Unresolved target detection improvement by use of multiple matched filters approach at different spatial phases

Description

Background of the invention

1. Field of the Invention

[0001] The present invention relates generally to unresolved target detection using an infrared focal plane array and matched filters, and it more particularly relates to an unresolved target detection technique using multiple matched filters.

2. Description of the Related Art

[0002] With the advancement of IR FPA manufacturing technology, IR FPA has been broadly used for sensors in all three major platform: airborne sensing, satellite sensing, as well as ground sensing. For example, passive IR (Infrared) sensors are widely used to detect the energy emitted from targets, backgrounds, incoming threats, and the atmosphere for a plurality of applications including military surveillance, missile target and detection systems, crop and forest management, weather forecasting, and other applications.

[0003] U.S. Patent Application publication US 2003/0184468 by Hai-Wen Chen and Teresa Olson, entitled "Integrated Spatio-Temporal Multiple Sensor Fusion System Design" provides a theoretical evaluation for different averaging processes that can reduce random noise and enhance target signatures. The inventor is a coauthor of several related papers including Hat-Wen Chen and Teresa Olson, "Integrated Spatio-Temporal Multiple Sensor Fusion System Design," SPIE AeroSense, Proceedings of Sensor and Data Fusion Conference, vol. 4731, pp. 204-215, Orlando, FL, 1-5 April, 2002; Hal-Wen Chen and Teresa Olson, "Adaptive Spatiotemporal Multiple Sensor Fusion."Journal of Optical Engineering. vol. 42 (5), pp. 1481-1495, May, 2003.

[0004] The Matched Filter (MF) method is currently a popular approach for unresolved target detection using IR FPAs as sensor detectors. In the MF method, DPSF (discrete point spread function sampled by discrete pixels in IR FPA) is estimated from CPSF (continuous point spread function). CPSF is available based on the sensor optical and lens designs. A matched spatial filter is obtained by dividing the DPSF with the co-variance matrix of background clutter. This matched filter is optimal In an MSE (mean-square-error) sense in that it provides a maximum SCNR (signal to clutter noise ratio) for a point source (unresolved) target.

[0005] In current advanced optical designs, most energy of a DPSF can be contained within a 3x3 pixel area, and the PVF (point visibility function) can be as high as 0.6-0.75. A 0.7 PVF means that if the peak of a CPSF is located at the center of a pixel, this pixel will contain 70% of the energy of the CPSF and 30% of its energy is spread out in the neighbor pixels. Although CPSF is a symmetrical Mexican-hat shape function, the shape of a DPSF varies depending on the spatial phases. Spatial phase means the location of the CPSF peak at the sub-pixel space. If the peak is aligned with the center of a pixel, we call it a center phase. Similarly, a corner phase means that the peak falls down on a corner of a pixel. In this case, all the four pixels nearby that corner will receive equal energy from the CPSF. Therefore, it is clear that a 3x3 DPSF of a center phase has a totally different shape of energy distribution from a 3x3 DPSF of a corner phase, as evidenced in Figs. 2(a) and (b). The PVF of the CPSF is 0.73.

[0006] Theoretically there are infinite different phases. In practice, we can approximate the infinite phases by dividing a pixel into multiple sub-pixels. For example, if we divide a pixel into 11x11 sub-pixels, then we have 121 different phases to approximate all the infinite phases. At any time moment, any sub-pixel location should have an equal probability to be aligned with the CPSF peak. That is, the spatial phase is a random variable with a uniform distribution.

[0007] Samson et al. "Point target detection and subpixel position estimation in optical imagery", Applied Optics, vol. 43, no 2, 10 January 2004, pages 257-263, disclose detectors that are alternatives to the conventional matched filters. Samson et al. take their starting point in a likelihood ratio test corresponding to the conventional matched filtering approach. The likelihood ratio test is then generalized in different ways in order to take the subpixel phase of the target into account. In one example the likelihood ratio test is combined with maximum likelihood estimates of the amplitude and the phase of the point spread function. In another example a Bayesian model is used where the amplitude and the phase are modelled by random variables having a known a priori distribution.

