Method and device for sorting security documents

Abstract

 The invention relates to a method for sorting security documents, preferably banknotes, wherein from the reflection of light in a wavelength range of approximately 400-500 nm of at least a part of a security document a degree of soiling of the security document is determined. In one embodiment the B component from the RGB components of a colour image is used for determining the reflection of light in the wavelength range. In an alternative embodiment the wavelength range has been selected from approximately 400-450 nm, preferably approximately 410-430 nm.

Description

BACKGROUND OF THE INVENTION

[0001] The invention relates to a device and method for sorting security documents, particularly banknotes. The invention more particularly relates to sorting collected banknotes that are in circulation, particularly classifying banknotes into those that can be brought into circulation again, and those that can no longer be brought into circulation.

[0002] Considering the importance of security documents it is of great importance that the security documents in circulation are of a good quality. A bad quality may render it more difficult to the public to verify the genuineness. This applies in particular to banknotes. Moreover it is of great importance to the reputation and image of the issuing authorities, particularly the image of reliability of the national central banks and the European Central Bank, that the banknotes in circulation are of a good quality.

[0003] On the other hand there are large numbers of banknotes in circulation, and the sorting thereof nonetheless has to be as consistent as possible.

[0004] An additional challenge is that for instance Euro banknotes have a large area of distribution, and sorting preferably has take place nationally, using the same criteria and quality and according to the same standards.

[0005] A device for sorting banknotes is generally described in applicant's European patent EP-B1-1043700.

SUMMARY OF THE INVENTION

[0006] It is an object of the invention to provide a method and device for sorting banknotes as efficiently as possible, as regards speed, costs as well as quality criteria.

[0007] It is a particular object of the invention to effect that the banknotes are sorted into those that can be brought into circulation again and those that can no longer be brought into circulation again.

[0008] For that purpose the invention relates to a method for sorting security documents, preferably banknotes, wherein from the reflection of light in the wavelength range of approximately 400-500 nm of at least a part of a security document a degree of soiling of the security document is determined.

[0009] According to another aspect the invention further relates to method for sorting security documents, preferably banknotes, wherein a security document is exposed to infrared light, by means of a first image acquisition device at least one image of at least a part of the security document is taken in the infrared light, and a degree of creases of the security document is determined from the at least one infrared image.

[0010] According to another aspect the invention further relates to a method for sorting security documents, preferably banknotes, wherein at least one image is taken of at least a part of a security document, the image or images is or are subdivided into predetermined, fixed sub areas, of each of the sub areas an average reflection is determined, said average reflection values are each compared with predetermined calibration values, and from the comparison a degree of smudginess is determined.

[0011] According to another aspect the invention further relates to a method for sorting security documents, preferably banknotes, wherein from at least one image of at least a part of a security document a degree of soiling, and a crease level and a smudginess is determined, and from among others those determinations the security document is classified, preferably into a class of security documents that can be brought into circulation again and a class of security documents that can no longer be brought into circulation.

[0012] The said methods make it possible to sort particularly banknotes reliably and rapidly, particularly to sort banknotes into those that can be brought into circulation again, and those that have to taken out of circulation. Especially because the number of notes to check is very large, a rapid and reliable check is imperative.

[0013] Particularly determining the degree of soiling is of importance for sorting banknotes that can either be brought back into circulation or have to be taken out of circulation. An additional determination of the crease level and/or smudginess may lead to the banknotes being sorted more reliably. The degree of soiling particularly relates to both the seriousness of the soiling and the nature thereof.

[0014] In one embodiment the blue (B) component from the RGB components of a colour image is used for determining the reflection of light in the wavelength range.

[0015] In one embodiment the wavelength range has been selected from approximately 400-450 nm. One of the properties of dirt on banknotes is that said dirt absorbs more visible light as the wavelength becomes lower. In short in case of a soiled note more light is absorbed (and therefore reflected less) at 400 nm than at 420 nm, at 450 nm more light is absorbed than at 470 nm and this tendency can be seen over the entire visual spectrum, albeit that the differences at the lowest wavelengths may be the largest and as the wavelengths approach the red and infrared they seem to become smaller. Therefore a detector appears to operate better when the wavelength range is selected as low as possible, because the difference between clean and soiled notes is largest then.

