Method and apparatus for video frame sequence-based object tracking

1. A method for the analysis of a sequence of captured images showing a scene for detecting and tracking of at least one moving or static object and for matching behavior patterns of the at least one moving or static object in the captured images to object behavior in predetermined scenarios, the method comprising the steps of:

capturing an at least one long term reference image representing a background of a scene, an at least one short term reference image representing the background of the scene and an at least one static object within the scene, and an at least one current image of the scene;

pre-processing the at least one current image of the scene for generating a long term difference image between the at least one long term reference image and the at least one current image of the scene, and a short term difference image between the short term reference image and the at least one current image of the scene; and

clustering the at least one moving or static object from the short term difference image and the long term difference image and generating at least one new object and at least one existing object, the step of clustering comprising:

building groups of clustered objects from at least two dynamic or static objects in accordance with the relative locations of each of the at least two dynamic or static objects, the step of building groups comprising:

measuring the distance between each of the at least two dynamic or static objects;

determining neighborhood relations between each of the at least two dynamic or static objects and in accordance with the results of the distance measurement;

clustering the at least two dynamic or static objects in accordance with the determined neighborhood relations into distinct object groups; and

adjusting the distinct object groups in order to determine the optimal spatial parameters of each of the at least two dynamic or static objects in the distinct object groups;

adjusting the parameters of each of the at least two dynamic or static objects clustered within each group; and

updating the parameters and status of each of the at least two dynamic or static objects.

2. The method as claimed in claim 1 further comprising the steps of characterizing the scene and updating the at least one short term reference image and the at least one long term reference image with background changes.

3. The method as claimed in claim 1 further comprising a step of providing at least one reference image, at least one timing parameter and at least one visual parameter.

4. The method as claimed in claim 1 further comprising a step of defining at least one region of interest.

5. The method as claimed in claim 1 further comprising the steps of:

constructing an initial short term reference image and an initial long term reference image;

providing the initial short term reference image and the initial long term reference image; and

providing timing parameters; and assigning visual parameters.

6. The method as claimed in claim 5 wherein the step of constructing comprises creating the short term reference image and the long term reference image from a captured image.

7. The method as claimed in claim 5 wherein the step of constructing comprises creating the short term reference image and the long term reference image from internally stored images.

8. The method as claimed in claim 5 wherein the step of constructing comprises creating the short term reference image and the long term reference image through a learning process utilizing a set of sequentially ordered and captured images.

9. The method as claimed in claim 1 wherein the step of pre-processing comprises the steps of:

obtaining the short term reference image;

obtaining the long term reference image;

obtaining a currently captured image;

generating a short term difference image from the short term reference image and the currently captured image;

generating a long term difference image from the long term reference image and the currently captured image.

10. The method of claim 1 wherein the step of clustering comprises: a step of predicting the motion of the at least one moving object by predictive filtering; and a step of adapting a parameter of the at least one moving object.

11. The method as claimed in claim 10 wherein the step of adapting the parameters of the at least one moving object comprises the steps of locating the center of the at least one moving object; locating an at least one boundary point constituting the boundary line of the at least one moving object; re-calculating the location of the center of the at least one moving object; and inserting the at least one moving object into an objects table.

12. The method as claimed in claim 11 further comprising the steps of adjusting an at least one spatial parameter of the at least one moving object and retrieving objects similar to the at least one moving object from a discarded object archive.

13. The method as claimed in claim 2 wherein the step of characterizing comprises the steps of: measuring the movement of the at least one moving object to determine the behavior of the at least one moving object; merging spatially overlapping objects; and generating an alarm trigger in accordance with the results of the behavior of the at least one moving object or in accordance with the spatial or visual parameters of the at least one moving object.

14. The method as claimed in claim 13 wherein the alarm trigger is generated in accordance with the texture of the moving object.

15. The method as claimed in claim 13 wherein the alarm trigger is generated in accordance with the shape of the moving object.

16. The method as claimed in claim 13 wherein the alarm trigger is generated in accordance with the velocity of the at least one moving object.

17. The method as claimed in claim 13 wherein the alarm trigger is generated in accordance with the trajectory of the at least one moving object.

18. The method as claimed in claim 1 wherein the step of updating the background comprises the steps of: updating the background draft; updating the short term reference image; and updating the long term reference image.