Object segmentation with background extraction and moving boundary techniques

• US 6,731,799 B1
• Filed: 06/01/2000
• Issued: 05/04/2004
• Est. Priority Date: 06/01/2000
• Status: Expired due to Term

• Current Assignee: Washington University In St Louis
• Original Assignee: University of Washington
• Inventors: Sun, Shijun, Kim, Yongmin
• Primary Examiner(s): Mehta, Bhavesh M.
• Assistant Examiner(s): SUKHAPHADHANA, CHRISTOPHER T
• Application Number:
  US09/587,210
• Time in Patent Office: 1,433 Days
• Field of Search: 382/173, 382/242, 382/190, 382/103, 382/131, 382/266, 348/77, 348/135, 348/142, 348/143, 348/148, 348/149, 348/152, 348/154, 348/155, 348/169, 378/901
• US Class Current: 382/173
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06T 7/149 involving deformable models, e.g. active contour models
• CPC Class Codes: G06T 7/174 involving the use of two or more images
• CPC Class Codes: G06T 7/246 using feature-based methods, e.g. the tracking of corners or segments
• CPC Class Codes: G06V 40/103 Static body considered as a whole, e.g. static pedestrian or occupant recognition

Abstract

Image data for an image frame are allocated among three groups. In one group are image data which are part of a derived motion boundary, along with image data which differ by at least a threshold amount from a corresponding point among normalized background data. In another group are image data which closely correspond to the normalized background image data. A third group includes the remaining pixels. An initial estimate for the object boundary is adjusted iteratively based on the groupings, followed by application of an active contour model.

Claims

1. A method for estimating a desired contour of an image object which is tracked over a sequence of image frames, each image frame comprising a plurality of image data, comprising the steps of:

extracting a background model of an image background from the sequence of image frames;

for a given image frame among the sequence of image frames, applying the background model to the given image frame to derive background information for the given image frame;

for the given image frame, extracting motion information corresponding to said given image frame, said motion information comprising image data at which movement occurs within said given image frame relative to either one or both of a prior image frame and a subsequent image frame;

for the given image frame, allocating the plurality of image data for said given image frame among a plurality of groups based upon the background information for the given image frame and the motion information for said given image frame, wherein a first group of the plurality of groups corresponds to image data at which motion occurs and a second group corresponds to image data correlating to the background model;

receiving one estimate of the desired contour of the image object for the given image frame, said one estimate comprising a plurality of data points; and

adjusting the estimate of the desired contour for the given image frame by progressing from data point to data point among the plurality of data points of said one estimate and updating points included in the desired contour to achieve an updated estimate of the desired contour, said adjusting comprising, for a given point;

(a) when the given point is among the first group of the plurality of groups, adjusting the estimate of the desired contour outwards at said given point; and

(b) when the given point is among the second group among the plurality of groups, adjusting the estimate of the desired contour inwards at said given point.

2. The method of claim 1, further comprising the step of:

repeating the step of adjusting in an iterative fashion, wherein the updated estimate of the desired contour for one iteration serves as the estimate which is adjusted during a next iteration.

3. The method of claim 2, further comprising after the step of repeating, the step of:

applying an active contour model to the updated estimate of the desired contour to derive a final estimate of the desired contour for the given image frame.

4. The method of claim 1, wherein for the sequence of image frames there is a corresponding plurality of image pixel arrays, wherein the image pixels of each one array of the plurality of image pixel arrays have corresponding image pixels in the other arrays of said plurality of image pixel arrays based upon relative pixel locations, each one array of the plurality of image pixel arrays comprising an array of image pixels, each one image pixel having a luminance value, and wherein the step of extracting the background model comprises:

for each one relative pixel location, compiling the luminance value corresponding to each corresponding pixel in each corresponding image pixel array for a plurality of image frames among the sequence of image frames to achieve a plurality of sets of luminance values, wherein there is one set for said each one relative pixel location; and

deriving a luminance value in the background model for said each one relative pixel location based upon a frequency distribution of luminance values at said each one relative pixel location among the plurality of image pixel arrays corresponding to the sequence of image frames.