[0008] Pohlig, "An algorithm for detection of moving optical targets", IEEE Trans. on Aerospace & Electr. Systems, vol. 25, no 1, 1 January 1989, discloses an algorithm for detecting moving optical targets against a spatially non-stationary Poisson background and noise. The algorithm has applications in optical detection of objects such as meteors and satellites against a stellar background. The method uses a maximum likelihood approach to reduce the interference from stars. The algorithm may be implemented using convolution with filters.

[0009] Fernandez et al., "Optimal subpixel-level IR frame-to-frame registration", Proceedings SPIE signal and data processing of small targets, vol. 1481, 1 April 1991, discloses a method for image registration at a subpixel level. The method uses several data frames which are simultaneously matched to a common reference frame. By considering several data frames which are collected over time and not each frame separately, the performance is improved compared to traditional methods which use interpolation in one frame at a time.

[0010] From the discussion above, it is clear that the random phase causes problems in target detection using the MF method. In the traditional MF method approach, the DPSF of center phase (or averaged phase) is used to obtain the matched filter. Therefore, if the target center is located near the center of a pixel, the MF method performs well. However, if the target center is located near pixel corners or edges, the performance will be worse because the matched filter is not matched to the DPSF of the corner (or edge) phase.

[0011] Accordingly, there is a need to improve target detection of a point source target, when utilizing a matched filter and when the target center is located away from the pixel center.

Summary of the Invention

[0012] The method and system of the present invention, which includes multiple matched filters, substantially overcome the previously described problem of a point source target center not being aligned with a pixel center when utilizing a matched filter. A sensor, which is preferably an Infrared focal plane array, includes a plurality of pixels for providing image data. A processor processes the image data from a selected pixel corresponding to a potential point source target and from a plurality of neighboring pixels. The multiple filters have a plurality of different phases, including at least a center phase and four corner phases. The multiple filters filter the image data to obtain different filtered spatial images. The filtered spatial images are averaged and preferably normalized. The intensities of the filtered spatial images are also preferably summed. A detector, which is preferably a CFAR detector, detects a target based upon the spatially filtered images from the multiple matched filters. Preferably, either five or nine matched filters are utilized.

[0013] In the present invention, an MMF (multiple matched filters) method is described that can significantly improve detection performance for unresolved targets. Performance evaluations were conducted using images from an airborne sensor and a satellite sensor.

Brief Description of the Drawings

[0014] 

Fig. 1 illustrates an exemplary SMF (single matched filter) image processing system 100 found in the prior art.

Figs. 2A and 2B are matrices which respectively represent a 3x3 DPSF of a pixel center phase, and a 3x3 DPSF of a pixel corner phase.

Fig. 3 is a photograph of a cloud scene that appears as a background clutter image.

Fig. 4 is a diagram illustrating the performance of a traditional single MF method having six phases among a possible 121 phases.

Figs. 5A and 5B are graphs that compare the performances of seven different matched filters.

Fig. 6A and 6B are diagrams illustrating both a five MMF approach and a nine MMF approach.

Fig. 7 is a block diagram of the five MMF detection process.

Fig. 8 is an illustration of the averaging process for the phase associated with an upper left corner.

Fig. 9 is an illustration of the averaging process for the nine MMF Method.

Figs. 10A and 10B are graphs that show detection performance comparisons between the five-MMF method (dashed curve with triangles) and the traditional single MF method using the averaged phase filter (solid curve with circles).

Figs. 11A and 11B are graphs that show detection performance comparisons between the nine-MMF method (dashed curve with triangles) and the traditional single MF method using the averaged phase filter (solid curve with circles).

Fig, 12 shows a relatively low clutter scene in a satellite image.

Fig. 13 shows a relatively medium/heavy clutter scene in a satellite image.

Fig. 14 shows the residual image after LS time-differencing from the medium/heavy clouds clutter shown in Fig. 13.

Fig. 15 is a graph showing ROC curve comparisons for a low clutter scene.

Fig. 16 is a graph showing ROC curve comparisons: before and after LS time-differencing.

Fig. 17 is a graph showing ROC curve comparisons for the residual image after time-differencing.