[0016] For banknotes, however, it is a fact that there is quite a bit of variation between various banknote productions in the wavelength range between 390 and 410 nm. Said variation seems to be predominantly caused by a variety of one of the auxiliary substances in the current paper manufacturing, and particularly by using the rutile or anatase form of titanium dioxide. For the rutile form of titanium dioxide applies that the reflection below 410 nm drops dramatically, whereas this phenomenon in the anatase form does not occur until below 390nm.

[0017] In one embodiment the wavelength range in view of recognising dirt in the anatase form has been selected from 390-410 nm (after all as low as possible), whereas in the rutile form the wavelength range has been selected from a less optimal range between 410 and 430 nm.

[0018] In one embodiment the wavelength range has been selected from 410-430 nm. An advantage of applying said violet part at the blue side of the spectrum is that determining the degree of soiling can be less dependent on or even independent from the type of auxiliary substance in the security document.

[0019] In one embodiment the degree of soiling of the used security document is determined from reflection differences between a used security document and a new security document of light in the wavelength range of approximately 400-500 nm.

[0020] In one embodiment the degree of soiling of the security document is determined from reflection of light in a first wavelength range and in a second wavelength range for a same location on the security document, wherein the first wavelength range is the wavelength range of approximately 400-500 nm.

[0021] In one embodiment the degree of soiling of the security document is determined from reflection ratios between reflection of light in the first wavelength range and the second wavelength range for a same location on the security document, preferably from a quotient of the reflection.

[0022] On the one hand the second wavelength range may be a red wavelength range, and wherein the quotient preferably is a quotient of an R and B component from the RGB components of a colour image.

[0023] On the other hand the second wavelength range may also be a wavelength range at the blue side of the spectrum that differs from the first wavelength range. In one embodiment the first wavelength range has been selected from approximately 410-430 nm, preferably approximately 420-430 nm, and the second wavelength range has been selected from approximately 430-500 nm, preferably approximately 450-460 nm.

[0024] In one embodiment of the former method according to the invention, the determination of the reflection comprises an image, preferably a digital image of at least a part of the security document.

[0025] In one embodiment of the former method according to the invention, the security document is subdivided into several areas, preferably rectangles, and of each area an average reflection is then determined, wherein preferably the average reflection is compared and a predetermined average reflection.

[0026] In one embodiment of the former method according to the invention, the areas are predetermined areas having a predetermined position on the carrier.

[0027] In one embodiment of said former method the areas have individually been predetermined for each type of carrier.

[0028] In one embodiment of the former method according to the invention, a visual degree of soiling of a test set is first of all visually established, and the reflection of the same test set is determined, and the visual degree of soiling is correlated by means of statistical calculation to the reflection measurement or at least one parameter derived therefrom for determining a calibration, and said calibration is applied to the reflection measurement or the at least one parameter derived therefrom for determining a degree of soiling.

[0029] In one embodiment of the former method according to the invention, the reflection is measured in one or more areas on the security document that have been lightly printed or have not been printed on, for the determination of the soiling.

[0030] In one embodiment of the former method according to the invention, the area or the areas has or have a representative surface for the area in question of the security document.

[0031] In one embodiment of the former method according to the invention, the sorting criteria are adjusted using a test set of security documents that have been visually assessed, wherein soiling degrees are allocated to the degree of soiling determined by means of the reflection measurements.

[0032] In one embodiment of said method a relation is calculated between the allocated soiling degrees and measured degree of soiling, by means of statistical calculation for the test set, and the calculated relation is used to allocate a soiling degree to the security documents to be sorted, wherein preferably five soiling degrees are allocated, which preferably have a numerical value.

[0033] In one embodiment of the method for sorting security documents, preferably banknotes, wherein a security document is exposed to infrared light, by means of a first image acquisition device at least one image of at least a part of the security document is taken in the infrared light, and a degree of creases of the security document is determined from the at least one infrared image, in one embodiment the security document is exposed to infrared light that has an angle of incidence to the surface of the security document, and a reflection image is taken preferably approximately transverse to the plane of the security document.

[0034] In one embodiment thereof the infrared light has an angle of incidence of 20-40 degrees, preferably approximately 30 degrees.

[0035] In one further embodiment of the latter method an image, preferably a digital image, is taken by means of an image acquisition device that is sensitive in the infrared range.

[0036] In one further embodiment of the latter method, the image acquisition device is positioned over the security document, preferably approximately straight over it.

[0037] In one further embodiment of the latter method, a standard deviation of the brightness is determined from the image, from which deviation a crease level is determined.