5. The method of claim 1, wherein the step of extracting motion information comprises:

comparing the given image frame with the subsequent image frame and the prior image frame to derive a subset of motion image data for the given image frame which corresponds to motion which occurs among said prior image frame, given image frame and subsequent image frame.

6. The method of claim 1, wherein the step of allocating the plurality of image data for said given image frame among a plurality of groups, comprises allocating a given image data of the plurality of image data to the second group when the given image data corresponds within a prescribed variation with the background model as applied to the given image frame.

7. The method of claim 1, wherein the step of allocating the plurality of image data for said given image frame among a plurality of groups, comprises allocating a given image data of the plurality of image data to the first group when the given image data varies by a prescribed threshold with the background model as applied to the given image frame.

8. The method of claim 1, further comprising before the step of adjusting the estimate, the steps of:

receiving a tracked image object for the given image frame;

for the given image frame, comparing the received one estimate of the object boundary with the tracked image object to identify discontinuities in the tracked image object occurring within a periphery defined by the one estimate of the object boundary;

filling in an identified discontinuity by altering image data for image pixels in an area of discontinuity; and

assigning the altered image data to said first group.

9. A method for estimating a desired contour of an image object which is tracked over a sequence of image frames, each image frame comprising a plurality of image data, comprising the steps of:

extracting a background model of an image background from the sequence of image frames;

for a given image frame among the sequence of image frames, applying the background model to the given image frame to derive background information for the given image frame;

for the given image frame, extracting motion information corresponding to said given image frame, said motion information comprising image data at which movement occurs within said given image frame relative to either one or both of a prior image frame and a subsequent image frame;

for the given image frame, allocating the plurality of image data for said given image frame among a plurality of groups based upon the background information for the given image frame and the motion information for said given image frame;

receiving one estimate of the desired contour of the image object for the given image frame, said one estimate comprising a plurality of data points; and

adjusting the estimate of the desired contour for the given image frame by progressing from data point to data point among the plurality of data points of said one estimate and updating points included in the desired contour to achieve an updated estimate of the desired contour, said adjusting comprising, for a given point;

(a) when the given point is among a first group of the plurality of groups, adjusting the estimate of the desired contour outwards at said given point; and

(b) when the given point is among a second group among the plurality of groups, adjusting the estimate of the desired contour inwards at said given point;

wherein for the sequence of image frames there is a corresponding plurality of image pixel arrays, wherein the image pixels of each one array of the plurality of image pixel arrays have corresponding image pixels in the other arrays of said plurality of image pixel arrays based upon relative pixel locations, each one array of the plurality of image pixel arrays comprising an array of image pixels, each one image pixel having a luminance value, and wherein the step of extracting the background model comprises;

for each one relative pixel location, compiling the luminance value corresponding to each corresponding pixel in each corresponding image pixel array for a plurality of image frames among the sequence of image frames to achieve a plurality of sets of luminance values, wherein there is one set for said each one relative pixel location;

processing said each one set to achieve said background model, wherein said step of processing comprises deriving up to three model luminance values for each one set of luminance values.

10. The method of claim 9, wherein the up to three model luminance values derived for a given one set of said each one set of luminance values are derived using prominence as a criteria, and wherein luminance value is prominent when the number of pixels corresponding to the prominent luminance value is at least 5% of the number of image frames.

11. The method of claim 9, wherein the step of processing further comprises deriving an average background value for each one relative pixel location for each image frame, the step of deriving the average background value comprising:

for each image frame, testing said each one relative pixel location to determine whether it is part of the image object for the current image frame; and

when said each one relative pixel location is not part of the image object for the current frame, selecting one value of the corresponding up to three luminance values which is closest to the luminance value for said each one relative pixel location of the current image frame.

12. The method of claim 11, wherein when said each one relative pixel location is part of the image object for the current frame, rejecting each value of the corresponding up to three luminance values which are within a prescribed tolerance of the luminance value for said each one relative pixel location of the current image frame, and when there is a remaining value of the up to three luminance values for said one relative pixel location after the step of rejecting, choosing one value of the remaining values which is closest to an average background value for a plurality of neighboring pixel locations of the current image frame.