Detailed Description of the Invention

[0015] In order to facilitate the reader's understanding of the present invention, the following list of acronyms and their general meanings are provided:

CFAR - Constant False Alarm Rate

CPSF - Continuous Point Spread Function

DPSF - Discrete Point Spread Function

FOR - Field of Regard

FOV - Field of View

FPA - Focal Plane Array

IR - Infrared

IRST - Infrared Search and Track

LOS - Line of Sight

LS - Least-Square

MF - Matched filter

MMF - Multiple Matched Filters

MFA - Multiple-Frame-Association

MSE - Mean-Square-Error

MTI - Moving Target indicator

NEDT- Noise-Equivalent Temperature Difference

Pd - Probability of Detection

Pfa - Probability of False-Alarm

PSF - Point spread function

PVF - Point Visibility Function

ROC - Receiver Operating Characteristics

S/N - Signal-to-Noise Ratio

SCNR - Signal to Clutter Noise Ratio

STD - Spot Time-Differencing

[0016] The present invention is directed to a MMF (Multiple Matched Filters) method having multiple matched filters at different spatial phases to improve detection performance. This approach applies a neighbor-fusion strategy for spatial noise reduction to increase SCNR. The phasing problem is a universal problem to different sensors using IR FPA. In addition to improving detection performance, the outputs from MMF can also be used for improving sub-pixel centroiding performance.

[0017] The present invention has been tested with a cloud scene as the background clutter image, as shown in Fig. 3. In the preliminary testing, a point-source target was randomly inserted to the cloud background in 121 different phases. For each phase, it was randomly inserted twelve times. Accordingly, the total target insertion number was 121x12 = 1452.

[0018] Fig. 4 illustrates the performance of a traditional single MF method having six phases among a possible 121 phases. Phase P1 is the left-upper corner phase, and phase P6 is the center phase. The traditional MF method uses a single MF. Its performance is shown in Fig. 5 at two different SCNRs, The performances of seven different matched filters were compared. MFs # 1 to 6 are obtained by using different DPSFs with the six different phases shown In Fig. 4. MF # 7 is obtained by using the averaged DPSF of the 121 phases. It is seen that the averaged-phase MF performs the best among the seven MFs.

[0019] Instead of using a single MF, the present invention is directed to a multiple MF (MMF) method where the input image is filtered by MMFs, and the multiple filtered images are processed using a neighbor-fusion scheme to suppress random noise and to increase SCNR. The present invention includes both a five MMF approach and a nine MMF approach. As shown in Fig. 6, the phases of the five MMFs are the corner phases P1-P4 and the center phase P5, and the nine MMFs have four additional edge phases.

[0020] The block diagram of the five MMF detection process is illustrated in Fig. 7. A reference image (t) is processed in step 60 to estimate a variance of sigma and to choose a spatial filter 61-65. The "averaging process" step 66 in Fig. 7 is the core function that averages the four corner phasing filter outputs at different pixel locations. This function can reduce spatial random noise standard deviation (sigma) by half without reducing the target intensity at the corners. The next "additive process" step 67 then sums the detected intensities from all the filters 61-65 at the five phase locations (or nine phase locations in the nine MMF technique) before flowing to the "CFAR detection" step 68.

Averaging Process - Neighbor Fusion

[0021] As disclosed in the copending patent application S.N. 10/395,269 by Chen et al, entitled "Integrated Spatio-Temporal Multiple Sensor Fusion System Design", averaged fusion (equivalent to additive fusion) performs better than other fusion strategies such as MAX or MIN fusions. For example, the variance of a RV (random variable) can be reduced to one fourth of its original variance by averaging four of its events at different spaces or times.

[0022] As shown in Fig. 7, the inputs to the averaging process 66 are five SNR images. The original image is first filtered by use of the five MF's 61-65 of different phases to obtain five different spatial images. The five spatial images are then normalized (divided) by a background estimator to obtain five SNR images. It should be noted that the background normalization process is a nonlinear process.

[0023] As shown in Fig. 2(b) and Fig. 8, for the sub-pixel phase P1 (left-upper corner phase), the point source target will have its energy equally distributed in the pixel 84 and its three neighbor pixels: the left pixel 83, the left-upper pixel 81, and the upper pixel 82. As shown in Fig. 8, for the left pixel 83, the energy can be optimally detected using the MF of phase P2 (right-upper phase). For the left-upper pixel 81 and the upper pixel 82, the energy can be optimally detected using the MF of phase P4 (right-lower phase) and phase P3 (left-lower phase), respectively.