[0038] In one further embodiment of the latter method, the standard deviation of the brightness is determined for areas on the security document, preferably rectangles, more preferably a grid of rectangles, such as 5* 10 rectangles divided over the security document.

[0039] In one further embodiment of the latter method the security document is exposed by means of a second light source, positioned at an angle to a surface of the security document and to the first light source.

[0040] In one further embodiment of the latter method, a crease level is at least placed in a class of relatively new security documents with sharp folding lines and a class of worn security documents having creases.

[0041] In one further embodiment of the latter method, the image is taken in transmission, wherein the security document is exposed at the one surface and at least one image is taken from the opposite surface.

[0042] In one further embodiment of the latter method, at least one image is taken in transmission and at least one image is taken wherein infrared light is at an angle of incidence to the surface of the security document, and a reflection image is taken of said surface, preferably approximately straight above said surface.

[0043] In one embodiment of the method for sorting security documents, preferably banknotes, wherein at least one image is taken of at least a part of a security document, the image or images is or are subdivided into predetermined, fixed sub areas, of each of the sub areas an average reflection is determined, said average reflection values are each compared with predetermined calibration values, and from the comparison a degree of smudginess is determined, in one embodiment the established calibration values are determined from average values for a test set of several notes.

[0044] In one embodiment of said method the image is taken with an exposure having a wavelength component at the blue side of the spectrum, or the blue (B) component of the image is used.

[0045] In one embodiment of said method the security document is given a class indication based on the degree of smudginess.

[0046] In the method for sorting security documents, preferably banknotes, wherein from at least one image of at least a part of a security document a degree of soiling, a crease level and a smudginess is determined, and from among others those determinations the security document is placed in a class, preferably a class of security documents that can be brought back into circulation again and a class of security documents that can no longer be brought into circulation again, in one embodiment one or more images of at least a part of a security document is or are taken by means of a reflection measurement and from said image or images at least the degree of soiling, the crease level and the smudginess is or are determined.

[0047] In said method in one embodiment the degree of soiling is determined using the method described above.

[0048] In said method in one embodiment the crease level is determined using the method described above.

[0049] In said method in one embodiment the smudginess is determined using the method described above.

[0050] According to another aspect of the invention it relates to a method for sorting documents, particularly security documents, wherein a document is exposed to light having a wavelength component in the range of approximately 400-500 nm, by means of an image acquisition device an image of the security document is taken in the wavelength range of approximately 400-500 nm, the image taken is compared with an image taken beforehand in a similar, preferably the same, security document, after which on the basis of the comparison a degree of soiling of the document is determined.

[0051] According to another aspect of the invention it relates to an assembly provided with an exposure unit, an image acquisition device and a data processing unit provided with software for when said software is active in the data processing unit, carrying out a method as described above.

[0052] The invention further relates to a carrier provided with machine instructions, such as computer software, for when active in the data processing unit, carrying out a method as described above.

[0053] The invention further relates to a device, adapted for carrying out the method as described in the description.

[0054] The various methods described above make it possible to sort security documents rapidly and accurately as regards the various aspects that are for instance important to security documents such as banknotes. In sorting banknotes in particular, speed and accuracy, particularly a low rate of incorrect disapproval and a low rate of incorrect approval are of great importance. In the Netherlands alone millions of banknotes have to be sorted in a short time.

[0055] The aspects and measures described and/or shown in the application may where possible also be used individually. Said individual aspects, such as determining the degree, nature and spreading of the crease level, determining the degree, nature and spreading of the smudginess and other aspects may be combined, but may also each individually be the subject of divisional patent applications relating thereto.

SHORT DESCRIPTION OF THE DRAWINGS

[0056] The invention will be elucidated on the basis of a number of exemplary embodiments shown in the attached drawings, in which:

Figure 1 shows a graph of the correlation between visual quality and local reflection;

Figure 2 shows an image in infrared using exposure at an angle; and

Figure 3 shows an image in infrared in transmission.

DETAILED DESCRIPTION OF THE DRAWINGS

[0057] Research into the soiling on banknotes at the Applicant's proved the soiling on banknotes to have a strong red component.

[0058] This was among others established by measuring the reflection of a white light source on a number of banknotes of different qualities using a colour camera.