13. The method of claim 9, wherein for a case in which more than one model luminance value is achieved for a given set of luminance values, the corresponding pixel location is considered to be occluded by a foreground object over at least a portion of the sequence of image frames.

14. A method for estimating a desired contour of an image object which is tracked over a sequence of image frames, each image frame comprising a plurality of image data, comprising the steps of:

extracting a background model of an image background from the sequence of image frames;

for a given image frame among the sequence of image frames, applying the background model to the given image frame to derive background information for the given image frame;

for the given image frame, extracting motion information corresponding to said given image frame, said motion information comprising image data at which movement occurs within said given image frame relative to either one or both of a prior image frame and a subsequent image frame;

for the given image frame, allocating the plurality of image data for said given image frame among a plurality of groups based upon the background information for the given image frame and the motion information for said given image frame;

receiving one estimate of the desired contour of the image object for the given image frame, said one estimate comprising a plurality of data points; and

adjusting the estimate of the desired contour for the given image frame by progressing from data point to data point among the plurality of data points of said one estimate and updating points included in the desired contour to achieve an updated estimate of the desired contour, said adjusting comprising, for a given point;

(a) when the given point is among a first group of the plurality of groups, adjusting the estimate of the desired contour outwards at said given point; and

(b) when the given point is among a second group among the plurality of groups, adjusting the estimate of the desired contour inwards at said given point;

wherein for the sequence of image frames there is a corresponding plurality of image pixel arrays, wherein the image pixels of each one array of the plurality of image pixel arrays have corresponding image pixels in the other arrays of said plurality of image pixel arrays based upon relative pixel locations, each one array of the plurality of image pixel arrays comprising an array of image pixels, each one image pixel having a luminance value, and wherein the step of extracting the background model comprises;

for each one relative pixel location, compiling the luminance value corresponding to each corresponding pixel in each corresponding image pixel array for a plurality of image frames among the sequence of image frames to achieve a plurality of sets of luminance values, wherein there is one set for said each one relative pixel location;

processing said each one set to achieve said background model,

wherein the step of applying the background model to the given image frame comprises, for each one pixel of the given image frame, evaluating the corresponding luminance value using the background model.

15. A method for estimating a desired contour of an image object which is tracked over a sequence of image frames, each image frame comprising a plurality of image data, comprising the steps of:

extracting a background model of an image background from the sequence of image frames;

for a given image frame among the sequence of image frames, applying the background model to the given image frame to derive background information for the given image frame;

for the given image frame, extracting motion information corresponding to said given image frame, said motion information comprising image data at which movement occurs within said given image frame relative to either one or both of a prior image frame and a subsequent image frame;

for the given image frame, allocating the plurality of image data for said given image frame among a plurality of groups based upon the background information for the given image frame and the motion information for said given image frame;

receiving one estimate of the desired contour of the image object for the given image frame, said one estimate comprising a plurality of data points; and

adjusting the estimate of the desired contour for the given image frame by progressing from data point to data point among the plurality of data points of said one estimate and updating points included in the desired contour to achieve an updated estimate of the desired contour, said adjusting comprising, for a given point;

(a) when the given point is among a first group of the plurality of groups, adjusting the estimate of the desired contour outwards at said given point; and

(b) when the given point is among a second group among the plurality of groups, adjusting the estimate of the desired contour inwards at said given point;

wherein the step of extracting motion information comprises;

comparing the given image frame with the subsequent image frame and the prior image frame to derive a subset of motion image data for the given image frame which corresponds to motion which occurs among said prior image frame, given image frame and subsequent image frame;

wherein the step of allocating the plurality of image data for said given image frame among a plurality of groups, comprises allocating a given image data of the plurality of image data to the first group when the given image data is among said subset of motion image data.

