AUTOMATIC TARGET SELECTION FOR MULTI-TARGET OBJECT TRACKING

• US 20160140391A1
• Filed: 11/14/2014
• Published: 05/19/2016
• Est. Priority Date: 11/14/2014
• Status: Active Grant

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: PARK, MINJE, KIM, TAE-HOON
• Granted Patent: US 9,684,830 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 18/22 Matching criteria, e.g. proximity measures
• CPC Class Codes: G06T 7/20 Analysis of motion motion estimation for coding, decoding, compressing or decompressing digital video signals H04N19/43, H04N19/51
• CPC Class Codes: G06V 10/255 Detecting or recognising potential candidate objects based on visual cues, e.g. shapes
• CPC Class Codes: G06V 10/62 relating to a temporal dimension, e.g. time-based feature extraction; Pattern tracking
• CPC Class Codes: H04N 23/60 Control of cameras or camera modules
• CPC Class Codes: H04N 7/18 Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast

Abstract

Techniques related to automatic target object selection from multiple tracked objects for imaging devices are discussed. Such techniques may include generating one or more object selection metrics such as accumulated distances from frame center, accumulated velocities, and trajectory comparisons of predicted to actual trajectories for tracked objects and selecting the target object based on the object selection metric or metrics.

Claims

1. A method for providing automatic object selection from multiple tracked objects for an imaging device comprising:

tracking at least a first object and a second object across two or more image frames;

generating a first object selection metric associated with the first object and a second object selection metric associated with the second object; and

selecting the first object or the second object as a target object for tracking based at least in part on the first object selection metric and the second object selection metric.

2. The method of claim 1, wherein the first object selection metric comprises at least one of an accumulated distance from frame center for the first object across the two or more image frames, an accumulated velocity for the first object across the two or more image frames, or a trajectory comparison of a predicted trajectory to an actual trajectory for the first object across the two or more image frames.

3. The method of claim 1, wherein the first object selection metric comprises a weighted sum of an accumulated distance from frame center for the first object across the two or more image frames and an accumulated velocity for the first object across the two or more image frames.

4. The method of claim 1, wherein the first object selection metric comprises a weighted sum of an accumulated distance from frame center for the first object across the two or more image frames, an accumulated velocity for the first object across the two or more image frames, and a trajectory comparison of a predicted trajectory to an actual trajectory for the first object across the two or more image frames.

5. The method of claim 1, wherein the first object selection metric comprises a trajectory comparison of a predicted trajectory to an actual trajectory for the first object across the two or more image frames, wherein the predicted trajectory comprises a closed-loop control predicted trajectory, and wherein the trajectory comparison comprises at least one of a difference between the predicted trajectory and the actual trajectory or a dynamic time warping comparison of the predicted trajectory and the actual trajectory.

6. The method of claim 1, wherein the first object selection metric comprises a trajectory comparison of a predicted object trajectory and an actual object trajectory for the first object.

7. The method of claim 1, further comprising:

detecting multiple objects comprising at least the first object and the second object;

performing single object tracking of the target object;

performing at least one of an auto focus adjustment, an auto exposure adjustment, or an auto white balance adjustment based at least in part on the target object; and

capturing an image based on the auto focus adjustment, auto exposure adjustment, or auto white balance adjustment.

8. The method of claim 7, wherein detecting the multiple objects comprises at least one of a saliency based object detection, a face based object detection, or an image descriptor based object detection.

9. The method of claim 1, wherein tracking the first object comprises generating a filtered trajectory of the first object.

10. The method of claim 1, further comprising:

performing object recognition based on the target object.

11. A system for providing automatic object selection from multiple tracked objects for an imaging device comprising:

a memory configured to store image data; and

a central processing unit coupled to the memory, wherein the central processing unit comprises;

object tracking circuitry to track at least a first object and a second object across two or more image frames; and

automatic object selection circuitry to generate a first object selection metric associated with the first object and a second object selection metric associated with the second object and to select at least the first object or the second object as a target object for tracking based at least in part on the first object selection metric and the second object selection metric.