[0059] Research proved that a difference measurement at wavelength ranges of about 420-430 and about 450-460 may yield a better colour result than a difference measurement between the red and blue extremities of the visual spectrum. A consequence thereof is that the difference measurement can no longer be carried out using the relatively simple colour camera.

[0060] In an exemplary embodiment a monochrome camera is used, wherein two illumination sources are being used; a first illumination source emitting light in the wavelength range of approximately 420-430 nm and a second illumination source emitting light in the wavelength range of approximately 450-460 nm. With this arrangement images of the document may be obtained by alternately switching the first and second illumination source on and off, wherein switching takes place per image line of the camera. In this example a selection of the wavelength range with the illumination source is opted for. Said selection may additionally or alternatively also take place by using colour filters that are alternately placed in front of the camera.

[0061] In addition the reflection on this series of banknotes has also been measured using a monochrome camera while exposing with an infrared source.

[0062] In this way it is possible to measure the reflection of the series of banknotes in the infrared, red, green and blue parts of the spectrum.

[0063] In practice it turned out that the difference in reflection between a new and a soiled note in the red and infrared part of the spectrum is relatively low, whereas the difference in reflection between a new note and a soiled note in the part of the spectrum of approximately 400-500 nm turned out the be the largest (see Table 1 below). Generally the soiling turned out to become better visible when the wavelength of the light (in the visible wavelength range) decreased. The classification was made on the basis of visual assessment of approximately 400-700 mm.

Table 1 Connection reflection and soiling class

Class	Infrared	Red	Green	Blue
Very clean	107.0	155.9	164.1	167.6
Clean	105.0	156.2	164.8	167.7
Acceptable	105.3	154.7	161.7	162.0
Soiled	104.7	152.3	156.1	153.9
Highly soiled	101.4	145.9	147.4	142.2

1. The use of blue exposure or a filter allowing blue through or the blue component of an RGB colour image.

[0064] In order to recognise soiling on a banknote as well as possible, in a method according to the invention preferably a blue light source emitting light having a wavelength or a wavelength range of 400-500 nm, or a white light source with a blue filter, absorbing light having a wavelength or wavelength range of 400-500 nm, in front of a monochrome camera is used or the B component of an RGB colour camera is used.

2. Use of a ratio measurement.

[0065] A further improvement can be achieved by instead of using a direct reflection measurement under blue exposure or with a filter allowing blue through as qualification for the test decision, carrying out a ratio measurement between the red and blue component of an RGB colour camera. If for instance the quotient of B and R is selected as soiling determining relation, this quotient in case of a dirty note will be generally low with respect to said quotient of a clean note. After all the B component for a soiled note will decrease relatively stronger than the R component. The advantage of a relative measurement is that the initial variation in the reflected light quantity, which occurs in new notes, is largely eliminated. In practice it turned out that the (for that matter present) variation in tone is less large than the variation in reflection.

[0066] A further improvement can be achieved by a difference measurement at wavelength ranges about 420-430 and about 450-460 nm.

3. Optimisation measurement position on note.

[0067] A further improvement can be achieved by optimising the location on which the light measurements are carried out. In order to determine the optimal location for a measurement a note is divided into a grid of for instance 5*10 rectangles. Subsequently it is determined of each of these 50 squares to what extend the reflection measurements correlate to the general soiling condition of the note. This general soiling condition is visually determined (beforehand).

[0068] In table 2 (see below) the correlation of the reflection per rectangle is shown for a 5 Euro note, wherein the rectangle having coordinates 1,1 is top left in the note and the element 10,5 is the bottom right corner. In figure 1 said correlation is also graphically shown.

Table 2. Correlation local reflection using general visual assessment of the note.

	1	2	3	4	5	6	7	8	9	10
1	0.710	0.327	0.755	0.691	0.723	0.568	0.538	0.443	0.288	0.242
2	0.765	0.808	0.737	0.769	0.761	0.557	0.245	0.434	0.450	0.343
3	0.778	0.839	0.777	0.813	0.777	0.429	0.371	0.737	0.549	0.381
4	0.756	0.806	0.802	0.809	0.774	0.431	0.374	0.616	0.441	0.155
5	0.710	0.667	0.615	0.715	0.740	0.446	0.250	0.386	0.465	0.364

[0069] Practice shows that the squares that are in an area of the note that is lightly printed or not printed on, provide a more reliable assessment about the note quality than the more heavily printed areas of a note. It furthermore shows that the squares in the corners and the squares adjacent to a heavily printed area of a note give a relatively less reliable indication of the soiling than the squares that are entirely within the areas of the note that are lightly printed or not printed on. The number of rectangles that is selected (in this case 50) may vary, but some demands are made on the number. For instance the surface of the rectangles must be a reasonable surface in order to be representative for the note area in question. On the other hand said surface should not be so large that the various note details can no longer be distinguished from each other.

