Methods and systems for model based automatic target recognition in SAR data

• US 10,732,277 B2
• Filed: 04/29/2016
• Issued: 08/04/2020
• Est. Priority Date: 04/29/2016
• Status: Active Grant

• Current Assignee: The Boeing Co.
• Original Assignee: The Boeing Co.
• Inventors: Korchev, Dmitriy V., Owechko, Yuri, Curry, Mark A.
• Primary Examiner(s): Magloire, Vladimir
• Assistant Examiner(s): Braswell, Donald H B
• Application Number:
  US15/141,905
• Publication Number:
  US 20170350974A1
• Time in Patent Office: 1,558 Days
• Field of Search: None
• CPC Class Codes: G01S 13/9027 Pattern recognition for feature extraction
• CPC Class Codes: G01S 7/412 based on a comparison between measured values and known or stored values

Abstract

A method for automatic target recognition in synthetic aperture radar (SAR) data, comprising: capturing a real SAR image of a potential target at a real aspect angle and a real grazing angle; generating a synthetic SAR image of the potential target by inputting, from a potential target database, at least one three-dimensional potential target model at the real aspect angle and the real grazing angle into a SAR regression renderer; and, classifying the potential target with a target label by comparing at least a portion of the synthetic SAR image with a corresponding portion of the real SAR image using a processor.

Claims

1. A method for automatic target recognition in synthetic aperture radar (SAR) data, comprising:

capturing a real SAR image of a potential target at a real aspect angle and a real grazing angle;

generating a synthetic SAR image of the potential target by inputting, from a potential target database, at least one three-dimensional potential target model at the real aspect angle and the real grazing angle into a SAR regression renderer; and

,

classifying the potential target with a target label by comparing only a far edge of at least one shadow area of the synthetic SAR image with a corresponding far edge of at least one shadow area of the real SAR image using a processor.

2. A method according to claim 1, further comprising segmenting the synthetic SAR image and the real SAR image before classifying the potential target with a target label.

3. A method according to claim 2, wherein:

the synthetic SAR image is segmented into the at least one shadow area of the synthetic SAR image, and

the real SAR image is segmented into the at least one shadow area of the real SAR image.

4. A method according to claim 3, further comprising:

extracting the far edge of the at least one shadow area of the synthetic SAR image, and

extracting the far edge of the at least one shadow area of the real SAR image.

5. A method according to claim 3, wherein:

the synthetic SAR image is segmented into at least one bright area and the at least one shadow area, and

the real SAR image is segmented into at least one bright area and the at least one shadow area.

6. A method according to claim 1, wherein the classifying the potential target includes:

comparing, at the same aspect angles and grazing angles, a plurality of models from the potential target database to the real SAR image, and

choosing a best match from the plurality of models.

7. A method according to claim 1, wherein geolocation is used in addition to the real aspect angle and the real grazing angle in the capturing and generating.

8. A method according to claim 1, wherein the at least one three-dimensional potential target model also includes material information.

9. A method according to claim 1, wherein:

the generating the synthetic SAR image and the classifying the potential target are performed in parallel, and

the generating the synthetic SAR image is performed by a plurality of SAR regression renderers and/or the classifying the potential target is performed by processors.

10. A method according to claim 1, wherein the at least one three-dimensional potential target model is a model of at least one of a tank, an armored car, a car, a truck, an artillery piece, a vehicle, a boat, and/or combinations thereof.

11. A method according to claim 1, wherein at least one of the capturing the real SAR image, the generating the synthetic SAR image and/or the classifying the potential target are repeated for target label verification.

12. A method for automatic target recognition in maritime-derived synthetic aperture radar (SAR) data, comprising:

capturing a real cross-range projection SAR image of a potential maritime target at a real grazing angle and a real aspect angle;

generating a first synthetic cross-range projection SAR image of the potential maritime target by inputting, from a potential target database, at least one three-dimensional potential target model at the real grazing angle and the real aspect angle into a SAR regression renderer;

generating a second synthetic cross-range projection SAR image of the potential maritime target by inputting, from a potential target database, the at least one three-dimensional potential target model at the real grazing angle and a second aspect angle into the SAR regression renderer; and

,

classifying the potential maritime target with a target label by comparing only a far edge of at least one shadow area of the first synthetic cross-range projection SAR image and the second synthetic cross-range projection SAR image with a corresponding far edge of at least one shadow area of the real cross-range projection SAR image using a processor.

