Image anomaly detection in a target area using polarimetric sensor data

Patent metadata

Field	Value
Publication number	US-9495594-B2
Application number	US-201414296653-A
Country	US
Kind code	B2
Filing date	Jun 5, 2014
Priority date	Jul 18, 2013
Publication date	Nov 15, 2016
Grant date	Nov 15, 2016

Abstract

A methodology for detecting image anomalies in a target area for classifying objects therein, in which at least two images of the target area are obtained from a sensor representing different polarization components. The methodology can be used to classify and/or discriminate manmade objects from natural objects in a target area, for example. A data cube is constructed from the at least two images with the at least two images being aligned, such as on a pixel-wise basis. A processor computes the global covariance of the data cube and thereafter locates a test window over a portion of the data cube. The local covariance of the contents of the test window is computed and objects are classified within the test window when an image anomaly is detected in the test window. For example, an image anomaly may be determined when a matrix determinant ratio of the local covariance and the global covariance exceeds a probability ratio threshold. The window can then be moved, e.g., by one or more pixels to form a new test window in the target area, and the above steps repeated until all of the pixels in the data cube have been included in at least one test window.

First claim

We claim:

1. A computer-implemented method for detecting image anomalies in a target area for classifying objects therein, the method comprising:
a) receiving at least two images of said target area from a sensor, said at least two images representing different polarization components,
b) creating a data cube from said at least two images by aligning pixels of the at least two-images,
c) computing a global covariance of the data cube,
d) computing a local covariance of a content of a test window located over a portion of the data cube, the test window being smaller in size than the data cube, and
e) classifying at least one object in said target area, if an image anomaly is determined in the test window.

2. The method as defined in claim 1 , wherein an anomaly is determined when a ratio of matrix determinants between said local covariance and said global covariance exceeds a probability-ratio threshold.

3. The method as defined in claim 1 , further comprising:
f) moving said test window by one or more pixels to form a new test window,
g) repeating steps d) to f) until all pixels in the data cube have been included in at least one test window.

4. The method as defined in claim 1 , wherein said different polarizations are spaced approximately 90 degrees apart from each other.

5. The method as defined in claim 2 , wherein said classification comprises: applying a Bayes decision rule to a natural log of said ratio of said local covariance and said global covariance, and determining an anomaly when said ratio exceeds a natural log of said probability ratio threshold.

6. The method as defined in claim 3 , wherein said one or more pixels in said moving step is one pixel.

7. The method as defined in claim 1 , wherein a filter is used for obtaining said at least two images at different polarization angles.

8. The method as defined in claim 1 , further comprising calculating pixel intensity values of the at least two images using Stokes parameters output from the sensor.

9. The method as defined in claim 8 wherein said pixel intensity values are calculated by the formula
I θ =0.5( S 0 +S 1 cos 2θ+ S 2 sin 2θ)

where S 0 , S 1 and S 2 are Stokes parameters and I is an intensity value at polarization angle θ.

10. The method as defined in claim 1 , wherein said sensor comprises a passive polarimetric longwave infrared sensor.

11. The method as defined in claim 1 , further comprising: discriminating and classifying type of object from another in the target area.

12. The method as defined in claim 11 , wherein the objects include manmade objects and natural objects in the target area.

13. A system for detecting image anomalies in a target area for classifying objects therein, the system comprising: at least one processor configured to execute a method according to claim 1 .

14. The system as defined in claim 13 , further comprising: a sensor configured to detect radiance.

15. The system as defined in claim 14 , wherein the sensor is a passive sensor.

16. The system as defined in claim 14 , wherein the sensor is a passive polarimetric longwave infrared sensor.

17. The system as defined in claim 14 , further comprising a filter configured to filter radiance at a polarization angle.

18. The system as defined in claim 17 , wherein the filter is a rotatable polarimetric filter.

19. A non-transient computer readable medium for storing computer instructions that, when executed by at least one processor, causes the at least one processor to perform a method for detecting image anomalies in a target area for identifying objects therein, according to claim 1 .

Assignees

• Us Army Res Lab
• Us Army

Inventors

• Rosario Dalton S
• Romano Joao M
• Mccarthy James P

Classifications

• G06V10/803

  of input or preprocessed data · CPC title

• G06V20/00 Primary

  Scenes; Scene-specific elements (control of digital cameras H04N23/60) · CPC title

• G06F18/251

  of input or preprocessed data · CPC title

• G06V10/147

  Details of sensors, e.g. sensor lenses (fingerprint or palmprint sensors G06V40/13; vascular sensors G06V40/145; eye sensors G06V40/19) · CPC title

• G06K9/00624 Primary

  Physics · mapped topic