4. Use of five soiling classes.

[0070] It furthermore turned out that the use of a test set of five soiling degrees is quite usable for determining test thresholds and test algorithms. In the various studies the following classifications was used:

Class	Name	Description
1	very clean	as good as new
2	clean	a waste to destroy
3	acceptable	suitable to be brought into circulation again, but if this note is disapproved of this will not be a great loss
4	soiled	a note to be disapproved of in principle, but if the machine approves of it as yet this will not be a qualitative disaster
5	highly soiled must not be brought into circulation again

[0071] Practice shows that proper results can already be achieved with a class size of 100 notes. By allocating a numerical value to the five classes and using them in developing and determining test algorithms and test thresholds, they can be determined relatively easy using mathematical and statistical techniques.

5. Crease level determination.

[0072] It furthermore appears that in addition to the soiling degree of a note other properties play a part in assessing the suitability to be brought into circulation. In addition to a number of mechanical properties that are relatively easy to establish (for instance holes, dog-ears, tears), two properties that are harder to establish play a part, namely the crease level and smudges.

[0073] Although said properties play a less prominent part in ascertaining whether a note is still suitable for re-circulation than the soiling degree does, in practice the test result turns out to improve nonetheless when said properties are included in the general test decision. For the crease level measurement two methods can be used. In one method an infrared source is radiated at an angle to the note surface. In the other method an infrared source is radiated through the note (transmission).

5.1 Crease level determination through shadow effects

[0074] In one embodiment this method consists of a monochrome camera having a filter allowing infrared through, which filter is directed perpendicular to the note surface. Furthermore an infrared light source is directed at an angle of approximately 30° ± 10° to the note surface on the note area under the camera.

[0075] As a large part of the Euro notes is transparent in the infrared part of the spectrum, creases in the banknote can be made visible in the form of a light and shadow image. A crease on a note of which one side faces the light source will show a pattern wherein this side will reflect more light than the uncreased part of the note, whereas the side of the crease facing away from the light source will reflect less light towards the camera than the uncreased part of the note. The part of the note that is transparent in infrared will in this way be built up from parts that show more or less average note reflection (uncreased parts), parts showing more than the average note reflection and parts showing less than the average note reflection. By determining the standard deviation of the note image taken in this way, a measure for the crease level of a note can be obtained.

[0076] Said standard deviation can be determined for the full infrared transparent part, but use can also be made here of the grid of 5* 10 squares. In this way it is possible to consider the specific weak locations of the note (for instance the main folding lines) separately.

[0077] In this way creases that are parallel to the light beam cannot be detected or hardly so. A solution to solve this could be to use a second infrared light source which is placed at an angle of 90° to the first light source and which light sources are alternately switched on and off so that in the end a full crease image can be formed.

5.2 Crease level determination by transmission in IR range.

[0078] Another method for determining the crease level of a note is making an infrared image in transmission of the note. In figure 3 an example is shown of such an image.

[0079] Creases in this case become visible because a crease ensures that the light has to traverse a larger distance in paper than a ray of light that has to traverse a perpendicularly positioned surface. A second reason why creases can be properly recognised in transparency is caused by the fibre structure being disturbed by the crease, which results in light at the location of the disturbed fibre structure being more strongly scattered and the light yield through the crease being less than through a note part without creases.

[0080] The combination of both images can further be used to analyse the nature of the creases. In practice there seem to be two main groups of crease problems. The first group is the group with relatively new notes which in one or another have been creased very strongly. Because the paper is still rather new, notes with this type of creases show a rough surface having relatively large differences in height.

[0081] A second main type of crease level is formed by paper that has been in circulation for a long time. This paper has lost its firmness and because of the numerous disturbances of the fibre structure the note has become limp.

[0082] It has turned out that the two groups can be distinguished from each other because in the first group a strong light shadow pattern arises. Said shadow pattern should then be reasonably found back in the second image, the infrared image in transparency.