[0024] The averaging process (Neighbor Fusion) is expressed as

where snr_ 1(ii,jj) is the SNR image filtered using MF of phase 1, snr_2(ii,jj-1) is the SNR Image filtered using MF of phase P2 with one pixel shift to the left, snr_3(ii-1,jj) is the SNR image filtered using MF of phase P3 with one pixel shift to the upper direction, and snr_4(ii-1,jj-1) is the SNR image filtered using MF of phase P2 with one pixel shift to the left and one pixel shift to the upper direction.
Similarly we can obtain neighbor fusion for the other three corner phases:



and

[0025] For the nine MMF method of Fig. 9, there are four additional MF's of edge phases: phase P2 (upper-edge phase), phase P4 (left-edge phase), phase P5 (right-edge phase), and phase P7 (lower-edge phase). The averaging process (Neighbor Fusion) is expressed as





and

An Alternative Way of Averaging Process (Neighbor Fusion)

[0026] Since the spatial MF process and the averaging process are both linear processes, an alternative averaging process is to first average the four MFs with corner phases with appropriate pixel shifts. The resulting averaged MF is of a larger size. The original MF's have a size of 3x3. The averaged MF will have a size of 5x5. The four 5x5 averaged MFs are then used to filter the original image to obtain four different spatial images. They are then normalized by a background estimator to obtain four SNR images. Similarly, for the nine-MMF method, the edge MFs can be averaged to a larger 3x5 or 5x3 MF depending on its edge locations.

Modification of SNR image Processed by MF of Center Phase

[0027] It is desirable for the output of the MF with a center phase to be large when the point source target phase is near the pixel center. On the other hand, it is desirable that its output be small when the point source target phase is near the pixel corners, It has been observed that the values of the eight surrounding pixels are quite random with positive and negative values when the target phase is close to the center, and most of the 8 surrounding values are relatively large positive values, when the target phase is close to the corners. Based on this observation, the SNR image processed by the MF of center phase is modified.

[0028] For the five-MMF method.

where Residue = Sum(8 surrounding pixels) /8.
For the nine-MMF method,

where Residue = Sum(8 surrounding pixels) / 8.

Last Stage- Additive Process

[0029] All the neighbor-fused images are then added up as the sum of all the detection intensities. The resulting image is called neighbor-fused SNR image.
For the five-MMF method,

For the nine-MMF method,

The obtained neighbor-fused SNR image goes to the traditional CFAR process to report detections for a specie false detection rate Pfa, or go to a multiple thresholding process to generate ROC curves to evaluate the performance of the algorithms.

Airborne IR Sensor Clouds Background Clutter

[0030] Figs. 10A and 10B show detection performance comparison between the five-MMF method (dashed curve with triangles) and the traditional single MF method using the averaged phase filter (solid curve with circles) at SCNR = 2.2 (Fig. 10(a)) and at SCNR = 3.2 (Fig. 10(b)). It seen that detection performance is much improved by use of the five-MMF method. The Pfa is reduced by about 7-10 times.

[0031] Figs. 11A and 11B show detection performance comparisons between the nine-MMF method (dashed curve with triangles) and the traditional single MF method using the averaged phase filter (solid curve with circles) at SCNR = 2.2 (Fig. 11(a)) and at SCNR = 3.2 (Fig. 11 (b)) It is seen that detection performance is much improved by use of the nine-MMF method. The Pfa is reduced by about 20-30 times. That is, with 4 additional MFs of edge phases, the Pfa is further reduced by about 3 times, compared with the results using the five-MMF method

Satellite IR Sensor Clouds Background Clutter

[0032] A low clutter and a medium/heavy clutter scene of satellite images are shown in Figs. 12 and 13, respectively. The STD (standard deviation) of the medium/heavy clutter is almost 15 times as large as the STD of the low clutter (the medium/heavy clutter in the satellite image is similar to heavy clutter in airborne sensor images). The MMF method of the present invention is most efficient for improving detection for low and medium background clutter where random noise contributes as well as the correlated clutter noise. For example, the clouds scene in Fig. 3 is considered as a medium clutter for airborne IR sensors.

