dimensions in which the extraction regions are partially overlap, if the overlapped region of the image is falsified, that is, if the pixel value of the overlapped region is changed, the feature (region feature) extracted therefrom is affected, whereby it is highly likely that the feature may be different from the feature extracted from the region which was not falsified. If the region feature calculation methods in those dimensions are the same, the features extracted from the extraction regions of the respective dimensions are affected in a similar manner, whereby the possibility that the features are changed at the same time would be high, so that the robustness is low. If the region feature calculation methods in those dimensions are different, even if the overlapped region is falsified, the possibility that the features are affected in a similar manner so that the features are changed at the same time can be reduced, whereby the robustness can be secured. As such, it becomes difficult to falsify the overlapped region by changing the features of a plurality of dimensions sharing the overlapped region at the same time. Accordingly, with the region feature calculation methods being different from each other between the dimensions (pairs of dimensions) in which the extraction regions partially overlap, tolerance against falsification in an image can be enhanced.

[0132] From an opposite point of view, between dimensions (pair of dimensions) in which the region feature calculation methods are the same, it is desirable that the extraction regions in those dimensions (pair of dimensions) do not overlap each other. For example, between dimensions (pair of dimensions) in which the region feature calculation methods are the same, the extraction regions in such dimensions (pair of dimensions) may not overlap each other. In that case, between all dimensions (pairs of dimensions) in which the region feature calculation methods are the same, the extraction regions in those dimensions (pairs of dimensions) may not overlap each other. It is also possible that among the dimensions (pairs of dimensions) in which the region feature calculation methods are the same, the extraction regions may not overlap each other in the dimensions (pairs of dimensions) of a predetermined proportion or larger, or among the dimensions (pairs of dimensions) in which the region feature calculation methods are the same, the extraction regions may not overlap each other in at least one pair of dimensions.

[0133] From a similar point of view, between dimensions (pair of dimensions) in which the region feature calculation methods are the same, it is desirable that the combinations of the shapes of the pairs of extraction regions are different. Further, between dimensions (pair of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, it is desirable that the region feature calculation methods are different. For example, between dimensions (pair of dimensions) in which the region feature calculation methods are the same, the combinations of the shapes of the pairs of extraction regions may be different. Further, between dimensions (pair of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the region feature calculation methods may be different. In that case, between all dimensions (pairs of dimensions) in which the region feature calculation methods are the same, the combinations of the shapes of the pairs of extraction regions may be different. Further, between all dimensions (pairs of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the region feature calculation methods may be different. It is also possible that among the dimensions (pairs of dimensions) in which the region feature calculation methods are the same, the combinations of the shapes of the pairs of extraction regions may be different in the dimensions of a predetermined proportion or larger. Further, it is also possible that among the dimensions (pairs of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the region feature calculation methods may be different in the dimensions of a predetermined proportion or larger. It is also possible that among the dimensions (pairs of dimensions) in which the region feature calculation methods are the same, the combinations of the shapes of the pairs of extraction regions may be different in at least one pair of dimensions. Further, it is also possible that among the dimensions (pairs of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the region feature calculation methods may be different in at least one pair of dimensions.

[0134] The region feature calculation method acquisition unit 5 acquires, from the each-dimension region feature calculation method information supplied as an input, information indicating the region feature calculation method associated with the dimension supplied from the dimension determination unit 1, and supplies the information to the region feature calculation unit 3A.

[0135] The region feature calculation unit 3A calculates, based on information indicating a first extraction region and a second extraction region supplied from the extraction region acquisition unit for each dimension, a feature of the first extraction region and a feature of the second extraction region as a first region feature and a second region feature, respectively, from the image supplied as an input, according to the information indicating the region feature calculation method supplied from the region feature calculation method acquisition unit 5, and supplies the features to the comparison unit 4.

[0136] In the region feature calculation unit 3A, it is necessary that the dimension of the information indicating the extraction region to be supplied and the dimension of the information indicating the region feature calculation method are synchronized.

[Operation of Third Embodiment]



[0137] Next, with reference to the flowchart of Fig. 10, operation of the image signature extraction device according to the third embodiment will be described. In the flowchart of Fig. 10, a dimension (number) of a feature vector is indicated by "n", and there are N dimensions in total from 1 to N.

[0138] First, the dimension determination unit 1 determines a dimension 1 as the first dimension (n=1) for extracting a feature vector, and supplies it to the extraction region acquisition unit 2 and the region feature calculation method acquisition unit 5 (step C1). Next, the extraction region acquisition unit 2 acquires information indicating a first extraction region and a second extraction region of the dimension n from each-dimension extraction region information supplied as an input, and supplies the information to the region feature calculation unit 3A (step C2).

[0139] Then, the region feature calculation method acquisition unit 5 acquires information indicating the region feature calculation method corresponding to the dimension n from the each-dimension region feature calculation method information supplied as an input, and supplies the information to the region feature calculation unit 3A (step C3).

[0140] Then, the region feature calculation unit 3A calculates a first region feature and a second region feature of the dimension n from the image supplied as an input, and supplies the features to the comparison unit 4 (step C4). Then, the comparison unit 4 compares the first region feature with the second region feature of the dimension n, quantizes the comparison result, and outputs a quantization index (stepC5). Then, it is determined whether or not output of the quantization indexes for all dimensions has been completed (step C6). If output of the quantization indexes for all dimensions has been completed, the processing ends. If output of the quantization indexes for all dimensions has not been completed, the processing proceeds to step C7. At step C7, the dimension determination unit 1 determines the next dimension for extracting a feature vector (n=n+1), and supplies it to the extraction region acquisition unit 2 and the region feature calculation method acquisition unit 5. Then, the processing returns to step C2.

[0141] It should be noted that although the extraction processing is performed in order from the dimension 1 to the dimension N in this embodiment, any order may be taken without being limited to this order. Further, it is also possible to perform the extraction processing for a plurality of dimensions in parallel, without limiting to such processing procedures. Further, step C2 and step C3 may be in reverse order.

[Effect of Third Embodiment]



[0142] In addition to the advantageous effects of the first embodiment, the third embodiment has an advantageous effect that the discrimination capability which is a degree of discriminating different images can be further improved.

[0143] This is because as the region feature calculation methods are different among the dimensions (variable region feature calculation methods are used), correlation among dimensions can be small.

[Fourth Embodiment]


[Configuration of Fourth Embodiment]



[0144] Next, a fourth embodiment of the present invention will be described in detail with reference to the drawings.

[0145] Referring to Fig. 11, the configuration of the fourth embodiment of the present invention is different from that of the first embodiment shown in Fig. 1 in that a comparison method acquisition unit 6 is added, and the comparison unit 4 is replaced with a comparison unit 4B. As the other components are the same as those of the first embodiment, the explanation thereof is omitted in this embodiment. It should be noted that although a combination with the first embodiment is described in this embodiment, a combination with the second embodiment and a combination with the third embodiment are also acceptable.

[0146] To the comparison unit acquisition unit 6, a dimension from the dimension determination unit 1 and each-dimension comparison method information are supplied.

[0147] The each-dimension comparison and quantization method information is information indicating a method of comparing region features in a dimension which is associated with each dimension of a feature vector and performing quantization, and a prerequisite is that comparison and quantization methods must be different among the dimensions. It should be noted that comparison and quantization methods being different includes applying different parameters (threshold, quantization index number, or the like) to the identical procedure.

[0148] In this embodiment, comparison and quantization methods include various types of comparison and quantization methods described in the explanation of the comparison unit 4 of the first embodiment, and the parameters associated thereto (threshold, the number of quantization indexes, or the like), and various types of comparison and quantization methods described in the explanation of the comparison unit 4A of the second embodiment, and the parameters associated thereto (threshold, the number of quantization indexes, or the like).

[0149] It should be noted that the comparison and quantization method for each dimension indicated by the each-dimension comparison and quantization method information has a minimum condition that at least one pair of dimensions in which the comparison and quantization method is different should be included in the all dimensions of the feature vector. It is desirable that the number of dimensions in which comparison and quantization methods are different is larger, because as the number of such dimensions is larger, the number of dimensions having small correlation between them is larger in the feature vector, whereby the discrimination capability becomes higher. For example, the comparison and quantization methods may be different in all dimensions in the feature vector.

[0150] It should be noted that the information showing the comparison and quantization method for each dimension may take any form if the method of comparing and quantizing the region feature is uniquely specified.

[0151] Fig. 12 shows examples of the comparison and quantization methods for respective dimensions. As shown in Fig. 12, comparison and quantization methods are different among the dimensions. Further, different parameters (thresholds th) may be set in the same comparison and quantization methods as in the 3rd, 5th, and 12th dimensions. It should be noted that the examples of comparison and quantization methods for the respective dimensions shown in Fig. 12 are associated with the region feature calculation methods for the respective dimensions shown in Fig. 9. As such, comparison and quantization methods for scalar quantities are shown as examples for the region features of scalar quantities, and comparison and quantization methods for vector quantities are shown as examples for the region features of vector quantities.

[0152] Further, between dimensions (pair of dimensions) in which the extraction regions are close to each other, it is desirable that the comparison and quantization methods in such dimensions (pair of dimensions) are different from each other. For example, between dimensions (pair of dimensions) in which the extraction regions are close to each other, the comparison and quantization methods in such dimensions (pair of dimensions) may be different from each other. In that case, between all dimensions (pairs of dimensions) in which the extraction regions are close to each other, the comparison and quantization methods in those dimensions may be different from each other. It is also possible that among the dimensions (pairs of dimensions) in which the extraction regions are close to each other, the comparison and quantization methods may be different from each other in the dimensions (pairs of dimensions) of a predetermined proportion or larger, or among the dimensions (pairs of dimensions) in which the extraction regions are close to each other, the comparison and quantization methods may be different from each other in at least one pair of dimensions.

