Method and apparatus for tracking an object using a continuously adapting mean shift

• US 6,394,557 B2
• Filed: 05/15/1998
• Issued: 05/28/2002
• Est. Priority Date: 05/15/1998
• Status: Expired due to Fees

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: Bradski, Gary Rost
• Primary Examiner(s): Bella, Matthew C.
• Assistant Examiner(s): CHOOBIN, BARRY
• Application Number:
  US09/079,917
• Publication Number:
  US 20020012449A1
• Time in Patent Office: 1,474 Days
• Field of Search: 353/7 382/116,207 700/259 345/427 348/169 701/207
• US Class Current: 382/103
• CPC Class Codes: G06T 2200/24 involving graphical user interfaces [GUIs]
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 7/277 involving stochastic approaches, e.g. using Kalman filters
• CPC Class Codes: G06V 40/162 using pixel segmentation or colour matching

Abstract

A tracking method is disclosed. The method of the present invention tracks a object using a probability distribution of the desired object. The method operates by first calculating a mean location of a probability distribution within a search window. Next, the search window is centered on the calculated mean location. The steps of calculating a mean location and centering the search window may be performed until convergence. The search window may then be resized. Successive iterations of calculating a mean, centering on the mean, and resizing the search window track an object represented by the probability distribution. In one embodiment, a flesh hue probability distribution is generated from an input video image. The flesh hue probability distribution is used to track a human head within the video image.

Claims

1. A method of tracking a probability distribution of an object, said method comprising:

calculating a mean location of a probability distribution within a search window, the search window used to define the probability distribution whose mean location is to be calculated;

centering said search window onto said mean location;

resizing said search window; and

repeating said calculating, resizing and centering.

2. The method as claimed in claim 1 wherein said calculating and centering are performed until said search window converges.

3. The method as claimed in claim 1 wherein resizing said search window comprising resizing said search window dependent upon data located within said search window.

4. The method as claimed in claim 1 wherein resizing said search window comprising resizing said search window to size that is proportional to the Nth root of a zeroth moment of said data wherein N is a dimension of said data located within said search window.

5. The method as claimed in claim 1 wherein calculating said mean location comprises calculating a zeroth and first moment of data within said search window.

6. The method as claimed in claim 1, said method further comprising:

determining an orientation of said probability distribution.

7. A method of tracking an object within a video image, said method comprising:

converting said video image into a probability distribution;

calculating a mean location of a probability distribution within a search window, the search window used to define the probability distribution whose mean location is to be calculated;

centering said search window onto said mean location; and

repeating said calculating and centering.

8. The method as claimed in claim 7, said method further comprising:

resizing said search window.

9. The method as claimed in claim 8 wherein resizing said search window comprising resizing said search window dependent upon data located within said search window.

10. The method as claimed in claim 8 wherein resizing said search window comprising resizing said search window to a length and a width that is proportional to the square root of a zeroth moment of said data located within said search window.

11. The method as claimed in claim 7 wherein calculating said mean location comprises calculating a zeroth and first moment of data within said search window.

12. The method as claimed in claim 7, said method further comprising determining an orientation of said probability distribution.

13. The method as claimed in claim 12 wherein determining said orientation comprises determining a major axis of said probability distribution.

14. The method as claimed in claim 7 wherein said probability distribution comprises a flesh hue probability distribution.

15. An apparatus for tracking an object, said apparatus comprising:

a video camera, said video camera capturing an image of said object;

a video digitizer, said video digitizer digitizing said image; and

a computer system, said computer system converting said image into a probability distribution, said computer system iteratively calculating a mean location of a probability distribution within a search window of said image and centering said search window onto said mean location.

16. The apparatus as claimed in claim 15 wherein calculating said mean location comprises calculating a zeroth and first moment of data within said search window.

17. The apparatus as claimed in claim 15 wherein said computer system resizes said search window dependent upon data located within said search window.

18. The apparatus as claimed in claim 17 wherein resizing said search window comprising resizing said search window to size that is proportional to the Nth root of a zeroth moment of said data wherein N is a dimension of said data located within said search window.

19. The apparatus as claimed in claim 15 wherein calculating said mean location comprises calculating a zeroth and first moment of data within said search window.

20. The apparatus as claimed in claim 15 wherein said computer determines an orientation of said probability distribution.