Contraband detection system and method using variance data

• US 20050249416A1
• Filed: 05/07/2004
• Published: 11/10/2005
• Est. Priority Date: 05/07/2004
• Status: Active Grant

• Current Assignee: Morpho Detection (Safran SA)
• Original Assignee: General Electric Company
• Inventors: Avila, Ricardo Scott, Basu, Samit Kumar, Leue, William Macomber, Walter, Deborah Joy, Mc Culloch, Colin Craig
• Granted Patent: US 7,356,174 B2
• Time in Patent Office: Days
• US Class Current: 382/195
• CPC Class Codes: G01V 5/226 using tomography
• CPC Class Codes: G06T 2207/10081 Computed x-ray tomography [CT]
• CPC Class Codes: G06T 2207/30112 Baggage; Luggage; Suitcase
• CPC Class Codes: G06T 7/0004 Industrial image inspection
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/143 involving probabilistic approaches, e.g. Markov random field [MRF] modelling

Abstract

A technique is disclosed for detecting contraband by obtaining image data from a computed tomography machine and generating variance data and a variance map from the image data acquired. The method includes obtaining a mean density value and a variation value for each voxel of the image data, segmenting the voxels into discrete objects, and determining whether any of the discrete objects is contraband.

Claims

1. A method for detecting contraband in an enclosed container, comprising:

forming image data of the contents of the enclosed container;

obtaining a mean density value and a variation value for each voxel of the image data;

segmenting the voxels into discrete objects; and

determining whether any of the discrete objects is contraband.

2. The method of claim 1, wherein the forming step comprises:

acquiring projection data of the contents of the enclosed container; and

processing the projection data to form the image data.

3. The method of claim 2, wherein said acquiring step comprises acquiring the projection data with a computed tomography machine.

4. The method of claim 1, wherein the segmenting step comprises utilizing intensity values and the direction of the local normal vector for each voxel as criteria for segmenting voxels into the discrete objects.

5. The method of claim 4, wherein the segmenting step comprises using a vector connected components algorithm.

6. The method of claim 1, wherein the segmenting step comprises a partial volume deconvolution for estimating the boundaries of the discrete objects to a sub-voxel accuracy.

7. The method of claim 6, wherein the partial volume deconvolution utilizes variance data and knowledge of a system point spread function to affect voxels having only a partial volume.

8. The method of claim 1, further comprising classifying the discrete objects.

9. The method of claim 8, wherein the mean density value and the variation value for each voxel of the image data is propagated through the obtaining and segmenting steps to obtain a confidence of the decision reached in the classifying step.

10. The method of claim 1, wherein the determining step determines whether any of the discrete objects comprises an explosive.

11. The method of claim 1, further comprising displaying images based upon the image data on a display screen.

12. The method of claim 11, wherein the displayed images are color-coded to highlight areas having high variance data.

13. A method for detecting contraband in an enclosed container, comprising:

forming image data of the contents of the enclosed container by acquiring projection data of the contents of the enclosed container and processing the projection data to form the image data;

obtaining a mean density value and a variation value for each voxel of the image data;

segmenting the voxels into discrete objects, wherein the segmenting step comprises utilizing intensity values and the direction of the local normal vector for each voxel as criteria for segmenting voxels into the discrete objects; and

determining whether any of the discrete objects is contraband.

14. The method of claim 13, wherein the acquiring projection data step is accomplished with a computed tomography machine.

15. The method of claim 13, wherein the segmenting step comprises using a vector connected components algorithm.

16. The method of claim 13, wherein the segmenting step comprises a partial volume deconvolution for estimating the boundaries of the discrete objects to a sub-voxel accuracy.

17. The method of claim 16, wherein the partial volume deconvolution utilizes variance data and knowledge of a system point spread function to affect voxels having only a partial volume.

18. The method of claim 13, further comprising classifying the discrete objects.

19. The method of claim 18, wherein the mean density value and the variation value for each voxel of the image data is propagated through the obtaining and segmenting steps to obtain a confidence of the decision reached in the classifying step.

20. The method of claim 13, wherein the determining step determines whether any of the discrete objects comprises an explosive.

21. The method of claim 13, further comprising displaying images based upon the image data on a display screen.

22. The method of claim 21, wherein the displayed images are color-coded to highlight areas having high variance data.

23. A system for detecting contraband, comprising:

an imaging system for acquiring image data; and

a processor configured to;

obtain a mean density value and a variation value for each voxel of the image data; and

segment the voxels into discrete objects.

24. The system of claim 23, wherein the imaging system acquires projection data of the contents of the enclosed container and processes the projection data to form the image data.

25. The system of claim 24, wherein the imaging system comprises a computed tomography machine.

26. The system of claim 23, wherein the processor is configured to utilize intensity values and the direction of a local normal vector for each voxel as criteria for segmenting voxels into the discrete objects.

27. The system of claim 26, wherein the processor comprises a vector connected components algorithm.

28. The system of claim 23, wherein the processor is configured to perform partial volume deconvolution for estimating the boundaries of the discrete objects to a sub-voxel accuracy.

29. The system of claim 28, wherein the processor utilizes variance data and knowledge of a system point spread function to affect voxels having only a partial volume.