[0033] The medium/heavy clutter shown in Fig. 13 is dominated by spatially correlated clutter noise. For this kind of images, we first apply an LS (least-square) time-differencing process of the type described in M. Fernandez, A. Aridgides, D. Randolph, and D. Ferris, "Optimal Subpixel-Level IR Frame-to-Frame Registration," SPIE Proceedings of Signal and. Data Processing of Small Targets, vol. 1481, pp. 172-179, 1991. This technique reduces most of the correlated clutter and the MMF method is applied to further improve detection performance. Fig. 14 shows the residual image after the LS time-differencing (using a 5x5 correlated filter) from the medium/heavy clouds clutter shown in Fig. 13. The clutter STD is significantly reduced from 31 counts to 1.84 counts.

[0034] Fig. 15 shows detection performance comparison between the nine-MMF method (dashed curve with triangles) and the traditional single MF method using the averaged phase filter (solid curve with circles) at SCNR = 2.2 for the low clutter background shown in Fig. 12. It is seen that detection performance is much improved by use of the nine-MMF method, The Pfa is reduced by more than 10 times.

[0035] Fig. 16 shows detection performance comparison before and after LS time-differencing, and also shows the performance of direct time-differencing without using LS process. The background clutter used is the one shown in Fig. 13. The curve with 'triangle' symbols In Fig. 16 shows the detection performance on the original clutter image (before time-differencing), The target intensity relative to the original clutter STD is SCNR = 0.5. The other three curves in Fig. 16 also used the same target intensity. The curve with 'star' symbols in Fig. 16 shows the detection performance of direct time-differencing (direct subtraction between the current frame and the previous frame without using LS correlation process). It is seen that direct time-differencing performs better than no time-differencing.

[0036] The curves with 'circle' symbols and 'cross' symbols in Fig. 16 show the detection performance of LS time-differencing using a 5x5 and a 3x3 correlation filters, respectively. The 5x5 filter performs a little better. It is seen that LS time-differencing performs much better than no time-differencing. The Pfa is reduced by more than 100 times. All the four curves in Fig. 15 are performances using the traditional single MF method with the averaged phase filter. In Fig. 16, we will show that by applying the nine-MMF method we can further reduce the Pfa by another factor of 30 times, so that the total false alarms reduction is about 3,000 times.

[0037] Fig. 17 shows detection performance comparison between the nine-MMF method (dashed curve with triangles) and the traditional single MF method using the averaged phase filter (solid curve with circles) at SCNR = 2.2 for the residual clutter background (after LS time-differencing) shown in Fig. 14. It is seen that detection performance is much improved by use of the nine-MMF method. The Pfa is reduced by about 30 times. Note that the SCNR (=2.2) is relative to the residual clutter (STD = 1.84). If we convert the target intensity relative to the original clutter (STD = 31), the SCNR is 0.13. It is a very low SCNR but a relatively good detection performance can be obtained by combining the LS time-differencing and MMF method as shown in Fig. 17.

[0038] It should be noted that the MMF (Multiple Matched Filters) approach can significantly reduce the Pfa for a specific Pd, or can significantly increase Pd for a specific Pfa. For example, as shown in Fig. 11 (a), for a Pfa = 1 E-4, Pd is increased from 50% to 82%.

[0039] Since the MF size is small (3x3), the processing time for MMF method is still fast enough. For example, for the five-MMF method, the processing time required for spatial processing the five multiple MFs is less than that required for a single MF process with a larger filter size of 7x7.

[0040] It should be also be noted that the nine MMF method performs better than the five MMF method. Therefore, the MMF method can be further improved by adding more MFs with edge phases.

[0041] Effort in developing the MMF method has a especially beneficial effect, since the outputs from MMF can be further used for improving sub-pixel centroiding performance. For a low/medium SCNR, the centroiding errors can be reduced by a factor of 40%, and for a high SCNR, the centroiding errors can be further reduced by more than 50%.