[0153] It should be noted that a pair of dimensions in which the extraction regions are close to each other means a pair of dimensions in which a distance between the extraction regions (first extraction region and second extraction region) of the two dimensions, calculated by a predetermined method, is smaller than a predetermined threshold. The distance between the extraction regions (first extraction region and second extraction region) of two dimensions may be calculated as, for example, an average value of the distances between all pixels included in the extraction regions (first extraction region and second extraction region) of the respective dimensions. The distance may also be calculated as, for example, a minimum value of the distances between all pixels included in the extraction regions (first extraction region and second extraction region) of the respective dimensions. It is also possible to calculate center-of-gravity coordinates of the extraction regions (center-of-gravity coordinate of the first extraction region and the center-of-gravity coordinate of the second extraction region) of the respective dimensions, and calculate the distances between them (four distances are calculated in total), and then define the distance between the extraction regions of the two dimensions as an average value or a minimum value thereof, for example. However, methods of calculating the distance between the extraction regions of two dimensions are not limited to these calculation methods.

[0154] With the comparison and quantization methods being different from each other between the dimensions (pairs of dimensions) in which the extraction regions are close to each other, the following advantageous effect can be achieved. Between dimensions in which the extraction regions are close to each other, correlation between the extracted features tends to be large. If the comparison and quantization methods in those regions are the same, the correlation between the features is larger, whereby the redundancy in the entire feature vector becomes high, so that the discrimination capability is lowered. As such, by using different comparison and quantization methods in those regions in order to offset the correlation between the extracted features becoming large due to the fact that the extraction regions are close to each other, the correlation between the features can be reduced, whereby the redundancy in the entire feature vector can be low, so that it is possible to prevent deterioration in the discrimination capability.

[0155] From an opposite point of view, between dimensions (pair of dimensions) in which the comparison and quantization methods are the same, it is desirable that a distance between the extraction regions in those dimensions (pair of dimensions) is large. For example, between dimensions (pair of dimensions) in which the comparison and quantization methods are the same, the distance between such dimensions (pair of dimensions) may be larger than a predetermined threshold. In that case, between all dimensions (pairs of dimensions) in which the comparison and quantization methods are the same, the distances between those dimensions (pairs of dimensions) may be larger than a predetermined threshold. It is also possible that among the dimensions (pairs of dimensions) in which the comparison and quantization methods are the same, the distances between dimensions (pairs of dimensions) may be larger than a predetermined threshold in the dimensions of a predetermined proportion or larger, or among the dimensions (pairs of dimensions) in which the comparison and quantization methods are the same, the distance between dimensions (pairs of dimension) may be larger than a predetermined threshold in at least one pair of dimensions.

[0156] Further, between dimensions (pair of dimensions) in which the extraction regions are adjacent to each other, it is desirable that the comparison and quantization methods in such dimensions (pair of dimensions) are different from each other. It should be noted that dimensions (pair of dimensions) in which the extraction regions are adjacent means that partial peripheries of the extraction regions contact each other. For example, between dimensions (pair of dimensions) in which the extraction regions are adjacent, the comparison and quantization methods in such dimensions (pair of dimensions) may be different from each other. In that case, between all dimensions (pairs of dimensions) in which the extraction regions are adjacent, the comparison and quantization methods in those dimensions may be different from each other. It is also possible that among the dimensions (pairs of dimensions) in which the extraction regions are adjacent, the comparison and quantization methods may be different from each other in the dimensions (pairs of dimensions) of a predetermined proportion or larger, or among the dimensions (pairs of dimensions) in which the extraction regions are adjacent, the comparison and quantization methods between dimensions may be different from each other in at least one pair of dimensions.

[0157] With the comparison and quantization methods being different from each other between the dimensions (pairs of dimensions) in which the extraction regions are adjacent to each other, the following advantageous effect can be achieved. Between dimensions in which the extraction regions are adjacent, correlation between the extracted features tends to be large. This is because the distance between the extraction regions of the two dimensions is close between the dimensions in which the extraction regions are adjacent to each other (correlation is high between close regions within an image). If the comparison and quantization methods in those regions are the same, the correlation between the features is larger, whereby the redundancy in the entire feature vector becomes high, so that the discrimination capability is lowered. As such, by using different comparison and quantization methods in those regions in order to offset the correlation between the extracted features becoming large due to the fact that the extraction regions are adjacent, the correlation between the features can be reduced, whereby the redundancy in the entire feature vector can be low, so that it is possible to prevent deterioration in the discrimination capability.

[0158] From an opposite point of view, between dimensions (pair of dimensions) in which the comparison and quantization methods are the same, it is desirable that the extraction regions in those dimensions (pair of dimensions) are not adjacent to each other. For example, between dimensions (pair of dimensions) in which the comparison and quantization methods are the same, the extraction regions in such dimensions (pair of dimensions) may not be adjacent. In that case, between all dimensions (pairs of dimensions) in which the comparison and quantization methods are the same, the extraction regions in those dimensions (pairs of dimensions) may not be adjacent. It is also possible that among the dimensions (pairs of dimensions) in which the comparison and quantization methods are the same, the extraction regions in the dimensions (pairs of dimensions) may not be adjacent in the dimensions of a predetermined proportion or larger, or among the dimensions (pairs of dimensions) in which the comparison and quantization methods are the same, the extraction regions may not be adjacent in at least one pair of dimensions.

[0159] Further, between dimensions (pair of dimensions) in which the extraction regions partially overlap, it is desirable that the comparison and quantization methods in such dimensions (pair of dimensions) are different from each other. For example, between dimensions (pair of dimensions) in which the extraction regions partially overlap, the comparison and quantization methods in such dimensions (pair of dimensions) may be different from each other. In that case, between all dimensions (pairs of dimensions) in which the extraction regions partially overlap, the comparison and quantization methods in those dimensions may be different from each other. It is also possible that among the dimensions (pairs of dimensions) in which the extraction regions partially overlap, the comparison and quantization methods may be different from each other in the dimensions (pairs of dimensions) of a predetermined proportion or larger, or among the dimensions (pairs of dimensions) in which the extraction regions partially overlap, the comparison and quantization methods may be different from each other in at least one pair of dimensions.

[0160] With the comparison and quantization methods being different from each other between the dimensions (pairs of dimensions) in which the extraction regions partially overlap, the following advantageous effect can be achieved. Between dimensions in which the extraction regions partially overlap, correlation between the extracted features tends to be large. This is because between the dimensions in which the extraction regions partially overlap, the regions used for calculating the features are shared partially. If the comparison and quantization methods in those regions are the same, the correlation between the features is larger, whereby the redundancy in the entire feature vector becomes high, so that the discrimination capability is lowered. As such, by using different comparison and quantization methods in those regions in order to offset the correlation between the extracted features becoming large due to the fact that the extraction regions partially overlap, the correlation between the features can be reduced, whereby the redundancy in the entire feature vector can be low, so that it is possible to prevent deterioration in the discrimination capability.

[0161] With the comparison and quantization methods being different from each other between the dimensions (pairs of dimensions) in which the extraction regions partially overlap, another advantageous effect can be achieved. With the comparison and quantization methods being different from each other between the dimensions (pairs of dimensions) in which the extraction regions partially overlap, there is an advantage that tolerance against falsification (particularly, malicious falsification) in an image can be enhanced. Malicious falsification to an image is conducted by changing the pixel value of a partial region of the image, for example, in an attempt to deteriorate accuracy in identity determination by the image signature (feature vector) extracted from the image. Between the dimensions in which the extraction regions are partially overlap, if the overlapped region of the image is falsified, that is, if the pixel value of the overlapped region is changed, the feature (region feature) extracted therefrom is affected, whereby it is highly likely that the feature may be different from the feature extracted from the region which was not falsified. If the comparison and quantization methods in those dimensions are the same, the features extracted from the extraction regions of the respective dimensions are affected in a similar manner, whereby the possibility that the features are changed at the same time would be high, so that the robustness is low. If the comparison and quantization methods in those dimensions are different, even if the overlapped region is falsified, the possibility that the features are affected in a similar manner so that the features are changed at the same time can be reduced, whereby the robustness can be secured. As such, it becomes difficult to falsify the overlapped region by changing the features of a plurality of dimensions sharing the overlapped region at the same time. Accordingly, by using different comparison and quantization methods between the dimensions (pairs of dimensions) in which the extraction regions partially overlap, tolerance against falsification (particularly, malicious falsification) in an image can be enhanced.

[0162] From an opposite point of view, between dimensions (pair of dimensions) in which the comparison and quantization methods are the same, it is desirable that the extraction regions in those dimensions (pair of dimensions) do not overlap each other. For example, between dimensions (pair of dimensions) in which the comparison and quantization methods are the same, the extraction regions in such dimensions (pair of dimensions) may not overlap each other. In that case, between all dimensions (pairs of dimensions) in which the comparison and quantization methods are the same, the extraction regions in those dimensions (pairs of dimensions) may not overlap each other. It is also possible that among the dimensions (pairs of dimensions) in which the comparison and quantization methods are the same, the extraction regions may not overlap each other in the dimensions (pairs of dimensions) in a predetermined proportion or larger, or among the dimensions (pairs of dimensions) in which the comparison and quantization methods are the same, the extraction regions may not overlap each other in at least one pair of dimensions.

[0163] From a similar point of view, between dimensions (pair of dimensions) in which the comparison and quantization methods are the same, it is desirable that the combinations of the shapes of the pairs of extraction regions are different. Further, between dimensions (pair of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, it is desirable that the comparison and quantization methods are different. For example, between dimensions (pair of dimensions) in which the comparison and quantization methods are the same, the combinations of the shapes of the pairs of extraction regions may be different. Further, between dimensions (pair of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the comparison and quantization methods may be different. In that case, between all dimensions (pairs of dimensions) in which the comparison and quantization methods are the same, the combinations of the shapes of the pairs of extraction regions may be different. Further, between all dimensions (pairs of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the comparison and quantization methods may be different. It is also possible that among the dimensions (pairs of dimensions) in which the comparison and quantization methods are the same, the combinations of the shapes of the pairs of extraction regions may be different in the dimensions of a predetermined proportion or larger. Further, it is also possible that among the dimensions (pairs of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the comparison and quantization methods may be different in the dimensions of a predetermined proportion or larger. It is also possible that among the dimensions (pairs of dimensions) in which the comparison and quantization methods are the same, the combinations of the shapes of the pairs of extraction regions may be different in at least one pair of dimensions. Further, it is also possible that among the dimensions (pairs of dimensions) in which the combinations of the shapes of the pairs of extraction regions are the same, the comparison and quantization methods may be different in at least one pair of dimensions.