16. A method for estimating a desired contour of an image object which is tracked over a sequence of image frames, each image frame comprising a plurality of image data, comprising the steps of:

extracting a background model of an image background from the sequence of image frames;

for a given image frame among the sequence of image frames, applying the background model to the given image frame to derive background information for the given image frame;

for the given image frame, extracting motion information corresponding to said given image frame, said motion information comprising image data at which movement occurs within said given image frame relative to either one or both of a prior image frame and a subsequent image frame;

for the given image frame, allocating the plurality of image data for said given image frame among a plurality of groups based upon the background information for the given image frame and the motion information for said given image frame;

receiving one estimate of the desired contour of the image object for the given image frame, said one estimate comprising a plurality of data points; and

adjusting the estimate of the desired contour for the given image frame by progressing from data point to data point among the plurality of data points of said one estimate and updating points included in the desired contour to achieve an updated estimate of the desired contour, said adjusting comprising, for a given point;

(a) when the given point is among a first group of the plurality of groups, adjusting the estimate of the desired contour outwards at said given point; and

(b) when the given point is among a second group among the plurality of groups, adjusting the estimate of the desired contour inwards at said given point;

wherein the step of allocating the plurality of image data for said given image frame among a plurality of groups, comprises;

allocating a given image data of the plurality of image data to a third group when the given image data does not correspond to the extracted motion information and varies by more than the prescribed variation and less than the prescribed threshold with the background model as applied to the given image frame.

17. An apparatus for estimating a desired contour of an image object which is tracked over a sequence of image frames, each image frame comprising a plurality of image data, comprising the steps of:

means for extracting a background model of an image background from the sequence of image frames, wherein the background model comprises a plurality of pixels, the extracting means comprising means for defining a luminance value for each pixel among a second plurality of pixels from a frequency distribution of luminance values of correspondingly located pixels among the sequence of image frames;

means for applying, for a given image frame among the sequence of image frames, the background model to the given image frame to derive background information for the given image frame;

means for extracting, for the given image frame, motion information corresponding to said given image frame, said motion information comprising image data at which movement occurs within said given image frame relative to either one or both of a prior image frame and a subsequent image frame;

means for allocating, for the given image frame, the plurality of image data for said given image frame among a plurality of groups based upon the background information for the given image frame and the motion information for said given image frame;

means for receiving one estimate of the desired contour of the image object for the given image frame, said one estimate comprising a plurality of data points; and

means for adjusting the estimate of the desired contour for the given image frame by progressing from data point to data point among the plurality of data points of said one estimate and updating points included in the desired contour to achieve an updated estimate of the desired contour, said adjusting means comprising, for a given point;

(a) means for adjusting, when the given point is among a first group of the plurality of groups, the estimate of the desired contour outwards at said given point; and

(b) means for adjusting, when the given point is among a second group among the plurality of groups, the estimate of the desired contour inwards at said given point.

18. An apparatus for estimating a desired contour of an image object which is tracked over a sequence of image frames, each image frame comprising a plurality of image data, comprising the steps of:

a first processor which extracts a background model of an image background from the sequence of image frames;

a second processor which applies, for a given image frame among the sequence of image frames, the background model to the given image frame to derive background information for the given image frame;

a third processor which extracts, for the given image frame, motion information corresponding to said given image frame, said motion information comprising image data at which movement occurs within said given image frame relative to either one or both of a prior image frame and a subsequent image frame;

a fourth processor which allocates, for the given image frame, the plurality of image data for said given image frame among a plurality of groups based upon the background information for the given image frame and the motion information for said given image frame, wherein a first group of the plurality of groups corresponds to image data at which motion occurs and a second group corresponds to image data correlating to the background model;

a fifth processor which receives one estimate of the desired contour of the image object for the given image frame, said one estimate comprising a plurality of data points; and

a sixth processor which adjusts the estimate of the desired contour for the given image frame by progressing from data point to data point among the plurality of data points of said one estimate and updating points included in the desired contour to achieve an updated estimate of the desired contour, said sixth processor;

(a) adjusting, when the given point is among the first group of the plurality of groups, the estimate of the desired contour outwards at said given point; and

(b) adjusting, when the given point is among the second group among the plurality of groups, the estimate of the desired contour inwards at said given point.

