VIDEOTRACKING

• US 20100278386A1
• Filed: 12/18/2007
• Published: 11/04/2010
• Est. Priority Date: 07/11/2007
• Status: Active Grant

• Current Assignee: Deltatre AG (Deltatre SpA)
• Original Assignee: Cairos Technologies Ag
• Inventors: Hoeflinger, Florian
• Granted Patent: US 8,542,874 B2
• Time in Patent Office: Days
• US Class Current: 382/103
• CPC Class Codes: G06T 2207/30221 Sports video; Sports image
• CPC Class Codes: G06T 7/292 Multi-camera tracking
• CPC Class Codes: G06V 20/40 in video content extracting overlay text G06V20/62; video retrieval G06F16/70; processing of video elementary streams in video servers H04N21/234; processing of video elementary streams in video clients H04N21/44

Abstract

A method for tracking an object in a sequence of video frames includes the following steps: creating a model with characteristic features for the object to be tracked; and performing a template matching algorithm in individual frames on the basis of the created model for determining a position of the object in the respective frame. An apparatus arrangement for performing the method includes at least one video camera (10, 12), at least one monitor (24, 26), one computer (20) and one input device (28) for an observer (22).

Claims

1. A method for tracking an object in a sequence of video frames, characterized by the following steps:

creating a model with characteristic features for the object to be tracked; and

performing a template matching algorithm in individual frames on the basis of the created model for determining a position of the object in the respective frame.

2. The method according to claim 1, characterized in that a first component of the model contains at least one geometric dimension of the object.

3. The method according to claim 2, characterized in that the first component is a geometric shape defined for the object.

4. The method according to claim 3, characterized in that the first component is determined on the basis of a background subtraction.

5. The method according to claim 3, characterized in that the first component is determined on the basis of a calibration with respect to another object with known position and known dimension.

6. The method according to claim 3, characterized in that the first component is determined on the basis of a manual annotation by an observer (22).

7. The method according to any of claims 4 to 6, characterized in that the first component is provided with a position-dependent correction function.

8. The method according to any of claims 1 or 4 to 6, characterized in that a second component of the model contains a color of the object.

9. The method according to claim 8, characterized in that the second component is a color dependent on the first component.

10. The method according to any of claims 1 or 4 to 6, characterized in that when creating the model, object regions are defined, to which certain colors are allocated.

11. The method according to claim 10, characterized in that when performing the template matching algorithm, a weighting of the object regions is included.

12. The method according to claim 10, characterized in that before performing the first template matching algorithm, a ROI, inside which the objects should be tracked, and further regions are defined.

13. The method according to claim 10, characterized in that the template matching algorithm is performed in a limited search region of the current frame, the search region being determined by means of a dynamic prediction.

14. The method according to claim 13, characterized in that when determining the search region, the positions of the object in a specific number of previous frames are considered.

15. The method according to claim 13, characterized in that when determining the search region, a maximum reach of the object relative to its last position is determined.

16. The method according to claim 1, characterized in that in dependence on the result of the template matching algorithm, an uncertainty factor is assigned to the position of the object determined by the template matching algorithm.

17. The method according to claim 16, characterized in that a correction algorithm is included in the determination of the position of the object, when the uncertainty factor exceeds an uncertainty threshold.

18. The method according to claim 17, characterized in that as a first measure, the correction algorithm continues to automatically output a position forecast by the prediction as the current object position over a certain period, until the uncertainty factor falls below the uncertainty threshold.

19. The method according to claim 18, characterized in that as a second measure, the correction algorithm continues to output a position, which was determined as an object position before exceeding the uncertainty threshold, as the current object position, until the uncertainty factor falls below the uncertainty threshold again.

20. The method according to claim 1, characterized in that an observer (22) can manually intervene, in order to correct a false allocation of positions to an object.

21. The method according to claim 20, characterized in that an overlap of two objects is detected and analyzed automatically by a merging algorithm, whereupon the observer (22) is given the opportunity to correct the result of the merging algorithm.

22. The method according to claim 17, characterized in that after a manual intervention by the observer (22), the correction algorithm checks when the uncertainty factor has exceeded the uncertainty threshold and automatically corrects the false allocation also for the period between the exceeding of the uncertainty threshold and the manual intervention.

23. The method according to claim 1, characterized in that the template matching algorithm defines colors according to a HSV color model.

24. The method according to claim 1, characterized in that each template matching algorithm is performed on a real-time basis and the positions of the object determined thereby are also stored on a real-time basis.

25. The method according to claim 1, characterized in that successively determined position data of the object are smoothed.

26. The method according to claim 1, characterized in that the determined position data of the object are transformed into coordinates of a three-dimensional reference system, preferably into world coordinates.

27. The method according to claim 1, characterized in that in the sequence of video frames a plurality of objects are tracked at the same time.

28. An apparatus arrangement for performing the method according to any of the preceding claims, comprising at least one video camera (10, 12), at least one monitor (24, 26), one computer (20) and one input device (28) for an observer (22).

29. The apparatus arrangement according to claim 28, characterized in that two video cameras (10, 12) are provided, which are arranged such that they detect two different regions (16, 18) of a ROI within which the object is to be tracked.