SYNTHETIC APERTURE RADAR CHIP LEVEL CROSS-RANGE STREAK DETECTOR

• US 20130033394A1
• Filed: 08/05/2011
• Published: 02/07/2013
• Est. Priority Date: 08/05/2011
• Status: Active Grant

• Current Assignee: Raytheon Company (Rtx Corporation)
• Original Assignee: Raytheon Company (Rtx Corporation)
• Inventors: McCleary, Brent
• Granted Patent: US 8,686,892 B2
• Time in Patent Office: Days
• US Class Current: 342/25.A
• CPC Class Codes: G01S 13/9027 Pattern recognition for feature extraction
• CPC Class Codes: G01S 7/415 Identification of targets based on measurements of movement associated with the target

Abstract

A method of reducing cross-range streaking in a radar image includes determining a number of on-pixels in each line of at least a portion of the radar image, determining which lines have a determined number of on-pixels that exceeds a threshold number, and removing the on-pixels of lines having the determined number of on-pixels exceeding the threshold number.

Claims

1. A method of reducing cross-range streaking in a radar image, the method comprising:

determining a number of on-pixels in each line of at least a portion of the radar image;

determining which lines have a determined number of on-pixels that exceeds a threshold number; and

removing the on-pixels of lines having the determined number of on-pixels exceeding the threshold number.

2. The method of claim 1, wherein determining the number of on-pixels in each line of at least a portion of the radar image comprises:

forming a target profile histogram; and

filtering the target profile histogram.

3. The method of claim 2, wherein filtering the target profile histogram comprises passing the target profile histogram through a low pass filter.

4. The method of claim 2, wherein determining which lines have the determined number of on-pixels exceeding the threshold number comprises:

determining a largest gradient of the target profile histogram; and

determining a thickness of the largest gradient of the target profile histogram.

5. The method of claim 4, wherein removing the on-pixels of lines having the determined number of on-pixels exceeding the threshold number comprises removing the on-pixels corresponding to the largest gradient when the thickness meets a threshold streak thickness.

6. The method of claim 4, wherein determining the largest gradient of the target profile histogram comprises:

incrementally varying a pose angle through a range of 90 degrees; and

determining a maximum gradient of the target profile histogram on a projection axis and on an axis perpendicular to the projection axis at each of increments corresponding to the range of 90 degrees.

7. The method of claim 1, wherein the lines of the at least a portion of the radar image correspond to horizontal rows of the at least a portion of the radar image.

8. The method of claim 1, further comprising:

forming a modified binary cluster after removing the on-pixels of lines having the determined number of on-pixels exceeding the threshold number;

determining a best fit of a rotating bounding box around a leading edge of the modified binary cluster; and

determining a pose estimate and determining a position estimate corresponding to the determined best fit.

9. The method of claim 8, wherein determining the pose estimate comprises:

computing length and width features of the binary cluster, wherein the computed length corresponds to a direction corresponding to a peak range of histogram bins.

10. The method of claim 1, wherein removing the on-pixels of lines having the determined number of on-pixels exceeding the threshold number comprises replacing the on-pixels with pixel values determined from other regions of the radar image.

11. The method of claim 1, further comprising:

receiving an unmodified radar image;

converting pixel data of the unmodified radar image into a binary cluster;

forming target pixel clusters corresponding to the binary cluster; and

analyzing the target pixel clusters as the at least a portion of the radar image.

12. The method of claim 1, further comprising:

forming a modified image corresponding to the removed on-pixels;

computing one or more attributes of a potential target corresponding to the modified image; and

categorizing the potential target corresponding to the computed attributes using automated target recognition.

13. A method of improving radar image analysis, the method comprising:

defining regions on a synthetic aperture radar (SAR) chip to be analyzed for cross-range streaking;

thresholding a magnitude chip corresponding to the defined regions;

forming a bi-level output chip corresponding to the thresholded magnitude chip and comprising on-pixels and off-pixels;

forming target pixel clusters comprising one or more of the on-pixels and corresponding to a potential target;

determining a number of the on-pixels in each of a plurality of sections of the target pixel clusters;

forming a cross-range profile corresponding to the determined number of on-pixels of each section;

filtering the cross-range profile;

thresholding the cross-range profile by categorizing sections having the determined number of on-pixels above a threshold value as potential problem areas;

testing the potential problem areas for at least one of minimum and maximum parameters; and

excluding pixels corresponding to the potential problem areas determined to be beyond at least one of the minimum or maximum parameters.

14. A method of automated target detection, the method including comprising:

downsampling radar data;

converting the downsampled radar data to a square-root-of-magnitude format;

identifying potential target pixels corresponding to local brightness;

clustering the potential target pixels to form regions of interest;

thresholding chip pixels corresponding to the regions of interest to determine which chip pixels correspond to a target;

clustering the thresholded chip pixels to create a binary cluster;

circumscribing the binary cluster;

computing length and width features of the binary cluster;

generating a joint feature discriminator score corresponding to the computed length and width features; and

categorizing a target detection corresponding to the joint feature discriminator score and at least one of an automatic target recognition model and an automatic target recognition template.

15. The method of claim 14, wherein the potential target pixels are identified using a classical Goldstein two-parameter constant false alarm rate method.

16. The method of claim 14, wherein circumscribing the binary cluster comprises determining a best fit of a bounding box around a leading edge of the binary cluster, and

wherein the length and width features are computed corresponding to the determined best fit.

17. The method of claim 16, wherein determining the best fit of a rotating bounding box around the leading edge of the binary cluster comprises:

determining a number of pixels at each integer coordinate value on a projection axis of a target profile histogram;

determining a largest gradient of the target profile histogram corresponding to the determined number of pixels; and

removing pixels corresponding to the largest gradient if the largest gradient meets a threshold thickness value.

18. The method of claim 14, wherein the joint feature discriminator score is used to determine whether the regions of interest belong to a target population or clutter population.