12. The system of claim 11, wherein the first object selection metric comprises at least one of an accumulated distance from frame center for the first object across the two or more image frames, an accumulated velocity for the first object across the two or more image frames, or a trajectory comparison of a predicted trajectory to an actual trajectory for the first object across the two or more image frames.

13. The system of claim 11, wherein the first object selection metric comprises a weighted sum of an accumulated distance from frame center for the first object across the two or more image frames and an accumulated velocity for the first object across the two or more image frames.

14. The system of claim 11, wherein the first object selection metric comprises a weighted sum of an accumulated distance from frame center for the first object across the two or more image frames, an accumulated velocity for the first object across the two or more image frames, and a trajectory comparison of a predicted trajectory to an actual trajectory for the first object across the two or more image frames.

15. The system of claim 11, wherein the first object selection metric comprises a trajectory comparison of a predicted trajectory to an actual trajectory for the first object across the two or more image frames, wherein the predicted trajectory comprises a closed-loop control predicted trajectory, and wherein the trajectory comparison comprises at least one of a difference between the predicted trajectory and the actual trajectory or a dynamic time warping comparison of the predicted trajectory and the actual trajectory.

16. The system of claim 11, further comprising:

object detection circuitry to detect multiple objects comprising at least the first object and the second object;

image capture adjustment circuitry to provide at least one of auto focus adjustment, an auto exposure adjustment, or an auto white balance adjustment based at least in part on the target object; and

image capture circuitry to capture an image based on the auto focus adjustment, auto exposure adjustment, or auto white balance adjustment,

wherein the object tracking circuitry is to perform single object tracking of the target object.

17. The system of claim 11, wherein the object tracking circuitry is to generate a filtered trajectory of the first object.

18. At least one machine readable medium comprising a plurality of instructions that in response to being executed on an imaging device, cause the imaging device to provide automatic object selection from multiple tracked objects by:

tracking at least a first object and a second object across two or more image frames;

generating a first object selection metric associated with the first object and a second object selection metric associated with the second object; and

selecting at least the first object or the second object as a target object for tracking based at least in part on the first object selection metric and the second object selection metric.

19. The machine readable medium of claim 18, wherein the first object selection metric comprises at least one of an accumulated distance from frame center for the first object across the two or more image frames, an accumulated velocity for the first object across the two or more image frames, or a trajectory comparison of a predicted trajectory to an actual trajectory for the first object across the two or more image frames.

20. The machine readable medium of claim 18, wherein the first object selection metric comprises a weighted sum of an accumulated distance from frame center for the first object across the two or more image frames and an accumulated velocity for the first object across the two or more image frames.

21. The machine readable medium of claim 18, wherein the first object selection metric comprises a weighted sum of an accumulated distance from frame center for the first object across the two or more image frames, an accumulated velocity for the first object across the two or more image frames, and a trajectory comparison of a predicted trajectory to an actual trajectory for the first object across the two or more image frames.

22. The machine readable medium of claim 18, wherein the first object selection metric comprises a trajectory comparison of a predicted trajectory to an actual trajectory for the first object across the two or more image frames, wherein the predicted trajectory comprises a closed-loop control predicted trajectory, and wherein the trajectory comparison comprises at least one of a difference between the predicted trajectory and the actual trajectory or a dynamic time warping comparison of the predicted trajectory and the actual trajectory.

23. The machine readable medium of claim 18 further comprising instructions that cause the computing device to provide automatic object selection from multiple tracked objects by:

detecting multiple objects comprising at least the first object and the second object;

performing single object tracking of the target object;

performing at least one of an auto focus adjustment, an auto exposure adjustment, or an auto white balance adjustment based at least in part on the target object; and

capturing an image based on the auto focus adjustment, auto exposure adjustment, or auto white balance adjustment.

24. The machine readable medium of claim 18, wherein tracking the first object comprises generating a filtered trajectory of the first object.