[0164] From a similar point of view, between dimensions (pair of dimensions) in which the comparison and quantization methods are the same, it is desirable that the region feature calculation methods are different. Further, between dimensions (pair of dimensions) in which the region feature calculation methods are the same, it is desirable that the comparison and quantization methods are different. For example, between dimensions (pair of dimensions) in which the comparison and quantization methods are the same, the region feature calculation methods may be different. Further, between dimensions (pair of dimensions) in which region feature calculation methods are the same, the comparison and quantization methods may be different In that case, between all dimensions (pairs of dimensions) in which the comparison and quantization methods are the same, the region feature calculation methods may be different. Further, between all dimensions (pairs of dimensions) in which the region feature calculation methods are the same, the comparison and quantization methods may be different. It is also possible that among the dimensions (pairs of dimensions) in which the comparison and quantization methods are the same, the region feature calculation methods may be different in the dimensions of a predetermined proportion or larger. Further, it is also possible that among the dimensions (pairs of dimensions) in which the region feature calculation methods are the same, the comparison and quantization methods may be different in the dimensions of a predetermined proportion or larger. It is also possible that among the dimensions (pairs of dimensions) in which the comparison and quantization methods are the same, the region feature calculation methods may be different in at least one pair of dimensions. Further, it is also possible that among the dimensions (pairs of dimensions) in which the region feature calculation methods are the same, the comparison and quantization methods may be different in at least one pair of dimensions.

[0165] The comparison method acquisition unit 6 acquires, from the each-dimension comparison and quantization method information supplied as an input, information indicating the comparison and quantization method corresponding to the dimension supplied from the dimension determination unit 1, and supplies the information to the comparison unit 4B.

[0166] The comparison unit 4B compares a first region feature with a second region feature supplied from the region feature calculation unit 3 for each dimension and quantizes according to the information indicating the comparison and quantization method supplied from the comparison method acquisition unit 6, and outputs a quantization index. The comparison unit 4B may have a configuration including both the comparison unit 4 of the first embodiment and the comparison unit 4B of the second embodiment if necessary, depending on the comparison and quantization method.

[0167]  In the comparison unit 4B, it is necessary that the dimension of the region feature to be supplied and the dimension of the information indicating the comparison and quantization method are synchronized.

[Operation of Fourth Embodiment]



[0168] Next, with reference to the flowchart of Fig. 13, operation of the image signature extraction device according to the fourth embodiment will be described. In the flowchart of Fig. 13, a dimension (number) of a feature vector is indicated by "n", and there are N dimensions in total from 1 to N.

[0169] First, the dimension determination unit 1 determines a dimension 1 as the first dimension (n=1) for extracting a feature vector, and supplies it to the extraction region acquisition unit 2 and the comparison method acquisition unit 6 (step D1). Next, the extraction region acquisition unit 2 acquires information indicating a first extraction region and a second extraction region of the dimension n from each-dimension extraction region information supplied as an input, and supplies it to the region feature calculation unit 3 (step D2).

[0170] Then, the comparison method acquisition unit 6 acquires information indicating the comparison and quantization method corresponding to the dimension n from the each-dimension comparison and quantization method information supplied as an input, and supplies it to the comparison unit 4B (step D3).

[0171]  Then, the region feature calculation unit 3 calculates a first region feature and a second region feature of the dimension n from the image supplied as an input, and supplies the features to the comparison unit 4B (step D4). Then, the comparison unit 4B compares the first region feature with the second region feature of the dimension n, quantizes the comparison result, and outputs a quantization index (stepD5). Then, it is determined whether or not output of the quantization indexes for all dimensions has been completed (step D6). If output of the quantization indexes for all dimensions has been completed, the processing ends. If output of the quantization indexes for all dimensions has not been completed, the processing proceeds to step D7. At step D7, the dimension determination unit 1 determines the next dimension for extracting a feature vector (n=n+ 1), and supplies it to the extraction region acquisition unit 2 and the comparison method acquisition unit 6. Then, the processing returns to step D2.

[0172] It should be noted that although the extraction processing is performed in order from the dimension 1 to the dimension N in this embodiment, any order may be taken without being limited to this order. Further, it is also possible to perform the extraction processing for a plurality of dimensions in parallel, without limiting to such processing procedures. Further, step D2 and step D3 may be in reverse order, and step D3 may be performed immediately before step D5.

[Effect of Fourth Embodiment]



[0173] In addition to the advantageous effects of the first embodiment, the fourth embodiment has an advantageous effect that the discrimination capability which is a degree of discriminating different images can be further improved.

[0174] This is because as the comparison and quantization methods are different among the dimensions (variable comparison and quantization methods are used), correlation between dimensions can be small.

[0175] Next, fifth to seventh embodiments of the present invention will be described.

[Fifth Embodiment]



[0176] In the fifth embodiment, the number of dimensions of a feature vector to be extracted is 300 dimensions (from 1st dimension to 300th dimension).

[0177] In the fifth embodiment, the extraction regions (first extraction regions and second extraction regions) for respective dimensions are formed of quadrangles in various shapes. Fig. 14 shows each-dimension extraction region information to be supplied to the extraction region acquisition unit 2 as an input, in the fifth embodiment. Fig. 14 shows XY coordinate values of the four corners of the quadrangles of the extraction regions (first extraction regions and second extraction regions) for the respective dimensions, with respect to the image size of 320 pixel wide and 240 pixel long, which is a prescribed image size. For example, the extraction regions for a 1st dimension is formed of a first extraction region consisting of a quadrangle with four corners having a coordinate value (262.000, 163.000), a coordinate value (178.068, 230.967), a coordinate value (184.594, 67.411), and a coordinate value (100.662, 135.378), and a first extraction region consisting of a quadrangle with four corners having a coordinate value (161.000, 133.000), a coordinate value (156.027. 132.477), a coordinate value (164.240, 102.170), and a coordinate value (159.268, 101.647).

[0178] The extraction regions (first extraction region and second extraction region) for each dimension is a set of pixels having coordinate values of integer values included in a region defined by these coordinate values of the four corners, with respect to the image normalized to an image size of 320 pixel wide and 240 pixel long. However, negative coordinate values included in the region defined by the four corners are not included in the extraction region.

[0179] Fig. 15 shows each-dimension region feature calculation method information supplied to the region feature calculation method acquisition unit 5 as an input, in the fifth embodiment. In the fifth embodiment, an average luminance of a group of pixels included in each of the extraction region (first extraction region and second extraction region) serves as a region feature of each of the extraction regions, for every dimension.

[0180] Fig. 17 shows each-dimension comparison and quantization method information supplied to the comparison method acquisition unit 6 as an input, in the fifth embodiment. In the fifth embodiment, the comparison and quantization method B or the comparison and quantization method G is used for each dimension, and the value of the parameter is also different for each dimension. For example, in the 1st dimension, the comparison and quantization method G is used, and the threshold th is D(floor(300*5.0/100)). In the 2nd dimension, the comparison and quantization method G is used, and the threshold th is D(floor(300*10.0/100)). Further, in the 9th dimension, for example, the comparison and quantization method B is used, and the threshold th is 3.0.

[Sixth Embodiment]



[0181] In the sixth embodiment, the number of dimensions of a feature vector to be extracted is 300 dimensions (from 1st dimension to 300th dimension), as in the fifth embodiment. In the sixth embodiment, as each-dimension extraction region information supplied to the extraction region acquisition unit 2 as an input, the information shown in Fig. 14 is used, as in the fifth embodiment. Further, in the sixth embodiment, as each-dimension comparison and quantization method information supplied to the comparison method acquisition unit 6 as an input, the information shown in Fig. 17 is used, as in the fifth embodiment.

[0182] Fig. 16 shows each-dimension region feature calculation method information supplied to the region feature calculation method acquisition unit 5 as an input in the sixth embodiment. In the sixth embodiment, for each dimension, an average luminance of a group of pixels included in the extraction regions (first extraction region and second extraction region) or a percentile luminance feature is used, and the feature is different for each dimension even if the same percentile luminance feature is used. For example, in the 1st dimension, an average luminance of the pixels included in the extraction regions is used. In the 4th dimension, for example, a percentile luminance feature is used, and the value is Y(floor(N*20.0/100)). Further, in the 8th dimension, a percentile luminance feature is used, and the value is Y(floor(N*80.0/100)).

[Seventh Embodiment]



[0183] In the seventh embodiment, the number of dimensions of a feature vector to be extracted is 325 dimensions (1st dimension to 325th dimension). In the seventh embodiment, each region consists of a combination of blocks among 1024 pieces of blocks formed by dividing an image into 32 pieces vertically and 32 pieces horizontally. To the respective blocks, indexes starting from 0 are assigned from the upper left part as shown in Fig. 27, and the regions are described using those indexes. Specifically, a rectangle region is indicated using an index "a" of the upper left block and an index "b" of the lower right block in a manner of "a-b". For example, a rectangle, formed of four blocks having indexes of 0, 1, 32, and 33, is described as 0-33. Further, if rectangles formed in this manner are linked using a sign "|", they represent a region formed by linking the rectangles before and after the sign. For example, 0-33|2-67indicates a region formed by linking a rectangle defined by 0-33 and a rectangle defined by 2-67, that is, a region formed by the block numbers 0, 1, 2, 3, 32, 33, 34, 35, 66, and 67.

[0184] Fig. 25 shows the regions, described in this manner, corresponding to the respective dimensions of the seventh embodiment. In Fig. 25, the 325 dimensions are described by classifying them by the type in Fig. 25-a, Fig. 25-b, Fig. 25-c, Fig. 25-d, Fig. 25-e, Fig. 25-f, and Fig. 25-g. In these figures, the type of a region means a group consisting of the dimensions having similar region patterns determined by a relative position or a combination of shapes between the first and second extraction regions.