[0042] In general, the performance for heavy clutter is worse than that for low clutter, even though time-differencing techniques are applied This is because the time-differencing process will cause a 40% increase of random noise Sigma. The present has shown that by combining time-differencing techniques with the MMF method, the Pfa caused by both the correlated clutter and random noise can be significantly reduced, and thus the performance for heavy clutter will be close to that for low clutter.

[0043] The random phasing of unresolved targets falling onto an IR FPA pixel will cause reduced detection performance, and this problem is universal to all systems that use IR FPAs as sensor detectors. Therefore, the present invention has wide applicability to many different systems.

Claims

1. A method for detecting a point source target using multiple matched filters, comprising the steps of:

receiving image data from a sensor having a plurality of pixels;

processing the image data from a selected pixel (84) corresponding to a potential point source target and from a plurality of pixels neighboring the selected pixel (81, 82, 83);

selecting (60) a plurality of spatial filters (61-65) having a plurality of different phases (P1-P5), including at least a center phase (P5) and four corner phases (P1-P4);

filtering the image data with the selected spatial filters (61-65) to obtain filtered image data;

averaging (66) the filtered image data, wherein the averaging includes averaging the outputs of four corner phase filters of four neighboring corners with at least a one pixel shift; and

detecting (68) a target based upon the averaged filtered image data.

2. A method according to claim 1 wherein at least five spatial filters (61-65) are selected.

3. A method according to claim 1 wherein at least nine spatial filters (61-65) are selected.

4. A method according to claim 3 wherein the spatial filters (61-65) include at least four edge phases.

5. A method according to claim 1 including the step of normalizing the filtered image data by dividing the filtered image data with a background estimation.

6. A method according to claim 1 which further includes the step of summing (67) the intensities of the filtered image data.

7. A method according to claim 1 wherein the averaging includes an average of the four corner phasing filter outputs at different pixel locations.

8. A method according to claim 1 wherein the detecting step (68) includes a constant false alarm rate detection.

9. A method according to claim 1 wherein the detecting step (68) includes a multiple thresholding technique.

10. A method according to claim 1 wherein a least square time-differencing process is applied to the image data prior to filtering.

11. A system for detecting a point source target using multiple matched filters, comprising:

a sensor having a plurality of pixels for providing image data;

a processor for processing the image data from a selected pixel corresponding to a potential point source target (84) and from a plurality of pixels neighboring the selected pixel (81, 82, 83); and

a plurality of spatial filters (61-65) having a plurality of different phases, including at least a center phase (P5) and four corner phases (P1-P4), said spatial filters (61-65) filtering the image data to obtain filtered image data ;
characterized in that

said processor is arranged to average (66) the filtered image data, wherein the average includes an average of the outputs of four corner phase filters of four neighboring corners with at least a one pixel shift; and

the system further comprises a detector for detecting a target based upon the averaged filtered image data.

12. A system according to claim 11 wherein at least five spatial filters (61-65) are utilized.

13. A system according to claim 11 wherein at least nine spatial filters (61-65) are utilized.

14. A system according to claim 11 wherein the spatial filters (61-65) include at least four edge phases.

15. A system according to claim 11, wherein said processor is arranged to normalize the filtered image data by dividing the filtered image data with a background estimation.

16. A system according to claim 11 wherein the intensities of the filtered image data are summed.

17. A system according to claim 11 wherein the detector includes a constant false alarm rate detector.

18. A system according to claim 11 wherein the detector includes multiple thresholds.

19. A system according to claim 11 wherein the sensor includes an infrared focal plane array.

Ansprüche

1. Verfahren zum Detektieren eines Punktquellenziels unter Verwendung mehrerer angepasster Filter, wobei das Verfahren die folgenden Schritte umfasst:

Empfangen von Bilddaten von einem Sensor, der mehrere Pixel aufweist;

Verarbeiten der Bilddaten von einem ausgewählten Pixel (84), das einem potenziellen Punktquellenziel entspricht, und von mehreren Pixeln, die sich neben dem ausgewählten Pixel (81, 82, 83) befinden;

Auswählen (60) mehrerer räumlicher Filter (61-65), die mehrere verschiedene Phasen (P1-P5) haben, zu denen mindestens eine mittige Phase (P5) und vier Eckphasen (P1-P4) gehören;

Filtern der Bilddaten mit den ausgewählten räumlichen Filtern (61-65). um gefilterte Bilddaten zu erhalten;

Mitteln (66) der gefilterten Bilddaten, wobei das Mitteln das Mitteln der Ausgangssignale von vier Eckphasenfiltern von vier benachbarten Ecken mit mindestens einer Einpixelverschiebung enthält, und

Detektieren (68) eines Ziels anhand der gemittelten gefilterten Bilddaten.

