The invention relates to a method for detecting and tracking targets in a series of successive images, said method comprising limiting the number of spots which are the subject of simultaneous tracking or false leads. The operation of the tracking module is thereby improved without having to increas
The invention relates to a method for detecting and tracking targets in a series of successive images, said method comprising limiting the number of spots which are the subject of simultaneous tracking or false leads. The operation of the tracking module is thereby improved without having to increase a detection threshold of said spots. The detection threshold can even be reduced, such that the detection haul is increased and the tracking of each target is more continuous, without the probability of false alarms itself being increased.
대표청구항▼
1. A method for detecting and tracking targets in a series of electro-optical images of a same field of surveillance, comprising the following steps carried out for each image during processing: /1/ using a detection threshold, searching for pixels in the image which each have a signal-to-noise rati
1. A method for detecting and tracking targets in a series of electro-optical images of a same field of surveillance, comprising the following steps carried out for each image during processing: /1/ using a detection threshold, searching for pixels in the image which each have a signal-to-noise ratio value that exceeds the detection threshold;/2/ within a matrix of the image, grouping some of the pixels having signal-to-noise ratio values that exceed the detection threshold, when said pixels are close to one another within the image, so as to form a set of grouped spots;/3/ selecting some of the spots formed in step /2/, more limited in number than the set of spots formed in said step /2/, and/4/ testing a track formation sequence for the selected spots only, and declaring a target track for a chain of spots that are selected in different images if the track formation sequence is completely satisfied for said chain of spots;wherein the number of spots selected in step /3/ is limited based on an integrated intensity value Nc calculated for a target presence probability density distribution, produced for the track formation sequences already being tested before step /3/ is executed for the image currently being processed, and based on a number of spots Nt of said distribution which have individual weights exceeding a predetermined weight threshold,wherein the spots selected in step /3/ correspond to the track formation sequences which are already being tested, or have weight values greater than the weight values of the spots not selected, or have respective signal-to-noise ratio values that exceed a selection threshold, or each comprise a number of pixels that is greater than a minimum spot size, or satisfy criteria related to an assumed target type,the integrated intensity value Nc being an integral summation of the individual values of the probability density distribution for all points of the matrix of the image, andthe individual weight of each spot resulting from a combination of a spatial extension of the spot in the image and a contrast of this spot. 2. The method of claim 1, wherein the predetermined weight threshold used for the individual weight of each spot for which the track formation sequence is being tested, in order to determine the number of spots Nt, is equal to 0.5. 3. The method of claim 1, wherein the number of spots selected in step /3/ is between the number of spots Nt and the integrated intensity value Nc plus a predetermined margin Mf, or is equal to said number of spots Nt or to said integrated intensity value Nc plus the margin Mf. 4. The method of claim 3, wherein the number of spots selected in step /3/ is equal to the smallest among the number of spots Nt and the integrated intensity value Nc plus the margin Mf. 5. The method of claim 3, wherein the margin Mf is a non-zero integer less than or equal to five. 6. The method of claim 1, further comprising: producing the target presence probability density distribution from a current state of the track formation sequences initiated for images preceding the image currently being processed, and from the trajectories of spots corresponding to these track formation sequences initiated for the images preceding the image currently being processed, by applying kinematic criteria concerning spot movement between successive images;searching for a correlation between the target presence probability density distribution and the spots formed in step /2/; thenupdating the target presence probability density distribution by using a result of the correlation,and wherein the statistical characteristic relating to the track formation sequences already being tested when step /3/ is carried out for the image currently being processed, wherein the statistical characteristic is used to limit the number of spots selected in said step /3/, and wherein the statistical characteristic is obtained from the updated target presence probability density distribution. 7. The method of claim 1, wherein each spot has a Gaussian profile within the target presence probability density distribution. 8. A system for the detection and tracking of targets, comprising: a device for capturing electro-optical images;an image processing module adapted to calculate respective signal-to-noise ratio values for the pixels of each captured image, in order to group certain pixels having signal-to-noise ratio values that exceed a detection threshold, when said pixels are close to one another in the captured image, so as to form grouped spots;a tracking module, adapted to test a track formation sequence for at least some of the spots by using a chain of spots that are detected in different images, and to declare a target track for the chain of spots if the track formation sequence is satisfied for this chain of spots; anda spot selection module, arranged to receive as input, for each captured image, the spots formed by the image processing module, and to output some of the spots received as input, wherein the tracking module tests the track formation sequence only for those spots which are output by said selection module,wherein the spots which are output by said selection module correspond to track formation sequences which are already being tested, or have weight values greater than the weight values of the spots not selected, or have respective signal-to-noise ratio values that exceed a selection threshold, or each comprise a number of pixels that is greater than a minimum spot size, or satisfy criteria related to an assumed target type,wherein the spot selection module is adapted to select the spots that are output, while limiting the number of said output spots based on an integrated intensity value Nc calculated for a target presence probability density distribution, produced for track formation sequences that were already being tested before a selection is made for an image currently being processed, and based on a number of spots Nt of said distribution which have individual weights that exceed a predetermined weight threshold,the integrated intensity value being an integral summation of the individual values of the probability density distribution for all points of a matrix of the image, andthe individual weight of each spot resulting from a combination of a spatial extension of the spot in the image and a contrast of said spot. 9. The system of claim 8, adapted to carry out a method comprising the following steps: /1/ using the detection threshold, searching for pixels in the image which each have a signal-to-noise ratio value that exceeds the detection threshold;/2/ within the matrix of the image, grouping some of the pixels having signal-to-noise ratio values that exceed the detection threshold, when said pixels are close to one another within the image, so as to form grouped spots;/3/ selecting some of the spots formed in step /2/, more limited in number than the set of spots formed in said step /2/, and/4/ testing a track formation sequence for the selected spots only, and declaring a target track for a chain of spots that are selected in different images if the track formation sequence is completely satisfied for said chain of spots;wherein the number of spots selected in step /3/ is limited based on an integrated intensity value Nc calculated for a target presence probability density distribution, produced for the track formation sequences already being tested before step /3/ is executed for the image currently being processed, and based on the number of spots Nt of said distribution which have individual weights exceeding the predetermined weight threshold. 10. The system of claim 9, further comprising: a hypothetical probability density filter adapted to produce the target presence probability density distribution from a current state of track formation sequences that is established by the tracking module, and from spot trajectories corresponding to said track formation sequences, by applying kinematic criteria concerning spot movement between successive images, and adapted to search for a correlation between the target presence probability density distribution and the spots that have been formed by the image processing module, followed by updating the target presence probability density distribution by using a result of the correlation; andthe hypothetical probability density filter incorporating the spot selection module. 11. The system of claim 10, wherein the hypothetical probability density filter is adapted to calculate, from the updated target presence probability density distribution, the integrated intensity value Nc and the number of spots Nt for which the track formation sequences are currently being tested, and which have individual weights exceeding the weight threshold. 12. The system of claim 10, wherein the hypothetical probability density filter is at least in part based on a combination of Gaussian profiles.
Khosla, Deepak; Huber, David J.; Chen, Yang; VanBuer, Darrel J.; Martin, Kevin R., System for surveillance by integrating radar with a panoramic staring sensor.
Ferrier Nol H. L. (Montmorency FRX) de Ruffi de Ponteves Dominique C. N. E. M. (Versailles FRX), Target and missile angle tracking method and system for guiding missiles on to targets.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.