Superpixels for improved structure and terrain classification using multiple synthetic aperture radar image products

• US 10,042,048 B1
• Filed: 02/19/2015
• Issued: 08/07/2018
• Est. Priority Date: 02/20/2014
• Status: Active Grant

• Current Assignee: National Technology & Engineering Solutions Of Sandia LLC (Honeywell International Inc.)
• Original Assignee: National Technology & Engineering Solutions Of Sandia LLC (Honeywell International Inc.)
• Inventors: Moya, Mary M., Koch, Mark W., Perkins, David Nikolaus
• Primary Examiner(s): Brainard, Timothy A
• Application Number:
  US14/626,582
• Time in Patent Office: 1,265 Days
• Field of Search: 342 25
• CPC Class Codes: G01S 13/9023 combined with interferometric techniques
• CPC Class Codes: G01S 13/9027 Pattern recognition for feature extraction
• CPC Class Codes: G01S 13/9029 specially adapted for moving target detection within a single SAR image or within multiple SAR images taken at the same time
• CPC Class Codes: G01S 7/414 Discriminating targets with respect to background clutter

Abstract

Various embodiments presented herein relate to assigning labels to segments of a synthetic aperture radar (SAR) image, where the segments are based upon a speckle-reduced SAR image product. A plurality of SAR images of a scene are co-registered to form a registered stack of SAR images. A speckle-reduced SAR image product is generated based upon at least one registered SAR image in the registered stack of SAR images. The speckle-reduced SAR image product is segmented into a plurality of superpixels, and boundaries of the superpixels are applied to the at least one registered SAR image to form a segmented SAR image. A segment of the SAR image is then labeled as including or not including a feature, wherein the label is assigned based upon values of pixels in the segment.

Claims

1. A system comprising:

an antenna that is configured to receive radar signals that have reflected off of a scene;

a processor; and

memory that comprises components that are executed by the processor, the components comprising;

a registration component that is configured to register a baseline synthetic aperture radar (SAR) image with a plurality of SAR images to create a stack of co-registered SAR images, wherein the plurality of SAR images are generated based upon the radar signals received by the antenna;

an artifact reduction component that is configured to perform a speckle-reduction operation on at least one SAR image in the stack of co-registered SAR images to form an artifact-reduced SAR image product, a first number of artifacts in the artifact-reduced SAR image product is less than a second number of artifacts in the at least one SAR image;

a segmentation component that is configured to;

form a segmented SAR image product by segmenting the artifact-reduced SAR image product into a plurality of superpixels based upon values of pixels in the artifact-reduced SAR image product; and

apply boundaries of the superpixels to the at least one SAR image to segment the at least one SAR image into a plurality of segments, such that the plurality of segments respectively correspond to the plurality of superpixels in the segmented SAR image product; and

a classifier component that is configured to assign a label to a segment in the plurality of segments in the at least one SAR image based upon pixel values in the segment, the label indicative of an identity of an object or feature captured in the at least one SAR image.

2. The system of claim 1, the artifacts being speckle.

3. The system of claim 2, the artifact reduction component is configured to utilize at least one of a sub-aperture multilook (SA-ML) imaging process, a mean-over-time of coherent change detection (MCCD) imaging process, or a median-over-time of radar cross section (MRCS) imaging process to form the artifact-reduced SAR image product.

4. The system of claim 3, the artifact reduction component further configured to form the artifact-reduced SAR image product such that each pixel in the artifact-reduced SAR image product is assigned values for an n-dimensional vector, the n-dimensional vector is formed based upon the SA-ML imaging process, the MCCD imaging process, or the MRCS imaging process.

5. The system of claim 1, the segmentation component is configured to utilize one of QUICK-SHIFT or simple linear iterative clustering (SLIC) to segment the artifact-reduced SAR image product into the plurality of superpixels.

6. The system of claim 5, the artifact-reduced SAR image product is further based upon application of an interferometric height (IF-H) mapping process to the at least one SAR image.

7. The system of claim 1, further comprising a radar system that comprises the antenna, the radar system is configured to generate SAR data as the radar system passes over the scene, the at least one SAR image is based upon the SAR data.

