Method and apparatus for performing a clean background subtraction

• US 6,411,744 B1
• Filed: 10/15/1998
• Issued: 06/25/2002
• Est. Priority Date: 10/15/1997
• Status: Expired due to Term

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Planet Electric Incorporated
• Inventors: Edwards, Jeffrey L.
• Primary Examiner(s): Au, Amelia M.
• Assistant Examiner(s): Miller, Martin
• Application Number:
  US09/174,491
• Time in Patent Office: 1,349 Days
• Field of Search: 382/195, 382/199, 382/282, 382/294, 358/464, 345/131, 345/4, 345/474, 345/473, 345/677, 345/678, 345/662, 345/651, 345/633, 345/664-5, 345/634
• US Class Current: 382/294
• CPC Class Codes: G06T 2207/20164 Salient point detection; Corner detection
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06T 7/194 involving foreground-background segmentation
• CPC Class Codes: G06V 40/103 Static body considered as a whole, e.g. static pedestrian or occupant recognition

Abstract

A background subtraction apparatus of the present invention includes a key point locator for locating key points on a known object type, a boundary point locator for locating boundary points of the known object that make up the edges of the known object, and an edge processor for processing the edges to provide a clean-edged extraction of the known object from a background image. Preferably, the key point locator includes an alignment detector for detecting alignment of an image of the known object type with a skeleton image. Still more preferably, the skeleton image is an exoskeleton image and the known object type is a human being.

Claims

1. A background subtraction apparatus comprising:

key point location means for locating key points on a known object type, wherein said key point location means includes an alignment detector for detecting alignment of an image of said known object type with a skeleton image;

repeat keypoint location means if said alignment detector fails to align said image with said skeleton image;

boundary point location means for, after locating key points on a known object type, locating boundary points of said known object that comprise edges of said known object;

repeat keypoint location means and boundary point location means if locating boundary points fails; and

edge processing means for, after locating key points on a known object type, processing said edges to provide a clean-edged extraction of said known object from a background image.

2. A background subtraction apparatus as recited in claim 1 wherein said skeleton image is an exoskeleton image.

3. A background subtraction apparatus as recited in claim 2 wherein said known object is a human being.

4. A background subtraction apparatus as recited in claim 3 wherein said key points include primary key points at extremities of an image of said human being and secondary key points near a torso of an image of said human being.

5. A background subtraction apparatus as recited in claim 1 wherein said boundary point location means includes a scanner for scanning an image of said known object dependent upon the orientation of line segments between said key points.

6. A background subtraction apparatus as recited in claim 1 wherein said edge processing means includes at least one filter.

7. A background subtraction apparatus as recited in claim 6 wherein said filter includes at least one of a Gaussian model filter and a least-square filter.

8. A background subtraction apparatus as recited in claim 6 wherein said edge processing means further includes applying a snake algorithm to said edges.

9. A background subtraction apparatus as recited in claim 8 wherein said edge processing means further includes a boundary point connector.

10. A method for subtracting a background comprising:

locating key points on a known object type, wherein locating said key points includes detecting the alignment of an image of said known object type with a skeleton image;

repeat locating key points if the alignment fails to align said image with said skeleton image;

locating, after locating key points on a known object type, boundary points of said known object that comprise edges of said known object;

repeat locating key points and locating boundary points if locating boundary points of said known object fails; and

processing, after locating key points on a known object type, said edges to provide a clean-edged extraction of said known object from a background image.

11. A method for subtracting a background as recited in claim 10 wherein said skeleton image is an exoskeleton image.

12. A method for subtracting a background as recited in claim 11 wherein said known object is a human being.

13. A method for subtracting a background as recited in claim 12 wherein said key points include primary key points at extremities of an image of said human being and secondary key points near a torso of an image of said human being.

14. A method for subtracting a background as recited in claim 10 wherein locating boundary points includes scanning an image of said known object dependent upon the orientation of line segments between said key points.

15. A method for subtracting a background as recited in claim 10 wherein processing said edges includes filtering said edges.

16. A method for subtracting a background as recited in claim 15 wherein said filtering includes at least one of a Gaussian model filtering and a least-square filtering.

17. A method for subtracting a background as recited in claim 15 wherein said processing said edges further includes applying a snake algorithm to edges.

18. A method for subtracting a background as recited in claim 17 wherein said processing said edges further includes connecting said boundary points.

19. A computer readable media including program instructions implementing the method of claim 10.

20. A method for subtracting a background as recited in claim 10, wherein said skeleton image is a stick figure image.

21. A method for subtracting a background as recited in claim 10, wherein the known object adjusts a position to match the skeleton image.

22. A method for subtracting a background as recited in claim 10, wherein secondary points are located relative to the location of said key points.