Method and apparatus for object tracking and segmentation via background tracking

Patent metadata

Field	Value
Publication number	US-10249046-B2
Application number	US-201515314497-A
Country	US
Kind code	B2
Filing date	May 26, 2015
Priority date	May 28, 2014
Publication date	Apr 2, 2019
Grant date	Apr 2, 2019

Abstract

A method for tracking an object commences by first establishing the object (12) in a current frame. Thereafter, a background region (202) is established encompassing the object in the current frame. The location for the object (12) is then estimated in a next frame. Next, the propagation of the background region (202) is determined. Finally, the object is segmented from its background based on propagation of the background region, thereby allowing tracking of the object from frame to frame.

First claim

The invention claimed is:

1. A method for tracking an object, comprising:
establishing the object in a current frame by circumscribing the object with a first bounding box;
establishing a background region encompassing the object in the current frame by circumscribing the object with a second bounding box larger than the first bounding box to encompass the object in at least a portion of background surrounding the object;
estimating a location for the object in a next frame;
determining propagation of the background region; and
segmenting the object from its background based on propagation of the background region.

2. The method according to claim 1 wherein determining the propagation of the background region includes detecting pixels that originally reside outside the object but inside the tracking window in the current frame but move inside a tracking window encompassing the object in the next frame.

3. The method according to claim 1 further including:
applying superpixelation to the background to yield super pixels attached to borders of the object; and
tracking such superpixels.

4. The method according to claim 1 wherein segmenting the image includes refining the segmentation via a standard color-based segmentation technique.

5. The method according to claim 1 wherein determining propagation of the background region includes tracking all background pixels using a dense point tracking method.

6. Computer program comprising program code instructions executable by a processor for implementing a method according to claim 1 .

7. Computer program product which is stored on a non-transitory computer readable medium and comprises program code instructions executable by a processor for implementing a method according to claim 1 .

8. System for tracking an object, comprising:
an image acquisition device for capturing images of an object to undergo tracking; and
a processor for processing the image captured by the image acquisition device, for establishing the object in a current frame by circumscribing the object with a first bounding box; (b) establishing a background region encompassing the object in the current frame by of circumscribing the object with a second bounding box larger than the first bounding box to encompass the object in at least a portion of background surrounding the object; (c) estimating a location for the object in a next frame; (d) determining propagation of the background region; and (e) segmenting the object from its background based on propagation of the background region.

9. The system according to claim 8 wherein the processor determines the propagation of the background region by detecting pixels that originally reside outside the object but inside the tracking window in the current frame but move inside a tracking window encompassing the object in the next frame.

10. The system according to claim 8 wherein the processor applies superpixelation to the background to yield super pixels attached to borders of the object and tracks such superpixels.

11. The system according to claim 8 wherein segmenting the image includes refining the segmentation via a standard color-based segmentation technique.

12. The system according to claim 8 wherein the processor determines propagation of the background region by tracking all background pixels using a dense point tracking method.

Assignees

• Thomson Licensing
• Interdigital Ce Patent Holdings

Inventors

• Crivelli Tomas Enrique
• Perez Rua Juan Manuel
• Perez Patrick

Classifications

• G06T7/70

  Determining position or orientation of objects or cameras (camera calibration G06T7/80) · CPC title

• G06T7/215

  Motion-based segmentation · CPC title

• G06T7/269

  using gradient-based methods · CPC title

• G06T7/194 Primary

  involving foreground-background segmentation · CPC title

• G06T7/90

  Determination of colour characteristics · CPC title