Systems and methods for object tracking

1. A method performed by an electronic device, comprising:

determining a local motion pattern by determining a set of local motion vectors within a region of interest between a previous frame and a current frame;

determining a global motion pattern by determining a set of global motion vectors within an area between the previous frame and the current frame, wherein the area is larger than the region of interest;

calculating a global motion consistency measure, the global motion consistency measure indicative of a consistency of the global motion vectors relative to an average of the set of global motion vectors;

calculating a separation metric based on the local motion pattern and the global motion pattern in a case that the global motion consistency measure meets a consistency criterion, wherein the separation metric indicates a motion difference between the local motion pattern and the global motion pattern; and

tracking an object based on the separation metric in the case that the global motion consistency measure meets the consistency criterion.

2. The method of claim 1 , further comprising:

measuring a distance error of each of the local motion vectors; and

rejecting one or more of the local motion vectors that have an amount of distance error.

3. The method of claim 1 , wherein the global motion consistency measure indicates a number of global motion vectors within a range from the average.

4. The method of claim 1 , wherein calculating the separation metric comprises:

fitting the global motion pattern to a global motion model; and

calculating fitting errors between the global motion model and the local motion pattern.

5. The method of claim 4 , wherein the global motion model comprises at least one statistical measure of the global motion pattern.

6. The method of claim 1 , further comprising calculating a foreground weight map based on the separation metric.

7. The method of claim 6 , further comprising smoothing the foreground weight map over time.

8. The method of claim 6 , further comprising refining the foreground weight map over space.

9. The method of claim 8 , wherein refining the foreground weight map over space comprises multiplying the foreground weight map by a centrally weighted kernel.

10. The method of claim 1 , further comprising smoothing a historical weight map over time.

11. An electronic device, comprising:

a processor configured to:

determine a local motion pattern by determining a set of local motion vectors within a region of interest between a previous frame and a current frame;

determine a global motion pattern by determining a set of global motion vectors within an area between the previous frame and the current frame, wherein the area is larger than the region of interests;

calculate a global motion consistency measure, the global motion consistency measure indicative of a consistency of the global motion vectors relative to an average of the set of global motion vectors;

calculate a separation metric based on the local motion pattern and the global motion pattern in a case that the global motion consistency measure meets a consistency criterion, wherein the separation metric indicates a motion difference between the local motion pattern and the global motion pattern; and

track an object based on the separation metric in the case that the global motion consistency measure meets the consistency criterion.

12. The electronic device of claim 11 , wherein the processor is configured to measure a distance error of each of the local motion vectors and reject one or more of the local motion vectors that have an amount of distance error.

13. The electronic device of claim 11 , wherein the global motion consistency measure indicates a number of global motion vectors within a range from the average.

14. The electronic device of claim 11 , wherein the processor is configured to:

fit the global motion pattern to a global motion model; and

calculate fitting errors between the global motion model and the local motion pattern.

15. The electronic device of claim 14 , wherein the global motion model comprises at least one statistical measure of the global motion pattern.

16. The electronic device of claim 11 , wherein the processor is configured to calculate a foreground weight map based on the separation metric.

17. The electronic device of claim 16 , wherein the processor is configured to smooth the foreground weight map over time.

18. The electronic device of claim 16 , wherein the processor is configured to refine the foreground weight map over space.

19. The electronic device of claim 18 , wherein the processor is configured to multiply the foreground weight map by a centrally weighted kernel.

20. The electronic device of claim 11 , wherein the processor is configured to smooth a historical weight map over time.

21. An apparatus, comprising:

means for determining a local motion pattern, wherein the means for determining the local motion pattern comprises means for determining a set of local motion vectors within a region of interest between a previous frame and a current frame;

means for determining a global motion pattern, wherein the means for determining the global motion pattern comprises means for determining a set of global motion vectors within an area between the previous frame and the current frame, wherein the area is larger than the region of interest;

means for calculating a global motion consistency measure, the global motion consistency measure indicative of a consistency of the global motion vectors relative to an average of the set of global motion vectors;

means for calculating a separation metric based on the local motion pattern and the global motion pattern in a case that the global motion consistency measure meets a consistency criterion, wherein the separation metric indicates a motion difference between the local motion pattern and the global motion pattern; and

means for tracking an object based on the separation metric in the case that the global motion consistency measure meets the consistency criterion.

22. The apparatus of claim 21 , further comprising:

means for measuring a distance error of each of the local motion vectors; and

means for rejecting one or more of the local motion vectors that have an amount of distance error.

23. The apparatus of claim 21 , wherein the global motion consistency measure indicates a number of global motion vectors within a range from the average.

24. The apparatus of claim 21 , wherein the means for calculating the separation metric comprises:

means for fitting the global motion pattern to a global motion model; and

means for calculating fitting errors between the global motion model and the local motion pattern.

25. The apparatus of claim 21 , further comprising means for calculating a foreground weight map based on the separation metric.

26. A computer-program product, comprising a non-transitory computer-readable medium having instructions thereon, the instructions comprising:

code for causing an electronic device to determine a local motion pattern by determining a set of local motion vectors within a region of interest between a previous frame and a current frame;

code for causing the electronic device to determine a global motion pattern by determining a set of global motion vectors within an area between the previous frame and the current frame, wherein the area is larger than the region of interest;

code for causing the electronic device to calculate a global motion consistency measure, the global motion consistency measure indicative of a consistency of the global motion vectors relative to an average of the set of global motion vectors;

code for causing the electronic device to calculate a separation metric based on the local motion pattern and the global motion pattern in a case that the global motion consistency measure meets a consistency criterion, wherein the separation metric indicates a motion difference between the local motion pattern and the global motion pattern; and

code for causing the electronic device to track an object based on the separation metric in the case that the global motion consistency measure meets the consistency criterion.

27. The computer-program product of claim 26 , further comprising:

code for causing the electronic device to measure a distance error of each of the local motion vectors; and

code for causing the electronic device to reject one or more of the local motion vectors that have an amount of distance error.

28. The computer-program product of claim 26 , wherein the global motion consistency measure indicates a number of global motion vectors within a range from the average.

29. The computer-program product of claim 26 , wherein the code for causing the electronic device to calculate the separation metric comprises:

code for causing the electronic device to fit the global motion pattern to a global motion model; and

code for causing the electronic device to calculate fitting errors between the global motion model and the local motion pattern.

30. The computer-program product of claim 26 , further comprising code for causing the electronic device to calculate a foreground weight map based on the separation metric.