Systems and methods for tracking with discrete texture traces

1. A method for tracking an image of an object in a plurality of video frames, comprising:

a processor determining in a first video frame in the plurality of video frames a set of discrete texture traces to a target location in a patch in the object in the first video frame to establish a reference model, the patch being generated by the set of discrete texture traces that connect to the target location from locations in a finite support region defined by a discrete trace length of a predetermined length n and a neighborhood relation scale and each discrete texture trace is a finite sequence of n quantized descriptions, wherein each quantized description includes a quantized description of a first location in the finite support region and a quantized description of a relation between the first location and a second location in the finite support region; and

the processor detecting the target location in a second video frame by maximizing a score based on the reference model.

2. The method of claim 1 , further comprising:

the processor updating the reference model based on the detected target location in the second video frame.

3. The method of claim 1 , wherein the score based on the reference model includes a relative number of discrete texture traces.

4. The method of claim 1 , further comprising:

the processor determining an estimate of the target location in the second video frame.

5. The method of claim 4 , further comprising:

the processor determining a location with a highest score in the second video frame by an iterative process as the target location in the second video frame.

6. The method of claim 4 , further comprising:

the processor determining a confidence map of a set of locations in a window around the estimate of the target location in the second video frame; and

the processor determining that a highest score in the confidence map does not meet a threshold requirement.

7. The method of claim 6 , further comprising:

the processor making the target location of a preceding video frame the target location of the second video frame; and

the processor determining a new set of discrete texture traces to the target location in the second video frame to establish a new reference model.

8. A method for image tracking in a plurality of video frames, comprising:

determining a target location in a first video frame in the plurality of video frames;

a processor extracting from an area in the first video frame with a defined size that includes the target location, a set of discrete texture traces of predetermined length in a patch from locations in the patch that are not previously detected keypoints to the target location as an active set of discrete texture traces; and

the processor computing a score map of an area in a second video frame in the plurality of video frames based on the active set of discrete texture traces.

9. The method of claim 8 , further comprising:

the processor determining as the target location in the second video frame, a location with a maximum score in the score map.

10. The method of claim 9 , wherein the maximum score meets a threshold scoring value.

11. The method of claim 8 , wherein no location in the score map meets a threshold scoring value.

12. The method of claim 11 , further comprising:

the processor determining the target location of the first video frame as the target location of the second video frame;

the processor storing the active set of discrete texture traces as a stored model; and

the processor determining a new active set of discrete texture traces related to the target location of the second video frame.

13. The method of claim 11 , further comprising:

the processor computing a score map for the second video frame for each of one or more stored models; and

the processor updating the target location of the second video frame if the computed score map meets the threshold scoring value.

14. The method of claim 8 , wherein the score map is a confidence map.

15. The method of claim 8 , wherein the score map is created by an iterative process.

16. A system to track an image of an object in a plurality of video frames, comprising:

a memory to store data, including instructions;

a processor enabled to execute instructions upon data retrieved from the memory to perform the steps:

determining in a first video frame in the plurality of video frames a set of discrete texture traces of a predetermined length to a target location from trace locations in a patch in the object in the first video frame to establish a reference model, wherein the trace locations in the patch are not previously detected keypoints; and

detecting the target location in a second video frame in the plurality of video frames by determining a score based on the reference model.

17. The system of claim 16 , further comprising:

the processor enabled to update the reference model based on the detected target location in the second video frame.

18. The system of claim 16 , wherein the score based on the reference model includes a relative number of discrete texture traces.

19. The system of claim 16 , further comprising:

the processor enabled to determine an estimate of the target location in the second video frame.

20. The system of claim 19 , further comprising:

the processor enabled to determine a scoring map of a set of locations in an area around the estimate of the target location in the second video frame; and

the processor enabled to determine a location with a highest number of discrete texture traces in the scoring map as the target location in the second video frame.