[0083] It furthermore turned out that the second group shows a less pronounced shadow image, but a stronger pattern in transparency. In this case many lines may be detected that have no clear shadow patterns.

6. Balanced distribution of reflection over the note (smudge determination)

[0084] Another source of additional information may be found in a so-called balance coefficient. Said coefficient makes it possible to recognise local soiling on a note, which may manifest itself in larger or smaller smudges, but also in the form of unequally distributed soiling on one of the note halves or note quadrants.

[0085] When determining the balance coefficient it is first determined what the average part in the reflection of the overall note of each grid square is for a certain denomination. As described above in this case use is made of the blue component of the spectrum. The determination of this average therefore takes place over a larger number of notes of varying circulation quality. Subsequently in the individual note test the part of each grid square in the measured reflection is compared with the average and the overall balance coefficient is determined by means of an algorithm. Currently the algorithm consists of summing up the square of the relative deviations of the 50 grid squares, but this algorithm may be further optimised by further analysis.

[0086] In practice it turned out that this is a reasonably good method to recognise strongly deviating patterns, for instance caused by small smudges having strong contrast, or large smudges having little contrast. By connecting this information intelligently with the information found earlier on, it turns out that in addition to the generally soiled notes also locally soiled notes can be properly recognised with this algorithm.

7. Intelligent joining of the various sub tests into one test decision.

[0087] The final algorithm taking the decision about whether or not to approve of the note is an intelligent combination of the above-mentioned factors, optionally supplemented with other sources of information (for instance the time that the note has been in circulation).

[0088] By means of the multiple regression technique a formula has been derived wherein the response variable (classification 1-5) is predicted by the various predictor variables. Use is then made of a test set consisting of at least 5*100 notes that have been visually classified into the five classes.

[0089] The larger the test set becomes, the more reliable the developed algorithm will be. Advantages of this method are:

• the statistic technique itself determines the relevance of the various factors that have been measured;
• only one test threshold needs to be set instead of a number depending on the number of measured factors;
• the effect of a change of a test threshold can be predicted on the basis of simulation using test sets.

Claims

1. Method for sorting security documents, preferably banknotes, wherein from the reflection of light in a wavelength range of approximately 400-500 nm of at least a part of a security document a degree of soiling of the security document is determined.

2. Method according to claim 1, wherein for determining the reflection of light in the wavelength range, the B component from the RGB components of a colour image is used.

3. Method according to claim 1 or 2, wherein the wavelength range has been selected from approximately 400-450 nm, preferably approximately 410-430 nm.

4. Method according to claim 1, 2 or 3, wherein from reflection differences between a used security document and a new security document of light in the wavelength range the degree of soiling of the used security document is determined.

5. Method according to claim 1, 2 or 3, wherein from reflection of light in a first wavelength range and in a second wavelength range for a same location on the security document, the degree of soiling of the security document is determined, wherein the first wavelength range is the wavelength range of approximately 400-500 nm, preferably approximately 400-450 nm, preferably 410-430 nm.

6. Method according to claim 5, wherein from reflection ratios between reflection of light in the first wavelength range and the second wavelength range for a same location on the security document, the degree of soiling of the security document is determined, preferably from a quotient of the reflection.

7. Method according to claim 6, wherein the second wavelength range is a red wavelength range, and wherein the quotient preferably is a quotient of an R and B component from the RGB components of a colour image.

8. Method according to claim 6, wherein the first wavelength range has been selected from approximately 410-430 nm, preferably approximately 420-430 nm, and the second wavelength range from approximately 430-500 nm, preferably approximately 450-460 nm.

9. Method according to any one of the preceding claims, wherein the determination of the reflection comprises an image, preferably a digital image, of at least a part of the security document.

10. Method according to any one of the preceding claims, wherein the security document is subdivided into several areas, preferably rectangles, and wherein of each area an average reflection is determined, wherein preferably the average reflection is compared and a predetermined average reflection, wherein the areas preferably are predetermined areas having a predetermined position on the carrier, and wherein the areas preferably have individually been predetermined for each type of carrier.

11. Method according to any one of the preceding claims, wherein a visual degree of soiling of a test set is first of all visually established, and the reflection of the same test set is determined, and the visual degree of soiling is correlated by means of statistical calculation to the reflection measurement or at least one parameter derived therefrom for determining a calibration, and said calibration is applied to the reflection measurement or the at least one parameter derived therefrom for determining a degree of soiling.