2. Verfahren nach Anspruch 1, wobei mindestens fünf räumliche Filter (61-65) ausgewählt werden.

3. Verfahren nach Anspruch 1, wobei mindestens neun räumliche Filter (61-65) ausgewählt werden.

4. Verfahren nach Anspruch 3, wobei die räumlichen Filter (61-65) mindestens vier Randphasen enthalten.

5. Verfahren nach Anspruch 1, das den Schritt des Normalisierens der gefilterten Bilddaten durch Teilen der gefilterten Bilddaten mit einer Hintergrundschätzung enthält.

6. Verfahren nach Anspruch 1, das des Weiteren den Schritt des Summierens (67) der Intensitäten der gefilterten Bilddaten enthält.

7. Verfahren nach Anspruch 1, wobei das Mitteln einen Durchschnitt der vier Eckphasenfilter-Ausgangssignale an verschiedenen Pixelpositionen enthält.

8. Verfahren nach Anspruch 1, wobei der Schritt des Detektierens (68) das Detektieren einer konstanten Falschalarmrate enthält.

9. Verfahren nach Anspruch 1, wobei der Schritt des Detektierens (68) eine Mehrschwellentechnik enthält.

10. Verfahren nach Anspruch 1, wobei vor dem Filtern ein Fehlerquadrat-Zeitdifferenzierungsprozess an den Bilddaten vollzogen wird.

11. System zum Detektieren eines Punktquellenziels unter Verwendung mehrerer angepasster Filter, wobei das System Folgendes umfasst:

einen Sensor, der mehrere Pixel aufweist, zum Bereitstellen von Bilddaten;

einen Prozessor zum Verarbeiten der Bilddaten von einem ausgewählten Pixel, das einem potenziellen Punktquellenziel (84) entspricht, und von mehreren Pixeln, die sich neben dem ausgewählten Pixel (81, 82, 83) befinden; und

mehrere räumliche Filter (61-65), die mehrere verschiedene Phasen haben, zu denen mindestens eine mittige Phase (P5) und vier Eckphasen (P1-P4) gehören, wobei die räumlichen Filter (61-65) die Bilddaten filtern, um gefilterte Bilddaten zu erhalten;

dadurch gekennzeichnet, dass

der Prozessor dafür ausgelegt ist, die gefilterten Bilddaten zu mitteln (66), wobei der Durchschnitt einen Durchschnitt der Ausgangssignale von vier Eckphasenfiltern von vier benachbarten Ecken mit mindestens einer Einpixelverschiebung enthält; und

wobei das System des Weiteren einen Detektor zum Detektieren eines Ziels anhand der gemittelten gefilterten Bilddaten enthält.

12. System nach Anspruch 11, wobei mindestens fünf räumliche Filter (61 - 65) verwendet werden.

13. System nach Anspruch 11, wobei mindestens neun räumliche Filter (61-65) verwendet werden.

14. System nach Anspruch 11, wobei die räumlichen Filter (61-65) mindestens vier Randphasen enthalten.

15. System nach Anspruch 11, wobei der Prozessor dafür ausgelegt ist, die gefilterten Bilddaten durch Teilen der gefilterten Bilddaten mit einer Hintergrundschätzung zu normalisieren.