[0185]  Specifically, Fig. 25-a corresponds to the case where two regions, formed by dividing a square defined by four blocks vertically and four blocks horizontally into two in a vertical or horizontal direction, are used as first and second extraction regions, an example of which is shown in Fig. 27-a. As such, the shape of both the first and second extraction regions is a rectangle defined by four blocks vertically and two blocks horizontally or a rectangle defined by two blocks vertically and four blocks horizontally. Further, regarding the relative position between the first and second extraction regions, they are present at positions adjacent to each other such that the longitudinal sides of the rectangles overlap each other.

[0186] Fig. 25-b corresponds to the case where two regions, formed by equally dividing a square defined by eight blocks vertically and eight blocks horizontally in vertical and horizontal directions into four squares and combining the upper left and lower right squares and combining the upper right and lower left squares, respectively, are used as a first and second extraction regions, an example of which is shown in Fig. 27-b. As such, the shape of both the first and second extraction regions is that two squares, defined by two blocks vertically and two blocks horizontally, are arranged on a diagonal line at an angle of 45 degrees or 135 degrees so as to share one vertex. Further, regarding the relative position between the regions, the two squares constituting the second region are present at a position adjacent to the left and below of the upper left square of the first region.

[0187] In the case of Fig. 25-c, the shape of both the first and second extraction regions is a square defined by 10 blocks vertically and 10 blocks horizontally, an example of which is shown in Fig. 27-c. Regarding the relative position between the first and second extraction regions, they are present at positions distant by the integral multiple of 10 blocks vertically and horizontally from each other.

[0188] In the case of Fig. 25-d, the shape of both the first and second extraction regions is a square defined by 8 blocks vertically and 8 blocks horizontally, an example of which is shown in Fig. 27-d. Regarding the relative position between the first and second extraction regions, they are present at positions distant by the integral multiple of 6 blocks vertically and horizontally from each other.

[0189] Fig. 25-e corresponds to the case where two regions, formed by dividing a square region into a center portion and an outside portion, are used as a first and second extraction regions, an example of which is shown in Fig. 27-e. As such, the shape of the second extraction region is a square of the center portion, and the shape of the first extraction region is a square in which the second extraction region is cut out from the whole square. Further, regarding the relative position between the regions, the second region is present at the center hole of the first region.

[0190] In the case of Fig. 25-f, the shape of the first extraction region is a rectangle defined by 6 blocks vertically and 10 blocks horizontally, and the shape of the second extraction region is a rectangle defined by 10 blocks vertically and 6 blocks horizontally, an example of which is shown in Fig. 27-f. Regarding the relative position between the first and second extraction regions, they are arranged such that the center positions thereof coincide.

[0191] Fig. 25-g corresponds to the case where two regions, formed by dividing a rectangle defined by 4 blocks vertically and 12 blocks horizontally or a rectangle defined by 12 blocks vertically and 4 blocks horizontally into a center square region formed by trisecting the longitudinal side and the other region, are used as a first and second extraction regions, an example of which is shown in Fig. 27-g. As such, the shape of the first region is in two squares defined by four blocks vertically and four blocks horizontally separated from each other by four blocks vertically or horizontally, and the shape of the second extraction region is a square defined by four blocks vertically and four blocks horizontally. Further, regarding the relative position between the regions, the second region is present between the squares of the first region.

[0192] Hereinafter, the region types of Fig. 25-a, Fig. 25-b, Fig. 25-c, Fig. 25-d, Fig. 25-e, Fig. 25-f, and Fig. 25-g are respectively referred to as a region type a, a region type b, a region type c, a region type d, a region type e, a region type f, and a region type g.

[0193] In the seventh embodiment, an average luminance is calculated as a region feature in each region shown in Fig. 25 and a feature of each dimension is calculated. Of course, it is possible to obtain a value extracted by the previously-described various extraction methods, such as a median or a maximum value, instead of the average luminance as a region feature.

[0194] For quantizing the feature of each dimension, a threshold is set for each of the types of the regions to perform quantization. For example, when quantizing a feature into a ternary value according to Expression 2, a threshold th for quantization is determined such that the proportion of occurrence of 0, 1, and -1 becomes equal for each of the types of regions, and quantization is performed. Specifically, a threshold th is obtained by applying the method described in paragraph 0096 for each type of region, in which P=33.333% and N represents the number of dimensions for each type of region. For example, as N=113 in the region type a, a threshold is calculated by th=D(floor(11333.333/100))=D(37). It should be noted that D(i) (i=0, 1, .., N-1) is a set in which the absolute values of the difference values of the 1st dimension to the 113th dimension are sorted in ascending order. In this case, an index corresponding to the threshold is 37. Similarly, an index corresponding to a threshold can be obtained for another region type, as shown in Fig. 26. By obtaining a threshold for each region type, it is possible to uniform the occurrence probability of 0, 1, and -1 in each dimension, compared with the case of determining a threshold collectively, whereby the discrimination capability is improved. Of course, it is possible to perform quantization by the other various quantization methods which have been described above.

[0195] It should be noted that in the case of the seventh embodiment, it is also possible to first calculate a representative value for each block (e.g., an average luminance pixels within a block) shown in Fig. 24, and then extract a region feature. Thereby, extraction can be performed at a higher speed than the case of directly extracting a region feature from all pixels within the region. Further, extraction regions of each region type have a symmetric property as a whole. As such, even in the cases where right and left or up and down of images are inverted, by changing the correspondence relation and signs of the dimensions appropriately, it is possible to restore the features of the original image from the features extracted from the image in which right and left or up and down are inverted. As such, matching can also be performed on an image in which right and left or up and down are inverted.

[Embodiment of Matching Unit]



[0196] Next, a matching unit which performs matching between quantization index vectors output in the present invention will be described using a block diagram.

[0197] Referring to Fig. 20 showing a block diagram of a matching unit 100 for performing matching between quantization index vectors output in the present invention, the matching unit 100 includes a dimension determination unit 101, quantization value acquisition units 102 and 103, and a scale calculation unit 104.

[0198] The dimension determination unit 101 is connected to the quantization value acquisition units 102 and 103, and outputs determined dimension information. The quantization value acquisition unit 102 acquires, from a first quantization index vector, a quantization index value of the dimension input from the dimension determination unit 101, and outputs the value to the scale calculation unit 104 as a first quantization index value. The quantization value acquisition unit 103 acquires, from a second quantization index vector, a quantization index value of the dimension input from the dimension determination unit 101, and outputs the value to the scale calculation unit 104 as a second quantization index value. The scale calculation unit 104 calculates a scale indicating the identity from the first and second quantization index values output from the quantization value acquisition units 102 and 103, and outputs it.

[0199] Next, operation of the matching unit 100 shown in Fig. 20 will be described.

[0200] First, to the matching unit 100, the first quantization index vector which is a quantization index vector extracted from a first image, and a second quantization index vector which is a quantization index vector extracted from a second image, are input. The input first and second quantization index vectors are respectively input to the quantization value acquisition units 102 and 103.

[0201] To the quantization value acquisition units 102 and 103, dimension information output from the dimension determination unit 101 is also input. The dimension determination unit 101 sequentially outputs information designating respective dimensions of the quantization index vectors which are N dimension vectors. The output order is not necessary incremented by one from 1 to N, and may be in any order if all dimensions from 1 to N are designated without deficiency and excess.

[0202] The quantization value acquisition units 102 and 103 acquire, from the input quantization index vectors, quantization index values of the dimension designated in the dimension information output from the dimension determination unit 101, and output the acquired quantization index values to the scale calculation unit 104.

[0203]  The scale calculation unit 104 compares the first quantization index value output from the quantization value acquisition unit 102 with the second quantization index value. This comparison is performed on the respective dimensions, and a similarity scale (or distance scale) between the first and second quantization index vectors is calculated as an identity scale.

[0204] The acquired identity scale is compared with a predetermined threshold to determine the identity. If the identity scale is a scale indicating a similarity value, they are determined to be identical if the scale value is equal to or larger than the threshold. On the other hand, if the identity scale is a scale indicating the distance, they are determined to be identical if the scale value is smaller than or equal to the threshold.

[0205] Next, operation of the matching unit 100 shown in Fig. 20 will be described using a flowchart. First, operation in the case of using a similarity value as an identity scale will be described.

[0206] Fig. 21 is a flowchart showing the operation of the matching unit 100. In the flowchart of Fig. 21. a dimension (number) of a feature vector is indicated by "n", and there are N dimensions in total from 1 to N. Further, a variable for calculating the similarity value is indicated by B.

[0207] First, the dimension determination unit 101 determines a dimension 1 as the first dimension (n=1) for performing matching on a quantization index vector, and supplies it to the quantization acquisition units 102 and 103 and sets the variable B to be 0 in the scale calculation unit 104 (step S100).

[0208] Then, the quantization acquisition units 102 and 103 acquire a first quantization index value and a second quantization index value of the dimension n from the first quantization index vector and the second quantization index vector, and supply them to the scale calculation unit 104 (step S102).

[0209] Then, the scale calculation unit 104 calculates, from the first quantization index value and the second quantization index value, a similarity value ΔB between the features corresponding to the respective quantization indexes (step S104). For example, ΔB=1 when the quantization indexes conform to each other, and ΔB=0 in other cases. Alternatively, it is also possible to calculate representative values of the features before quantization from the quantization indexes and use a value, which is increased as the difference between the representative values is smaller, as ΔB. In that case, instead of obtaining a difference by calculating the representative values of the features, it is possible to hold a table in which the value of AB is acquired from a combination of quantization index values, and directly obtain the value of ΔB using the table from the combination of the quantization index values.

[0210] Next, the value of AB is added to the variable B (step S106). At this point, if the value of ΔB is 0, it is possible to control not to add, rather than adding 0 to the variable B.

[0211]  Next, it is checked whether the dimension number n reaches the number of dimension N (step S108), and if the number does not reach, the processing moves to step S112, while if the number reached, the value of the variable B at that point is output as an identity scale (scale indicating similarity value) (step S110) and the processing ends.