12. Method according to claim 10 or 11, wherein for the determination of the soiling the reflection is measured in one or more areas on the security document that have been lightly printed or have not been printed on, wherein the area or the areas preferably have a representative surface for the area in question of the security document.

13. Method according to any one of the preceding claims, wherein the sorting criteria are adjusted using a test set of security documents that have been visually assessed, wherein soiling degrees are allocated to the degree of soiling determined by means of the reflection measurements.

14. Method according to claim 13, wherein by means of statistical calculation for the test set a relation is calculated between the allocated soiling degrees and measured degree of soiling, and the calculated relation is used to allocate a soiling degree to the security documents to be sorted, wherein preferably five soiling degrees are allocated, which preferably have a numerical value.

15. Method for sorting security documents, preferably banknotes, wherein from at least one image of at least a part of a security document a degree of soiling, and a crease level and/or a smudginess is determined, and from among others those determinations the security document is classified, preferably in a class of security documents that can be brought into circulation again and a class of security documents that can no longer be brought into circulation.

16. Method according to claim 15, wherein by means of a reflection measurement wherein one or more images of at least a part of a security document is or are taken and from that image or those images at least the degree of soiling, and the crease level and/or the smudginess are determined.

17. Method according to claim 15 or 16, wherein the degree of soiling is determined by means of the method of any one of the claims 1-14.

18. Method according to claim 15, 16 or 17, wherein a security document is exposed to infrared light, by means of a first image acquisition device at least one image of at least a part of the security document is taken in the infrared light, and a degree of creases of the security document is determined from the at least one infrared image, wherein the security document is exposed to infrared light that preferably has an angle of incidence to the surface of the security document, and a reflection image is preferably taken approximately transverse to the plane of the security document, wherein infrared light preferably has an angle of incidence of 20-40 degrees, preferably approximately 30 degrees.

19. Method according to claim 18, wherein an image, preferably a digital image, is taken by means of an image acquisition device that is sensitive in the infrared range.

20. Method according to claim 18 or 19, wherein the image acquisition device is positioned over the security document, preferably approximately straight over it.

21. Method according to claim 18, 19 or 20, wherein from the image a standard deviation of the brightness is determined, from which a crease level is determined.

22. Method according to claim 21, wherein for areas on the security document, preferably rectangles, more preferably a grid of rectangles, such as 5*10 rectangles divided over the security document, the standard deviation of the brightness is determined.

23. Method according to any one of the claims 18-22, wherein the security document is exposed by means of a second light source, positioned at an angle to a surface of the security document and to the first light source.

24. Method according to any one of the claims 18-23, wherein a crease level is at least placed in a class of relatively new security documents with sharp folding lines and a class of worn security documents having creases.

25. Method according to any one of the claims 18-24, wherein the image is taken in transmission, wherein the security document is exposed at the one surface and at least one image is taken from the opposite surface.

26. Method according to claim 25, wherein at least one image is taken in transmission and at least one image is taken wherein infrared light is at an angle of incidence to the surface of the security document, and a reflection image is taken of said surface, preferably approximately straight above said surface.

27. Method according to any one of the preceding claims 15-26, wherein at least one image is taken of at least a part of a security document, the image or images is or are subdivided into predetermined, fixed sub areas, of each of the sub areas an average reflection is determined, said average reflection values are each compared with predetermined calibration values, and from the comparison a degree of smudginess is determined.

28. Method according to claim 27, wherein the established calibration values are determined from average values for a test set of several notes, and wherein the image preferably is taken using an exposure with light from the wavelength range of 400-500 nm, or is taken through a filter absorbing light in the wavelength range of 400-500 nm.

29. Method according to claim 27 or 28, wherein the security document is given a class indication based on the degree of smudginess.

30. Method for sorting documents, particularly security documents, wherein a document is exposed to light having a wavelength component in the range of approximately 400-500 nm, by means of an image acquisition device an image of the security document is taken in the wavelength range of approximately 400-500 nm, the image taken is compared with the image taken beforehand in a similar, preferably the same, security document, after which on the basis of the comparison a degree of soiling of the document is determined.

31. Assembly provided with an exposure unit, an image acquisition device and a data processing unit provided with software for when said software is active in the data processing unit, carrying out the method according to any one of the preceding claims.

32. Carrier provided with machine instructions, such as computer software, for when active in the data processing unit, carrying out the method according to any one of the preceding claims.

Drawing