16. System nach Anspruch 11, wobei die Intensitäten der gefilterten Bilddaten summiert werden.

17. System nach Anspruch 11, wobei der Detektor einen Detektor zum Detektieren einer konstanten Falschalarmrate enthält.

18. System nach Anspruch 11, wobei der Detektor mehrere Schwellen enthält.

19. System nach Anspruch 11, wobei der Sensor ein Infrarot-Brennebenen-Array enthält.

Revendications

1. Procédé de détection d'une cible source ponctuelle en utilisant de multiples filtres appariés, comprenant les étapes consistant à :

recevoir des données d'image d'un capteur ayant une pluralité de pixels ;

traiter les données d'image à partir d'un pixel sélectionné (84) correspondant à une cible source ponctuelle potentielle et à partir d'une pluralité de pixels voisins du pixel sélectionné (81, 82, 83) ;

sélectionner (60) une pluralité de filtres spatiaux (61-65) ayant une pluralité de phases différentes (P1-P5), comprenant au moins une phase de centre (P5) et quatre phases de coin (P1-P4) ;

filtrer les données d'image avec les filtres spatiaux sélectionnés (61-65) pour obtenir des données d'image filtrées ;

moyenner (66) les données d'image filtrées, dans lequel le moyennage comprend le moyennage des sorties des quatre filtres de phase de coin des quatre coins voisins avec au moins un décalage d'un pixel , et

détecter (68) une cible sur la base des données d'image filtrées moyennées.

2. Procédé selon la revendication 1, dans lequel au moins cinq filtres spatiaux (61-65) sont sélectionnés.

3. Procédé selon la revendication 1, dans lequel au moins neuf filtres spatiaux (61-65) sont sélectionnés.

4. Procédé selon la revendication 3, dans lequel les filtres spatiaux (61-65) comprennent au moins quatre phases de bord.

5. Procédé selon la revendication 1, comprenant l'étape consistant à normaliser les données d'image filtrées en divisant les données d'image filtrées par une estimation d'arrière-plan.

6. Procédé selon la revendication 1, comprenant en outre l'étape consistant à sommer (67) les intensités des données d'image filtrées.

7. Procédé selon la revendication 1, dans lequel le moyennage comprend une moyenne des quatre sorties de filtres de phase de coin à différents emplacements de pixels.

8. Procédé selon la revendication 1, dans lequel l'étape de détection (68) comprend une détection constante de taux de fausses alarmes.

9. Procédé selon la revendication 1, dans lequel l'étape de détection (68) comprend une technique de seuils multiples.

10. Procédé selon la revendication 1, dans lequel un procédé de différenciation temporelle par les moindres carrés est appliqué aux données d'image avant le filtrage.

11. Système de détection d'une cible source ponctuelle en utilisant de multiples filtres appariés, comprenant :

un capteur ayant une pluralité de pixels pour fournir des données d'image ;

un processeur pour traiter les données d'image à partir d'un pixel sélectionné correspondant à une cible source ponctuelle potentielle (84) et à partir d'une pluralité de pixels voisins du pixel sélectionné (81, 82, 83) ; et

une pluralité de filtres spatiaux (61-65) ayant une pluralité de phases différentes, comprenant au moins une phase de centre (P5) et quatre phases de coin (P1-P4), lesdits filtres spaciaux (61-65) filtrant les données d'image pour obtenir des données d'image filtrées ;

ledit processeur est agencé pour moyenner (66) les données d'image filtrées, dans lequel la moyenne comprend une moyenne des sorties des quatre filtres de phase de coin des quatre coins voisins avec au moins un décalage d'un pixel ; et

le système comprend en outre un détecteur pour détecter une cible sur la base des données d'image filtrées moyennées.

12. Système selon la revendication 11, dans lequel au moins cinq filtres spatiaux (61-65) sont utilisés.

13. Système selon la revendication 11, dans lequel au moins neuf filtres spatiaux (61-65) sont utilisés.

14. Système selon la revendication 11, dans lequel les filtres spatiaux (61-65) comprennent au moins quatre phases de bord.

15. Système selon la revendication 11, dans lequel ledit processeur est agencé pour normaliser les données d'image filtrées en divisant les données d'image filtrées par une estimation d'arrière-plan.

16. Système selon la revendication 11, dans lequel les intensités des données d'image filtrées sont sommées.

17. Système selon la revendication 11, dans lequel le détecteur comprend un détecteur constant de taux de fausses alarmes.

18. Système selon la revendication 11, dans lequel le détecteur comprend de multiples seuils.

19. Système selon la revendication 11, dans lequel le capteur comprend un réseau de plans focaux infrarouges.

Drawing