[0212] At step 112, the dimension determination unit 101 determines the next dimension from n=n+1 as the dimension for acquiring quantization indexes, and supplies it to the quantization value acquisition units 102 and 103. Then, the processing returns to step S102.

[0213] It should be noted that although the extraction processing is performed in order from the dimension 1 to the dimension N, any order may be taken without being limited to this order.

[0214] Next, operation in the case of using a distance as an identity scale will be described.

[0215] Fig. 22 is another flowchart showing the operation of the matching unit 100. Also in the flowchart of Fig. 22, a dimension (number) of a feature vector is indicated by "n", and there are N dimensions in total from 1 to N. Further, a variable for calculating a distance scale is indicated by C.

[0216] While the basic flow is similar to that of Fig. 21, Fig. 22 is different in that steps S100, S104, S106, and S110 are respectively replaced with steps S200, S204, S206, and S210.

[0217]  First, at step S200, the dimension determination unit 101 determines a dimension 1 as the first dimension (n=1) for performing matching on a quantization index vector, and supplies it to the quantization acquisition units 102 and 103 and sets the variable C to be 0 in the scale calculation unit 104.

[0218] At step S204, the scale calculation unit 104 calculates, from the first quantization index value and the second quantization index value, a distance AC between the features corresponding to the respective quantization indexes. For example, ΔC=0 when the quantization indexes conform to each other, and ΔC=1 in other cases. Alternatively, it is also possible to calculate representative values of the features before quantization from the quantization indexes, and use a value, which is decreased as the difference between the representative values is smaller, as ΔC. In that case, instead of obtaining a difference by calculating the representative values of the features, it is possible to hold a table in which the value of ΔC is acquired from a combination of quantization index values, and directly obtain the value of ΔC using the table from the combination of the quantization index values.

[0219] At step S206, the value of AC is added to the variable C. At this point, if the value of ΔC is 0, it is possible to control not to add, rather than adding 0 to the variable C.

[0220] At step S210, the value of the variable C at that point is output as an identity scale (scale indicating distance) and the processing ends.

[0221]  The other steps are the same as those in the case of Fig. 21. However, if the dimension number n reached the number of dimensions N at step S108, the processing moves to step S210.

[0222] It should be noted that although the extraction processing is performed in order from the dimension 1 to the dimension N, any order may be taken without being limited to this order.

[0223] Next, description will be given for the operation in the case where a dimension in which "both quantization indexes are 0" for the first quantization index value and the second quantization index value is eliminated, and a similarity value is used as an identity scale.

[0224] Fig. 23 is another flowchart showing the operation of the matching unit 100. Also in the flowchart of Fig. 23, a dimension (number) of a feature vector is indicated by "n", and there are N dimensions in total from 1 to N. Further, a variable for calculating the similarity value is indicated by B, and a variable for counting the dimensions in which not "both quantization indexes are 0" is indicated by A.

[0225] First, the dimension determination unit 101 determines a dimension 1 as the first dimension (n=1) for performing matching on a quantization index vector, and supplies it to the quantization acquisition units 102 and 103, and sets the variables A and B to be 0 in the scale calculation unit 104 (step S300), and then moves to step S102.

[0226] Step S102 is the same as the case of Fig. 21, and when step S102 ends, the processing moves to step S314.

[0227] At step S314, the scale calculation unit 104 checks whether or not both the first quantization index value and the second quantization index value are 0. If both values are 0, the processing moves to step S108, while if either of them is not 0, the value of the variable A is incremented by one (step S316), and the processing moves to step S104.

[0228] The processing at steps S104, S106, S108, and S112 is the same as that in the case of Fig. 21. If the dimension number n reached the number of dimensions N at step S108, the processing moves to step S310.

[0229] At step S310, the scale calculation unit 104 calculates the value of B/A and outputs it as an identity scale, and ends the processing. However, if A =0. the scale calculation unit 104 outputs a predetermined value (e.g., 0.5).

[0230] It should be noted that although the extraction processing is performed in order from the dimension 1 to the dimension N, any order may be taken without being limited to this order.

[0231] While the embodiments of the present invention have been described above, the present invention is not limited to these examples. It is understood by those skilled in the art that various changes in form and details may be made therein. Further, the image signature extraction device of the present invention is adapted such that the functions thereof can be realized by computers and programs, as well as hardware. Such a program is provided in the form of being written on a computer readable recording medium such as a magnetic disk, a semiconductor memory, or the like, is read by a computer when the computer is started for example, and controls operation of the computer, to thereby allow the computer to function as the dimension determination unit, the extraction region acquisition unit, the region feature calculation unit, the comparison unit, the region feature calculation method acquisition unit, and the comparison method acquisition unit, of the above-described embodiments.

[0232] This application is based upon and claims the benefit of priority from Japanese patent applications No. 2009-061023 filed on March 13, 2009, and No. 2009-097862 filed on April 14, 2009.


Claims

1. An image signature extraction device, comprising:

a region feature calculation unit (3) that calculates, with respect to each of a plurality of dimensions composing an image signature, region features, respectively, from two sub-regions in an image, the image signature being information for identifying the image;

a difference value calculation unit (43) that calculates a difference value between the calculated region features with respect to each of the plurality of dimensions;

a determination unit (45) that calculates absolute values of the difference values with respect to the plurality of dimensions, respectively, sorts, with respect to each type, among a plurality of types, to which a group of dimensions belong, the absolute values in ascending order, and determines, as a threshold of said type, an absolute value at a position of a predetermined percentage from the bottom of the sorted order; and

a quantization unit (44) that quantizes the difference values with respect to each group of dimensions into ternary values by using the threshold of the corresponding type;

wherein a group of dimensions having identical sub-region pair patterns belong to a same type, the sub-region pair pattern being determined by a relative position between the two sub-regions and a combination of shapes of the two sub-regions,

sub-region pairs belonging to the same type are located at different absolute positions,

the ternary values include a first value, a second value which is a value smaller than the first value, and a third value which is a value smaller than the second value,

a difference between the first value and the second value and a difference between the second value and the third value are equal, and

wherein

the two sub-regions are a first sub-region and a second sub-region; and

the types include at least one of:

a first type in which two rectangles, formed by equally dividing a square of a certain size in the image into two in a vertical direction or a horizontal direction, serve as the first sub-region and the second sub-region;

a second type in which among four squares formed by equally dividing a square of a certain size in the image into two in a vertical direction and into two in a horizontal direction, a region formed by combining an upper left square and a lower right square serves as the first sub-region and a region formed by combining an upper right square and a lower left square serves as the second sub-region;

a third type in which the first sub-region and the second sub-region are squares of the same size and are spaced apart in a vertical direction and in a horizontal direction by a distance equal to an integral multiple of one side of the square;

a fourth type in which the first sub-region and the second sub-region are squares of the same size but are smaller than those of the third type, and are spaced apart in a vertical direction and in a horizontal direction by a distance equal to an integral multiple of one side of the square;

a fifth type in which, of two different sized squares sharing the center of gravity, a region obtained by removing the smaller square from the larger square serves as the first sub-region, and the smaller square serves as the second sub-region;

a sixth type in which two rectangles of the same size sharing the center of gravity and being placed orthogonally to each other such that their longitudinal directions are horizontal and vertical, respectively, serve as the first sub-region and the second sub-region; and

a seventh type in which among three vertically or horizontally adjacent squares of the same size, squares at both ends serve as the first sub-region and a square at the center serves as the second sub-region.


 
2. The image signature extraction device, according to Claim 1, wherein with respect to each type,
the threshold is an n-th absolute value in the sorted order, n being floor (N x P/100),
where N is a number of dimensions belonging to said type and P is the predetermined percentage.
 
3. The image signature extraction device, according to Claim 1 or 2, wherein
the two sub-regions are formed of one or more blocks among 1024 blocks obtained by dividing the image into 32 in a vertical direction and dividing the image into 32 in a horizontal direction, and
the square of a certain size of the first type is a square constituted of 4 blocks vertically and 4 blocks horizontally;
the square of a certain size of the second type is a square constituted of 8 blocks vertically and 8 blocks horizontally;
the square of the third type is a square constituted of 10 blocks vertically and 10 blocks horizontally;
the square of the fourth type is a square constituted of 6 blocks vertically and 6 blocks horizontally;
the two different sized squares of the fifth type are a combination of a square constituted of 10 blocks vertically and 10 blocks horizontally and a square constituted of 4 blocks vertically and 4 blocks horizontally, or a combination of a square constituted of 6 blocks vertically and 6 blocks horizontally and a square constituted of 2 blocks vertically and 2 blocks horizontally;
the two rectangles of the sixth type are a rectangle constituted of 10 blocks vertically and 6 blocks horizontally and a rectangle constituted of 6 blocks vertically and 10 blocks horizontally; and
the square of the seventh type is a square constituted of 4 blocks vertically and 4 blocks horizontally.
 
4. A matching device that performs matching using an image signature output by the image signature extraction device according to any of Claims 1 to 3.
 
5. An identifying device that performs identification using an image signature output by the image signature extraction device according to any of Claims 1 to 3.
 