8. The system of claim 1, the classifier component assigns the label to the segment based upon comparing the pixel values in the segment with pre-classified data.

9. The system of claim 1, the registration component is further configured to resample a first SAR image in the plurality of SAR images to cause the first SAR image to comprise a same number of pixels as a second SAR image in the plurality of SAR images.

10. A method, comprising:

receiving, at an antenna of a radar system, radar signals that have reflected off of a scene, wherein a plurality of synthetic aperture radar (SAR) images are generated based upon the radar signals received at the antenna;

registering the plurality of SAR images to form a stack of registered SAR images;

generating an artifact-reduced SAR image product based upon at least one registered SAR image from the stack of registered SAR images, the artifact-reduced SAR image product comprises fewer artifacts than the at least one registered SAR image;

segmenting the artifact-reduced SAR image product into a plurality of superpixels to form a superpixel image product;

applying boundaries of the superpixels to the at least one registered SAR image to form a segmented SAR image, the segmented SAR image comprises a plurality of segments that correspond to the superpixels of the superpixel image product; and

assigning a label to a segment in segmented SAR image based upon values of pixels in the segment, the label indicative of an identity of an object or feature captured in at least one image in the plurality of SAR images.

11. The method of claim 10, the artifacts being speckle.

12. The method of claim 10, wherein generating the artifact-reduced image product comprises performing at least one of a sub-aperture multilook (SA-ML) imaging process, a mean-over-time of coherent change detection (MCCD) imaging process, or a median-over-time of radar cross section (MRCS) imaging process to the at least one registered SAR image.

13. The method of claim 12, each pixel in the artifact-reduced SAR image product has a n-dimensional vector assigned thereto, values of the n-dimensional vector are based upon at least one of the SA-ML imaging process, the MCCD imaging process, or the MRCS imaging process.

14. The method of claim 10, wherein segmenting the artifact-reduced SAR image product into the plurality of superpixels comprises utilizing at least one of QUICK-SHIFT or simple linear iterative clustering (SLIC) to segment the artifact-reduced SAR image product into the plurality of superpixels.

15. The method of claim 10, wherein generating the artifact-reduced SAR image product comprises:

acquiring interferometric height (IF-H) data for the at least one registered SAR image; and

generating the artifact-reduced SAR image product based upon the IF-H data.

16. The method of claim 10, wherein registering the plurality of SAR images comprises resampling a first SAR image in the plurality of SAR images to cause the first SAR image to comprise a same number of pixels as a second SAR image in the plurality of SAR images.

17. A non-transitory computer-readable storage medium comprising instructions that, when executed by a processor, cause the processor to perform acts comprising:

generating a plurality of synthetic aperture radar (SAR) images based upon radar signals received by an antenna of a radar system, wherein the radar signals have been directed towards a scene and have reflected off of the scene prior to being received by the antenna;

registering the plurality of SAR images to form a stack of registered SAR images;

applying a speckle reduction algorithm to at least one registered SAR image in the stack of registered SAR images to form an artifact-reduced SAR image product, the artifact-reduced SAR image product comprises fewer artifacts than the at least one registered SAR image;

segmenting the artifact-reduced SAR image product into a plurality of superpixels to form a superpixel image product;

applying boundaries of the superpixels in the superpixel image product to the at least one registered SAR image to form a segmented SAR image that comprises a plurality of segments, the plurality of segments respectively correspond to the plurality of superpixels; and

assigning a label to a segment in the plurality of segments based upon values of pixels in the segment, the label indicative of an identity of an object or feature captured in at least one SAR image in the plurality of SAR images.

18. The computer-readable storage medium of claim 17, the speckle reduction algorithm is one of a sub-aperture multilook (SA-ML) imaging algorithm, a mean-over-time of coherent change detection (MCCD) imaging algorithm, or a median-over-time of radar cross section (MRCS) imaging algorithm.

19. The computer-readable storage medium of claim 17, the artifacts being speckle.

20. The computer-readable storage medium of claim 17, wherein registering the plurality of SAR images comprises resampling a first SAR image in the plurality of SAR images to cause the first SAR image to comprise a same number of pixels as a second SAR image in the plurality of SAR images.