19. The apparatus of claim 18, wherein the first processor, second processor, third processor, fourth processor, fifth processor and sixth processor are the same processor.

20. The apparatus of claim 18, wherein the sixth processor adjusts the estimate in an iterative fashion, wherein the updated estimate of the desired contour for one iteration serves as the estimate which is adjusted during a next iteration.

21. The apparatus of claim 20, further comprising an active contour model applied to the updated estimate of the desired contour to derive a final estimate of the desired contour for the given image frame.

22. The apparatus of claim 18, wherein for the sequence of image frames there is a corresponding plurality of image pixel arrays, wherein the image pixels of each one array of the plurality of image pixel arrays have corresponding image pixels in the other arrays of said plurality of image pixel arrays based upon relative pixel locations, each one array of the plurality of image pixel arrays comprising an array of image pixels, each one image pixel having a luminance value, and wherein the first processor:

for each one relative pixel location, compiles the luminance value corresponding to each corresponding pixel in each corresponding image pixel array for a plurality of image frames among the sequence of image frames to achieve a plurality of sets of luminance values, wherein there is one set for said each one relative pixel location; and

derives a luminance value in the background model for said each one relative pixel location based upon a frequency distribution of luminance values at said each one relative pixel location among the plurality of image pixel arrays corresponding tp the sequence of image frames.

23. The apparatus of claim 18, wherein the third processor compares the given image frame with the subsequent image frame and the prior image frame to derive a subset of motion image data for the given image frame which corresponds to motion which occurs among said prior image frame, given image frame and subsequent image frame.

24. The apparatus of claim 18, wherein the fourth processor allocates a given image data of the plurality of image data to the second group when the given image data corresponds within a prescribed variation with the background model as applied to the given image frame.

25. The apparatus of claim 18, wherein the fourth processor allocates a given image data of the plurality of image data to the first group when the given image data varies by a prescribed threshold with the background model as applied to the given image frame.

26. The apparatus of claim 18, which receives a tracked image object for the given image frame, and further comprising, for the given image frame, a seventh processor which compares the received one estimate of the object boundary with the tracked image object to identify discontinuities in the tracked image object occurring within a periphery defined by the one estimate of the object boundary, the seventh processor filling in an identified discontinuity by altering image data for image pixels in an area of discontinuity;

and assigning the altered image data to said first group.

27. An apparatus for estimating a desired contour of an image object which is tracked over a sequence of image frames, each image frame comprising a plurality of image data, comprising the steps of:

a first processor which extracts a background model of an image background from the sequence of image frames;

a second processor which applies, for a given image frame among the sequence of image frames, the background model to the given image frame to derive background information for the given image frame;

a third processor which extracts, for the given image frame, motion information corresponding to said given image frame, said motion information comprising image data at which movement occurs within said given image frame relative to either one or both of a prior image frame and a subsequent image frame;

a fourth processor which allocates, for the given image frame, the plurality of image data for said given image frame among a plurality of groups based upon the background information for the given image frame and the motion information for said given image frame;

a fifth processor which receives one estimate of the desired contour of the image object for the given image frame, said one estimate comprising a plurality of data points; and

a sixth processor which adjusts the estimate of the desired contour for the given image frame by progressing from data point to data point among the plurality of data points of said one estimate and updating points included in the desired contour to achieve an updated estimate of the desired contour, said sixth processor;

(a) adjusting, when the given point is among a first group of the plurality of groups, the estimate of the desired contour outwards at said given point; and

(b) adjusting, when the given point is among a second group among the plurality of groups, the estimate of the desired contour inwards at said given point;

wherein for the sequence of image frames there is a corresponding plurality of image pixel arrays, wherein the image pixels of each one array of the plurality of image pixel arrays have corresponding image pixels in the other arrays of said plurality of image pixel arrays based upon relative pixel locations, each one array of the plurality of image pixel arrays comprising an array of image pixels, each one image pixel having a luminance value, and wherein the first processor;

for each one relative pixel location, compiles the luminance value corresponding to each corresponding pixel in each corresponding image pixel array for a plurality of image frames among the sequence of image frames to achieve a plurality of sets of luminance values, wherein there is one set for said each one relative pixel location; and

processes said each one set to achieve said background model;

wherein up to three model luminance values are derived for each one set of luminance values.