6. An image signature extraction method, comprising:

calculating, with respect to each of a plurality of dimensions composing an image signature, region features, respectively, from two sub-regions in an image, the image signature being information for identifying the image;

calculating a difference value between the calculated region features with respect to each of the plurality of dimensions;

calculating absolute values of the difference values with respect to the plurality of dimensions, respectively, sorting, with respect to each type, among a plurality of types, to which a group of dimensions belong, the absolute values in ascending order, and determining, as a threshold of said type, an absolute value at a position of a predetermined percentage from the bottom of the sorted order; and

quantizing the difference values with respect to each group of dimensions into ternary values by using the threshold of the corresponding type;

wherein a group of dimensions having identical sub-region pair patterns belong to a same type, the sub-region pair pattern being determined by a relative position between the two sub-regions and a combination of shapes of the two sub-regions,

sub-region pairs belonging to the same type are located at different absolute positions,

the ternary values include a first value, a second value which is a value smaller than the first value, and a third value which is a value smaller than the second value,

a difference between the first value and the second value and a difference between the second value and the third value are equal, and

wherein

the two sub-regions are a first sub-region and a second sub-region; and

the types include at least one of:

a first type in which two rectangles, formed by equally dividing a square of a certain size in the image into two in a vertical direction or a horizontal direction, serve as the first sub-region and the second sub-region;

a second type in which among four squares formed by equally dividing a square of a certain size in the image into two in a vertical direction and into two in a horizontal direction, a region formed by combining an upper left square and a lower right square serves as the first sub-region and a region formed by combining an upper right square and a lower left square serves as the second sub-region;

a third type in which the first sub-region and the second sub-region are squares of the same size and are spaced apart in a vertical direction and in a horizontal direction by a distance equal to an integral multiple of one side of the square;

a fourth type in which the first sub-region and the second sub-region are squares of the same size but are smaller than those of the third type, and are spaced apart in a vertical direction and in a horizontal direction by a distance equal to an integral multiple of one side of the square;

a fifth type in which, of two different sized squares sharing the center of gravity, a region obtained by removing the smaller square from the larger square serves as the first sub-region, and the smaller square serves as the second sub-region;

a sixth type in which two rectangles of the same size sharing the center of gravity and being placed orthogonally to each other such that their longitudinal directions are horizontal and vertical, respectively, serve as the first sub-region and the

second sub-region; and
a seventh type in which among three vertically or horizontally adjacent squares of the same size, squares at both ends serve as the first sub-region and a square at the center serves as the second sub-region.


 
7. The image signature extraction method, according to Claim 6, wherein with respect to each type,
the threshold is an n-th absolute value in the sorted order, n being floor (N x P/100),
where N is a number of dimensions belonging to said type and P is the predetermined percentage.
 
8. The image signature extraction method, according to Claim 6 or 7, wherein
the two sub-regions are formed of one or more blocks among 1024 blocks obtained by dividing the image into 32 in a vertical direction and dividing the image into 32 in a horizontal direction, and
the square of a certain size of the first type is a square constituted of 4 blocks vertically and 4 blocks horizontally;
the square of a certain size of the second type is a square constituted of 8 blocks vertically and 8 blocks horizontally;
the square of the third type is a square constituted of 10 blocks vertically and 10 blocks horizontally;
the square of the fourth type is a square constituted of 6 blocks vertically and 6 blocks horizontally;
the two different sized squares of the fifth type are a combination of a square constituted of 10 blocks vertically and 10 blocks horizontally and a square constituted of 4 blocks vertically and 4 blocks horizontally, or a combination of a square constituted of 6 blocks vertically and 6 blocks horizontally and a square constituted of 2 blocks vertically and 2 blocks horizontally;
the two rectangles of the sixth type are a rectangle constituted of 10 blocks vertically and 6 blocks horizontally and a rectangle constituted of 6 blocks vertically and 10 blocks horizontally; and
the square of the seventh type is a square constituted of 4 blocks vertically and 4 blocks horizontally.
 
9. A matching method for performing matching using an image signature extracted by the image signature extraction method according to any of Claims 6 to 8.
 
10. An identifying method for performing identification using an image signature extracted by the image signature extraction method according to any of Claims 6 to 8.
 
11. A program for causing a computer to perform the method according to any of Claims 6 to 8.
 


Ansprüche

1. Bildsignatur-Extraktionsvorrichtung, die Folgendes umfasst:

eine Bereichsmerkmal-Berechnungseinheit (3), die jeweils von zwei Teilbereichen in einem Bild Bereichsmerkmale in Bezug auf jede von mehreren Dimensionen, die eine Bildsignatur bilden, berechnet, wobei die Bildsignatur Informationen zum Identifizieren des Bildes sind;

eine Differenzwert-Berechnungseinheit (43), die einen Differenzwert zwischen den berechneten Bereichsmerkmalen in Bezug auf jede der mehreren Dimensionen berechnet;

eine Bestimmungseinheit (45), die jeweils Absolutwerte der Differenzwerte in Bezug auf die mehreren Dimensionen berechnet, die Absolutwerte in Bezug auf jeden Typ von mehreren Typen, zu dem eine Gruppe von Dimensionen gehört, in aufsteigender Reihenfolge sortiert und einen Absolutwert an einer Position eines vorgegebenen Prozentsatzes von der Basis der sortierten Reihenfolge als einen Schwellenwert des Typs bestimmt; und

eine Quantisierungseinheit (44), die die Differenzwerte in Bezug auf jede Gruppe von Dimensionen unter Verwendung des Schwellenwertes des entsprechenden Typs in dreigliedrige Werte quantisiert;

wobei eine Gruppe von Dimensionen, die identische Teilbereichs-Paarmuster aufweisen, zu demselben Typ gehört, wobei das Teilbereichs-Paarmuster durch eine relative Position zwischen den zwei Teilbereichen und eine Kombination von Formen der zwei Teilbereiche bestimmt ist,

Teilbereichspaare, die zu demselben Typ gehören, an verschiedenen Absolutpositionen angeordnet sind,

die dreigliedrigen Werte einen ersten Wert, einen zweiten Wert, der um einen Wert kleiner als der erste Wert ist, und einen dritten Wert, der um einen Wert kleiner als der zweite Wert ist, enthalten,

eine Differenz zwischen dem ersten Wert und dem zweiten Wert und eine Differenz zwischen dem zweiten Wert und dem dritten Wert gleich sind, und

wobei

die zwei Teilbereiche ein erster Teilbereich und ein zweiter Teilbereich sind; und

die Typen Folgendes enthalten:

einen ersten Typ, bei dem zwei Rechtecke, die durch gleichmäßiges Teilen eines Quadrates einer bestimmten Größe im Bild in zwei Teile in einer vertikalen Richtung oder einer horizontalen Richtung gebildet sind, als der erste Teilbereich und der zweite Teilbereich dienen; und/oder

einen zweiten Typ, bei dem aus vier Quadraten, die durch gleichmäßiges Teilen eines Quadrates einer bestimmten Größe im Bild in zwei Teile in einer vertikalen Richtung und in zwei Teile in einer horizontalen Richtung gebildet sind, ein Bereich, der durch Kombinieren eines oberen, linken Quadrates und eines unteren, rechten Quadrates gebildet ist, als der erste Teilbereich dient und ein Bereich, der durch Kombinieren eines oberen, rechten Quadrates und eines unteren, linken Quadrates gebildet ist, als der zweite Teilbereich dient; und/oder

einen dritten Typ, bei dem der erste Teilbereich und der zweite Teilbereich Quadrate derselben Größe sind und in einer vertikalen Richtung und in einer horizontalen Richtung um einen Abstand, der gleich einem ganzzahligen Vielfachen einer Seite des Quadrates ist, beabstandet sind; und/oder

einen vierten Typ, bei dem der erste Teilbereich und der zweite Teilbereich Quadrate derselben Größe, jedoch kleiner als jene des dritten Typs sind und in einer vertikalen Richtung und in einer horizontalen Richtung um einen Abstand, der gleich einem ganzzahligen Vielfachen einer Seite des Quadrates ist, beabstandet sind; und/oder

einen fünften Typ, bei dem von zwei unterschiedlich bemessenen Quadraten mit gemeinsamem Schwerpunkt ein Bereich, der erhalten wird, indem das kleinere Quadrat aus dem größeren Quadrat entfernt wird, als der erste Teilbereich dient, und das kleinere Quadrat als der zweite Teilbereich dient; und/oder

einen sechsten Typ, bei dem zwei Rechtecke derselben Größe mit gemeinsamem Schwerpunkt und die derart senkrecht zueinander angeordnet sind, dass ihre Längsrichtungen horizontal bzw. vertikal sind, als der erste Teilbereich und der zweite Teilbereich dienen; und/oder

einen siebten Typ, bei dem aus drei vertikal oder horizontal aneinander angrenzenden Quadraten derselben Größe die Quadrate an beiden Enden als der erste Teilbereich dienen und ein Quadrat in der Mitte als der zweite Teilbereich dient.


 
2. Bildsignatur-Extraktionsvorrichtung nach Anspruch 1, wobei in Bezug auf jeden Typ
der Schwellenwert ein n-ter Absolutwert in der sortierten Reihenfolge ist, wobei n die Grundfläche (N x P/100) ist,
wobei N eine Anzahl von Dimensionen ist, die zu dem Typ gehören, und P der vorgegebene Prozentsatz ist.
 
3. Bildsignatur-Extraktionsvorrichtung nach Anspruch 1 oder 2, wobei
die zwei Teilbereiche aus einem oder mehreren Blöcken von 1024 Blöcken gebildet sind, die erhalten werden, indem das Bild in einer vertikalen Richtung in 32 Teile geteilt wird und das Bild in einer horizontalen Richtung in 32 Teile geteilt wird, und
das Quadrat einer bestimmten Größe des ersten Typs ein Quadrat ist, das vertikal aus 4 Blöcken und horizontal aus 4 Blöcken gebildet ist;
das Quadrat einer bestimmten Größe des zweiten Typs ein Quadrat ist, das vertikal aus 8 Blöcken und horizontal aus 8 Blöcken gebildet ist;
das Quadrat des dritten Typs ein Quadrat ist, das vertikal aus 10 Blöcken und horizontal aus 10 Blöcken gebildet ist;
das Quadrat des vierten Typs ein Quadrat ist, das vertikal aus 6 Blöcken und horizontal aus 6 Blöcken gebildet ist;
die zwei unterschiedlich bemessenen Quadrate des fünften Typs eine Kombination aus einem Quadrat, das vertikal aus 10 Blöcken und horizontal aus 10 Blöcken gebildet ist, und einem Quadrat, das vertikal aus 4 Blöcken und horizontal aus 4 Blöcken gebildet ist, oder eine Kombination aus einem Quadrat, das vertikal aus 6 Blöcken und horizontal aus 6 Blöcken gebildet ist, und einem Quadrat, das vertikal aus 2 Blöcken und horizontal aus 2 Blöcken gebildet ist, sind;
die zwei Rechtecke des sechsten Typs ein Rechteck, das vertikal aus 10 Blöcken und horizontal aus 6 Blöcken gebildet ist, und ein Rechteck, das vertikal aus 6 Blöcken und horizontal aus 10 Blöcken gebildet ist, sind; und
das Quadrat des siebten Typs ein Quadrat ist, das vertikal aus 4 Blöcken und horizontal aus 4 Blöcken gebildet ist.
 