13. A method according to claim 12, further comprising segmenting the real cross-range projection SAR image with a segmentation module.

14. A method according to claim 13, further comprising extracting a binary mask from the segmented real cross-range projection SAR image.

15. A method according to claim 14, wherein the segmentation module determines the real aspect angle and the second aspect angle from the real cross-range projection SAR image to input into the SAR regression renderer.

16. A method according to claim 12, wherein the first synthetic cross-range projection SAR image and the second synthetic cross-range projection SAR image are compared with the real cross-range projection SAR image to adjust for ambiguity of target aspect angle estimation in maritime-derived SAR data.

17. A method according to claim 12, wherein the target label includes a specific maritime target ship class.

18. A method for automatic target recognition in synthetic aperture radar (SAR) data, comprising:

capturing a real cross-range projection SAR image of a potential target at a real grazing angle and a real aspect angle;

generating a first synthetic cross-range projection SAR image of the potential target by inputting, from a potential target database, at least one three-dimensional potential target model at the real grazing angle and the real aspect angle into a SAR regression renderer;

generating a second synthetic cross-range projection SAR image of the potential target by inputting, from a potential target database, the at least one three-dimensional potential target model at the real grazing angle and a second aspect angle into the SAR regression renderer; and

,

classifying the potential target with a target label by comparing only a far edge of at least one shadow area of the first synthetic cross-range projection SAR image and the second synthetic cross-range projection SAR image with a corresponding far edge of at least one shadow area of the real cross-range projection SAR image using a processor.

19. A method according to claim 18, wherein at least one of the capturing the real cross-range projection SAR image, the generating a first projection, the generating a second projection and the classifying the potential target are repeated for target label verification.

20. A system for automatic target recognition in synthetic aperture radar (SAR) data, comprising:

a SAR configured to generate a real SAR image of a potential target at a real aspect angle and a real grazing angle;

a database containing at least one three-dimensional model of a potential target;

a SAR regression renderer configured to generate a synthetic SAR image using the at least one three-dimensional model at the real aspect angle and the real grazing angle; and

,

a processor configured to compare only a far edge of at least one shadow area of the synthetic SAR image with a corresponding far edge of at least one shadow area of the real SAR image to classify the potential target with a target label.

21. A system according to claim 20, further comprising a segmentation module configured to segment at least a portion of at least one of the synthetic SAR image and/or the real SAR image.

22. A system according to claim 21, wherein the segmentation module is configured to:

segment at least a portion of the synthetic SAR image into the at least one shadow area of the synthetic SAR image; and

segment at least a portion of the real SAR image into the at least one shadow area of the real SAR image.

23. A system according to claim 22, further comprising a module for:

extracting the far edge of the shadow area for the synthetic SAR image; and

extracting the far edge of the shadow area for the real SAR image.

24. A system according to claim 22, wherein the segmentation module is further configured to:

segment at least a portion of the synthetic SAR image into a bright area of the synthetic SAR image; and

segment at least a portion of the real SAR image into a bright area of the real SAR image.

25. A system according to claim 20, wherein:

the SAR regression renderer is configured to generate a plurality of synthetic SAR images using a plurality of three-dimensional model, and

the plurality of synthetic SAR image are compared to the real SAR image.

26. A system according to claim 25, further comprising a best match analysis module configured to analyze the comparison of the plurality of synthetic SAR images to the real SAR image to identify one of the plurality of synthetic SAR images as a best match to the real SAR image.

27. A system according to claim 20, wherein the SAR is configured to communicate geolocation to the system.

28. A system according to claim 20, wherein the at least one three-dimensional model of the potential target includes material information.

29. A system for automatic target recognition in synthetic aperture radar (SAR) data, comprising:

a platform;

a SAR mounted on the platform and configured to generate a real SAR image of a potential target at a real aspect angle and a real grazing angle;

a database containing at least one three dimensional model of a potential target;

a SAR regression renderer configured to generate a synthetic SAR image using the at least one three dimensional model at the real aspect angle and the real grazing angle; and

,

a processor configured to compare only a far edge of at least one shadow area of the synthetic SAR image with a corresponding far edge of at least one shadow area of the real SAR image to classify the potential target with a target label.

30. A system according to claim 29, wherein the platform is selected from the group consisting of:

a manned aircraft;

an unmanned aircraft;

a manned spacecraft;

an unmanned spacecraft;

a manned rotorcraft;

an unmanned rotorcraft;

an ordnance; and

/or combinations thereof.