28. The apparatus of claim 27, wherein the up to three model luminance values derived for a given one set of said each one set of luminance values are derived using prominence as a criteria, and wherein luminance value is prominent when the number of pixels corresponding to the prominent luminance value is at least 5% of the number of image frames.

29. The apparatus of claim 27, wherein an average background value for each one relative pixel location for each image frame is derived, including:

for each image frame, testing said each one relative pixel location to determine whether it is part of the image object for the current image frame; and

when said each one relative pixel location is not part of the image object for the current frame, selecting one value of the corresponding up to three luminance values which is closest to the luminance value for said each one relative pixel location of the current image frame.

30. The apparatus of claim 29, wherein when said each one relative pixel location is part of the image object for the current frame, rejecting each value of the corresponding up to three luminance values which are within a prescribed tolerance of the luminance value for said each one relative pixel location of the current image frame, and when there is a remaining value of the up to three luminance values for said one relative pixel location after the step of rejecting, choosing one value of the remaining which is closest to an average background value for a plurality of neighboring pixel locations of the current image frame.

31. The apparatus of claim 27, wherein for a case in which more than one model luminance value is achieved for a given set of luminance values, the corresponding pixel location is considered to be occluded by a foreground object over at least a portion of the sequence of image frames.

32. An apparatus for estimating a desired contour of an image object which is tracked over a sequence of image frames, each image frame comprising a plurality of image data, comprising the steps of:

a first processor which extracts a background model of an image background from the sequence of image frames;

a second processor which applies, for a given image frame among the sequence of image frames, the background model to the given image frame to derive background information for the given image frame;

a third processor which extracts, for the given image frame, motion information corresponding to said given image frame, said motion information comprising image data at which movement occurs within said given image frame relative to either one or both of a prior image frame and a subsequent image frame;

a fourth processor which allocates, for the given image frame, the plurality of image data for said given image frame among a plurality of groups based upon the background information for the given image frame and the motion information for said given image frame;

a fifth processor which receives one estimate of the desired contour of the image object for the given image frame, said one estimate comprising a plurality of data points; and

a sixth processor which adjusts the estimate of the desired contour for the given image frame by progressing from data point to data point among the plurality of data points of said one estimate and updating points included in the desired contour to achieve an updated estimate of the desired contour, said sixth processor;

(a) adjusting, when the given point is among a first group of the plurality of groups, the estimate of the desired contour outwards at said given point; and

(b) adjusting, when the given point is among a second group among the plurality of groups, the estimate of the desired contour inwards at said given point;

wherein for the sequence of image frames there is a corresponding plurality of image pixel arrays, wherein the image pixels of each one array of the plurality of image pixel arrays have corresponding image pixels in the other arrays of said plurality of image pixel arrays based upon relative pixel locations, each one array of the plurality of image pixel arrays comprising an array of image pixels, each one image pixel having a luminance value, and wherein the first processor;

for each one relative pixel location, compiles the luminance value corresponding to each corresponding pixel in each corresponding image pixel array for a plurality of image frames among the sequence of image frames to achieve a plurality of sets of luminance values, wherein there is one set for said each one relative pixel location; and

processes said each one set to achieve said background model;

wherein the second processor, for each one pixel of the given image frame, evaluates the corresponding luminance value using the background model.