4. Abgleichvorrichtung, die unter Verwendung einer Bildsignatur, die durch die Bildsignatur-Extraktionsvorrichtung nach einem der Ansprüche 1 bis 3 ausgegeben wird, einen Abgleich durchführt.
 
5. Identifizierungsvorrichtung, die unter Verwendung einer Bildsignatur, die durch die Bildsignatur-Extraktionsvorrichtung nach einem der Ansprüche 1 bis 3 ausgegeben wird, eine Identifizierung durchführt.
 
6. Bildsignatur-Extraktionsverfahren, das Folgendes umfasst:

Berechnen von Bereichsmerkmalen in Bezug auf jede von mehreren Dimensionen, die eine Bildsignatur bilden, jeweils von zwei Teilbereichen in einem Bild, wobei die Bildsignatur Informationen zum Identifizieren des Bildes sind;

Berechnen eines Differenzwertes zwischen den berechneten Bereichsmerkmalen in Bezug auf jede der mehreren Dimensionen;

jeweils Berechnen von Absolutwerten der Differenzwerte in Bezug auf die mehreren Dimensionen, Sortieren der Absolutwerte in aufsteigender Reihenfolge in Bezug auf jeden Typ von mehreren Typen, zu dem eine Gruppe von Dimensionen gehört, und Bestimmen eines Absolutwertes an einer Position einen vorgegebenen Prozentsatz von der Basis der sortierten Reihenfolge als einen Schwellenwert des Typs; und

Quantisieren der Differenzwerte in Bezug auf jede Gruppe von Dimensionen in dreigliedrige Werte unter Verwendung des Schwellenwertes des entsprechenden Typs;

wobei eine Gruppe von Dimensionen, die identische Teilbereichs-Paarmuster aufweisen, zu demselben Typ gehört, wobei das Teilbereichs-Paarmuster durch eine relative Position zwischen den zwei Teilbereichen und eine Kombination von Formen der zwei Teilbereiche bestimmt ist,

Teilbereichspaare, die zu demselben Typ gehören, an verschiedenen Absolutpositionen angeordnet sind,

die dreigliedrigen Werte einen ersten Wert, einen zweiten Wert, der um einen Wert kleiner als der erste Wert ist, und einen dritten Wert, der um einen Wert kleiner als der zweite Wert ist, enthalten,

eine Differenz zwischen dem ersten Wert und dem zweiten Wert und eine Differenz zwischen dem zweiten Wert und dem dritten Wert gleich sind, und

wobei

die zwei Teilbereiche ein erster Teilbereich und ein zweiter Teilbereich sind; und

die Typen Folgendes enthalten:

einen ersten Typ, bei dem zwei Rechtecke, die durch gleichmäßiges Teilen eines Quadrates einer bestimmten Größe im Bild in zwei Teile in einer vertikalen Richtung oder einer horizontalen Richtung gebildet sind, als der erste Teilbereich und der zweite Teilbereich dienen; und/oder

einen zweiten Typ, bei dem von vier Quadraten, die durch gleichmäßiges Teilen eines Quadrates einer bestimmten Größe im Bild in zwei Teile in einer vertikalen Richtung und in zwei Teile in einer horizontalen Richtung gebildet sind, ein Bereich, der durch Kombinieren eines oberen, linken Quadrates und eines unteren, rechten Quadrates gebildet ist, als der erste Teilbereich dient und ein Bereich, der durch Kombinieren eines oberen, rechten Quadrates und eines unteren, linken Quadrates gebildet ist, als der zweite Teilbereich dient; und/oder

einen dritten Typ, bei dem der erste Teilbereich und der zweite Teilbereich Quadrate derselben Größe sind und in einer vertikalen Richtung und in einer horizontalen Richtung um einen Abstand, der gleich einem ganzzahligen Vielfachen einer Seite des Quadrates ist, beabstandet sind; und/oder

einen vierten Typ, bei dem der erste Teilbereich und der zweite Teilbereich Quadrate derselben Größe, jedoch kleiner als jene des dritten Typs sind und in einer vertikalen Richtung und in einer horizontalen Richtung um einen Abstand, der gleich einem ganzzahligen Vielfachen einer Seite des Quadrates ist, beabstandet sind; und/oder

einen fünften Typ, bei dem von zwei unterschiedlich bemessenen Quadraten mit gemeinsamem Schwerpunkt ein Bereich, der erhalten wird, indem das kleinere Quadrat aus dem größeren Quadrat entfernt wird, als der erste Teilbereich dient, und das kleinere Quadrat als der zweite Teilbereich dient; und/oder

einen sechsten Typ, bei dem zwei Rechtecke derselben Größe mit gemeinsamem Schwerpunkt und die derart senkrecht zueinander angeordnet sind, dass ihre Längsrichtungen horizontal bzw. vertikal sind, als der erste Teilbereich und der zweite Teilbereich dienen; und/oder

einen siebten Typ, bei dem aus drei vertikal oder horizontal aneinander angrenzenden Quadraten derselben Größe die Quadrate an beiden Enden als der erste Teilbereich dienen und ein Quadrat in der Mitte als der zweite Teilbereich dient.


 
7. Bildsignatur-Extraktionsverfahren nach Anspruch 6, wobei in Bezug auf jeden Typ
der Schwellenwert ein n-ter Absolutwert in der sortierten Reihenfolge ist, wobei n die Grundfläche (N x P/100) ist,
wobei N eine Anzahl von Dimensionen ist, die zu dem Typ gehören, und P der vorgegebene Prozentsatz ist.
 
8. Bildsignatur-Extraktionsverfahren nach Anspruch 6 oder 7, wobei
die zwei Teilbereiche aus einem oder mehreren Blöcken von 1024 Blöcken gebildet sind, die erhalten werden, indem das Bild in einer vertikalen Richtung in 32 Teile geteilt wird und das Bild in einer horizontalen Richtung in 32 Teile geteilt wird, und
das Quadrat einer bestimmten Größe des ersten Typs ein Quadrat ist, das vertikal aus 4 Blöcken und horizontal aus 4 Blöcken gebildet ist;
das Quadrat einer bestimmten Größe des zweiten Typs ein Quadrat ist, das vertikal aus 8 Blöcken und horizontal aus 8 Blöcken gebildet ist;
das Quadrat des dritten Typs ein Quadrat ist, das vertikal aus 10 Blöcken und horizontal aus 10 Blöcken gebildet ist;
das Quadrat des vierten Typs ein Quadrat ist, das vertikal aus 6 Blöcken und horizontal aus 6 Blöcken gebildet ist;
die zwei unterschiedlich bemessenen Quadrate des fünften Typs eine Kombination aus einem Quadrat, das vertikal aus 10 Blöcken und horizontal aus 10 Blöcken gebildet ist, und einem Quadrat, das vertikal aus 4 Blöcken und horizontal aus 4 Blöcken gebildet ist, oder eine Kombination aus einem Quadrat, das vertikal aus 6 Blöcken und horizontal aus 6 Blöcken gebildet ist, und einem Quadrat, das vertikal aus 2 Blöcken und horizontal aus 2 Blöcken gebildet ist, sind;
die zwei Rechtecke des sechsten Typs ein Rechteck, das vertikal aus 10 Blöcken und horizontal aus 6 Blöcken gebildet ist, und ein Rechteck, das vertikal aus 6 Blöcken und horizontal aus 10 Blöcken gebildet ist, sind; und
das Quadrat des siebten Typs ein Quadrat ist, das vertikal aus 4 Blöcken und horizontal aus 4 Blöcken gebildet ist.
 
9. Abgleichverfahren zum Durchführen eines Abgleichs unter Verwendung einer Bildsignatur, die durch das Bildsignatur-Extraktionsverfahren nach einem der Ansprüche 6 bis 8 extrahiert wird.
 
10. Identifizierungsverfahren zum Durchführen einer Identifizierung unter Verwendung einer Bildsignatur, die durch das Bildsignatur-Extraktionsverfahren nach einem der Ansprüche 6 bis 8 extrahiert wird.
 
11. Programm zum Bewirken, dass ein Computer das Verfahren nach einem der Ansprüche 6 bis 8 durchführt.
 


Revendications

1. Dispositif d'extraction de signature d'image, comprenant :

une unité de calcul de caractéristiques de région (3) qui calcule, par rapport à chacune d'une pluralité de dimensions composant une signature d'image, des caractéristiques de région, respectivement, à partir de deux sous-régions dans une image, la signature d'image constituant une information permettant d'identifier l'image ;

une unité de calcul de valeur de différence (43) qui calcule une valeur de différence entre les caractéristiques de région calculées par rapport à chacune de la pluralité de dimensions ;

une unité de détermination (45) qui calcule les valeurs absolues des valeurs de différence par rapport à la pluralité de dimensions, respectivement, trie, par rapport à chaque type, parmi une pluralité de types, auquel appartient un groupe de dimensions, les valeurs absolues par ordre croissant et détermine, comme seuil dudit type, une valeur absolue à une position d'un pourcentage prédéterminé à partir du bas de l'ordre trié ; et

une unité de quantification (44) qui quantifie les valeurs de différence par rapport à chaque groupe de dimensions en valeurs ternaires en utilisant le seuil du type correspondant ;

dans lequel un groupe de dimensions ayant des motifs de paires de sous-régions identiques appartient à un même type, le motif de paires de sous-régions étant déterminé par une position relative entre les deux sous-régions et une combinaison de formes des deux sous-régions,

des paires de sous-régions appartenant au même type sont situées à différentes positions absolues,

les valeurs ternaires incluent une première valeur, une deuxième valeur qui est une valeur inférieure à la première valeur, et une troisième valeur qui est une valeur inférieure à la deuxième valeur,

une différence entre la première valeur et la deuxième valeur et une différence entre la deuxième valeur et la troisième valeur sont égales, et

dans lequel

les deux sous-régions sont une première sous-région et une deuxième sous-région ; et

les types comportent au moins l'un parmi :

un premier type dans lequel deux rectangles, formés en divisant de manière égale un carré d'une certaine taille dans l'image en deux dans une direction verticale ou dans une direction horizontale, constituent la première sous-région et la deuxième sous-région ;

un deuxième type dans lequel, parmi quatre carrés formés en divisant de manière égale un carré d'une certaine taille dans l'image en deux dans une direction verticale et en deux dans un direction horizontale, une région formée en combinant un carré supérieur gauche et un carré inférieur droit constitue la première sous-région et une région formée en combinant un carré supérieur droit et un carré inférieur gauche constitue la deuxième sous-région ;

un troisième type dans lequel la première sous-région et la deuxième sous-région sont des carrés de la même taille et sont espacées dans une direction verticale et dans une direction horizontale d'une distance égale à un multiple entier d'un côté du carré ;

un quatrième type dans lequel la première sous-région et la deuxième sous-région sont des carrés de la même taille mais sont plus petites que celles du troisième type, et sont espacées dans une direction verticale et dans une direction horizontale d'une distance égale à un multiple entier d'un côté du carré ;

un cinquième type dans lequel, de deux carrés de tailles différentes partageant le centre de gravité, une région obtenue en retirant le petit carré du grand carré constitue la première sous-région, et le petit carré constitue la deuxième sous-région ;

un sixième type dans lequel deux rectangles de la même taille partageant le centre de gravité et étant placés orthogonalement l'un par rapport à l'autre de sorte que leurs directions longitudinales soient horizontales et verticales, respectivement, constituent la première sous-région et la deuxième sous-région ; et

un septième type dans lequel, parmi trois carrés verticalement ou horizontalement adjacents de la même taille, des carrés aux deux extrémités constituent la première sous-région et un carré au centre constitue la deuxième sous-région.


 
2. Dispositif d'extraction de signature d'image selon la revendication 1, dans lequel, par rapport à chaque type,
le seuil est une nième valeur absolue dans l'ordre trié, n étant le plancher (N x P/100),
où N est un nombre de dimensions appartenant audit type et P est le pourcentage prédéterminé.
 
3. Dispositif d'extraction de signature d'image selon la revendication 1 ou 2, dans lequel
les deux sous-régions sont formées d'un ou de plusieurs bloc(s) parmi 1024 blocs obtenus en divisant l'image en 32 dans une direction verticale et en divisant l'image en 32 dans une direction horizontale, et
le carré d'une certaine taille du premier type est un carré constitué de 4 blocs verticalement et 4 blocs horizontalement ;
le carré d'une certaine taille du deuxième type est un carré constitué de 8 blocs verticalement et 8 blocs horizontalement ;
le carré du troisième type est un carré constitué de 10 blocs verticalement et 10 blocs horizontalement ;
le carré du quatrième type est un carré constitué de 6 blocs verticalement et 6 blocs horizontalement ;
les deux carrés de tailles différentes du cinquième type sont une combinaison d'un carré constitué de 10 blocs verticalement et 10 blocs horizontalement et d'un carré constitué de 4 blocs verticalement et 4 blocs horizontalement, ou une combinaison d'un carré constitué de 6 blocs verticalement et 6 blocs horizontalement et d'un carré constitué de 2 blocs verticalement et 2 blocs horizontalement ;
les deux rectangles du sixième type sont un rectangle constitué de 10 blocs verticalement et 6 blocs horizontalement et un rectangle constitué de 6 blocs verticalement et 10 blocs horizontalement ; et
le carré du septième type est un carré constitué de 4 blocs verticalement et 4 blocs horizontalement.
 
4. Dispositif de correspondance qui effectue une correspondance en utilisant une signature d'image délivrée en sortie par le dispositif d'extraction de signature d'image selon l'une des revendications 1 à 3.
 
5. Dispositif d'identification qui effectue une identification en utilisant une signature d'image délivrée en sortie par le dispositif d'extraction de signature d'image selon l'une des revendications 1 à 3.
 
6. Procédé d'extraction de signature d'image, comprenant les étapes consistant à :

calculer, par rapport à chacune d'une pluralité de dimensions composant une signature d'image, des caractéristiques de région, respectivement, à partir de deux sous-régions dans une image, la signature d'image constituant une information permettant d'identifier l'image ;

calculer une valeur de différence entre les caractéristiques de région calculées par rapport à chacune de la pluralité de dimensions ;

calculer les valeurs absolues des valeurs de différence par rapport à la pluralité de dimensions, respectivement, trier, par rapport à chaque type, parmi une pluralité de types, auquel appartient un groupe de dimensions, les valeurs absolues par ordre croissant et déterminer, comme seuil dudit type, une valeur absolue à une position d'un pourcentage prédéterminé à partir du bas de l'ordre trié ; et

quantifier les valeurs de différence par rapport à chaque groupe de dimensions en valeurs ternaires en utilisant le seuil du type correspondant ;

dans lequel un groupe de dimensions ayant des motifs de paires de sous-régions identiques appartient à un même type, le motif de paires de sous-régions étant déterminé par une position relative entre les deux sous-régions et une combinaison de formes des deux sous-régions,

les paires de sous-régions appartenant au même type sont situées à des positions absolues différentes,

les valeurs ternaires incluent une première valeur, une deuxième valeur qui est une valeur inférieure à la première valeur et une troisième valeur qui est une valeur inférieure à la deuxième valeur,

une différence entre la première valeur et la deuxième valeur et une différence entre la deuxième valeur et la troisième valeur sont égales, et

dans lequel

les deux sous-régions sont une première sous-région et une deuxième sous-région; et

les types comportent au moins l'un parmi :

un premier type dans lequel deux rectangles, formés en divisant de manière égale un carré d'une certaine taille dans l'image en deux dans une direction verticale ou dans une direction horizontale, constituent la première sous-région et la deuxième sous-région ;

un deuxième type dans lequel, parmi quatre carrés formés en divisant de manière égale un carré d'une certaine taille dans l'image en deux dans une direction verticale et en deux dans une direction horizontale, une région formée en combinant un carré supérieur gauche et un carré inférieur droit constitue la première sous-région et une région formée en combinant un carré supérieur droit et un carré inférieur gauche constitue la deuxième sous-région ;

un troisième type dans lequel la première sous-région et la deuxième sous-région sont des carrés de la même taille et sont espacées dans une direction verticale et dans une direction horizontale d'une distance égale à un multiple entier d'un côté du carré ;

un quatrième type dans lequel la première sous-région et la deuxième sous-région sont des carrés de la même taille mais sont plus petites que celles du troisième type, et sont espacées dans une direction verticale et dans une direction horizontale d'une distance égale à un multiple entier d'un côté du carré ;

un cinquième type dans lequel, de deux rectangles de tailles différentes partageant le centre de gravité, une région obtenue en retirant le petit carré du grand carré constitue la première sous-région, et le petit carré constitue la deuxième sous-région ;

un sixième type dans lequel deux rectangles de la même taille partageant le centre de gravité et étant placés orthogonalement l'un par rapport à l'autre de sorte que leurs directions longitudinales soient horizontales et verticales, respectivement, constituent la première sous-région et la deuxième sous-région ; et

un septième type dans lequel parmi trois carrés verticalement ou horizontalement adjacents de la même taille, des carrés aux deux extrémités constituent la première sous-région et un carré au centre constitue la deuxième sous-région.


 
7. Procédé d'extraction de signature d'image selon la revendication 6, dans lequel, par rapport à chaque type,
le seuil est une nième valeur absolue dans l'ordre trié, n étant le plancher (N x P/100),
où N est un nombre de dimensions appartenant audit type et P est le pourcentage prédéterminé.
 
8. Procédé d'extraction de signature d'image selon la revendication 6 ou 7, dans lequel
les deux sous-régions sont formées d'un ou de plusieurs bloc(s) parmi 1024 blocs obtenus en divisant l'image en 32 dans une direction verticale et en divisant l'image en 32 dans une direction horizontale, et
le carré d'une certaine taille du premier type est un carré constitué de 4 blocs verticalement et 4 blocs horizontalement ;
le carré d'une certaine taille du deuxième type est un carré constitué de 8 blocs verticalement et 8 blocs horizontalement ;
le carré du troisième type est un carré constitué de 10 blocs verticalement et 10 blocs horizontalement ;
le carré du quatrième type est un carré constitué de 6 blocs verticalement et 6 blocs horizontalement ;
les deux carrés de tailles différentes du cinquième type sont une combinaison d'un carré constitué de 10 blocs verticalement et 10 blocs horizontalement et d'un carré constitué de 4 blocs verticalement et 4 blocs horizontalement, ou une combinaison d'un carré constitué de 6 blocs verticalement et 6 blocs horizontalement et d'un carré constitué de 2 blocs verticalement et 2 blocs horizontalement ;
les deux rectangles du sixième type sont un rectangle constitué de 10 blocs verticalement et 6 blocs horizontalement et un rectangle constitué de 6 blocs verticalement et 10 blocs horizontalement ; et
le carré du septième type est un carré constitué de 4 blocs verticalement et 4 blocs horizontalement.
 
9. Procédé de correspondance permettant d'effectuer une correspondance en utilisant une signature d'image extraite par le procédé d'extraction de signature d'image selon l'une des revendications 6 à 8.
 
10. Procédé d'identification permettant d'effectuer une identification en utilisant une signature d'image extraite par le procédé d'extraction de signature d'image selon l'une des revendications 6 à 8.
 
11. Programme amenant un ordinateur à exécuter le procédé selon l'une des revendications 6 à 8.
 




Drawing

























































































































































































Cited references

REFERENCES CITED IN THE DESCRIPTION



This list of references cited by the applicant is for the reader's convenience only. It does not form part of the European patent document. Even though great care has been taken in compiling the references, errors or omissions cannot be excluded and the EPO disclaims all liability in this regard.

Patent documents cited in the description




Non-patent literature cited in the description