33. An apparatus for estimating a desired contour of an image object which is tracked over a sequence of image frames, each image frame comprising a plurality of image data, comprising the steps of:

a first processor which extracts a background model of an image background from the sequence of image frames;

a second processor which applies, for a given image frame among the sequence of image frames, the background model to the given image frame to derive background information for the given image frame;

a third processor which extracts, for the given image frame, motion information corresponding to said given image frame, said motion information comprising image data at which movement occurs within said given image frame relative to either one or both of a prior image frame and a subsequent image frame;

a fourth processor which allocates, for the given image frame, the plurality of image data for said given image frame among a plurality of groups based upon the background information for the given image frame and the motion information for said given image frame;

a fifth processor which receives one estimate of the desired contour of the image object for the given image frame, said one estimate comprising a plurality of data points; and

a sixth processor which adjusts the estimate of the desired contour for the given image frame by progressing from data point to data point among the plurality of data points of said one estimate and updating points included in the desired contour to achieve an updated estimate of the desired contour, said sixth processor;

(a) adjusting, when the given point is among a first group of the plurality of groups, the estimate of the desired contour outwards at said given point; and

(b) adjusting, when the given point is among a second group among the plurality of groups, the estimate of the desired contour inwards at said given point;

wherein the third processor compares the given image frame with the subsequent image frame and the prior image frame to derive a subset of motion image data for the given image frame which corresponds to motion which occurs among said prior image frame, given image frame and subsequent image frame;

wherein the fourth processor allocates a given image data of the plurality of image data to the first group when the given image data is among said subset of motion image data.

34. An apparatus for estimating a desired contour of an image object which is tracked over a sequence of image frames, each image frame comprising a plurality of image data, comprising the steps of:

a first processor which extracts a background model of an image background from the sequence of image frames;

a second processor which applies, for a given image frame among the sequence of image frames, the background model to the given image frame to derive background information for the given image frame;

a third processor which extracts, for the given image frame, motion information corresponding to said given image frame, said motion information comprising image data at which movement occurs within said given image frame relative to either one or both of a prior image frame and a subsequent image frame;

a fourth processor which allocates, for the given image frame, the plurality of image data for said given image frame among a plurality of groups based upon the background information for the given image frame and the motion information for said given image frame;

a fifth processor which receives one estimate of the desired contour of the image object for the given image frame, said one estimate comprising a plurality of data points; and

a sixth processor which adjusts the estimate of the desired contour for the given image frame by progressing from data point to data point among the plurality of data points of said one estimate and updating points included in the desired contour to achieve an updated estimate of the desired contour, said sixth processor;

(a) adjusting, when the given point is among a first group of the plurality of groups, the estimate of the desired contour outwards at said given point; and

(b) adjusting, when the given point is among a second group among the plurality of groups, the estimate of the desired contour inwards at said given point;

wherein the fourth processor allocates a given image data of the plurality of image data to a third group when the given image data does not correspond to the extracted motion information and varies by more than the prescribed variation and less than the prescribed threshold with the background model as applied to the given image frame.

35. A method for estimating a desired contour of an image object which is tracked over a sequence of image frames, each image frame comprising a plurality of image data, comprising the steps of:

extracting a background model of an image background from the sequence of image frames, wherein the background model comprises a plurality of pixels, each pixel among a second plurality of pixels having a luminance value derived from a frequency distribution of luminance values of corresponding pixels among the sequence of image frames;

for a given image frame among the sequence of image frames, applying the background model to the given image frame to derive background information for the given image frame;

for the given image frame, extracting motion information corresponding to said given image frame, said motion information comprising image data at which movement occurs within said given image frame relative to either one or both of a prior image frame and a subsequent image frame;

for the given image frame, allocating the plurality of image data for said given image frame among a plurality of groups based upon the background information for the given image frame and the motion information for said given image frame;

receiving one estimate of the desired contour of the image object for the given image frame, said one estimate comprising a plurality of data points; and

adjusting the estimate of the desired contour for the given image frame by progressing from data point to data point among the plurality of data points of said one estimate and updating points included in the desired contour to achieve an updated estimate of the desired contour, said adjusting comprising, for a given point;

(a) when the given point is among a first group of the plurality of groups, adjusting the estimate of the desired contour outwards at said given point; and

(b) when the given point is among a second group among the plurality of groups, adjusting the estimate of the desired contour inwards at said given point.