DETECTION OF ABNORMAL BEHAVIOUR IN VIDEO OBJECTS

• US 20110128374A1
• Filed: 11/24/2010
• Published: 06/02/2011
• Est. Priority Date: 11/30/2009
• Status: Active Grant

• Current Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Original Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Inventors: McLeish, David Grant, Shellshear, Andrew John
• Granted Patent: US 8,619,135 B2
• Time in Patent Office: Days
• US Class Current: 348/135
• CPC Class Codes: H04N 7/18 Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast

Abstract

Disclosed herein are a method, system, and computer program product for classifying a behaviour of a detected object in a video frame, wherein the video frame includes a plurality of blocks and the detected object is associated with a subset of the plurality of blocks and a set of parameters. The method includes the step of associating a behaviour model with each of the plurality of blocks, wherein each behaviour model (500) includes a set of behaviour modes (504, 505, 506, 507, 508). The method then performs the steps, for each block in the video frame associated with the detected object, of: (i) determining a set of behaviour statistics associated with the detected object, based on the detected object and the set of parameters; and (ii) comparing the determined set of behaviour statistics with each behaviour mode in the set of behaviour modes associated with that block (704) to determine an abnormality measure associated with the block (707). The method then classifies the behaviour of the detected object based on the abnormality measures associated with the subset of the plurality of blocks associated with the detected object.

Claims

1. A method of classifying a behaviour of a detected object in a video frame, said video frame comprising a plurality of blocks and said detected object being associated with a subset of said plurality of blocks and a set of parameters, said method comprising the steps of:

(a) associating a behaviour model with each of said plurality of blocks, wherein each behaviour model includes a set of behaviour modes;

(b) for each block in said video frame associated with the detected object;

(i) determining a set of behaviour statistics associated with the detected object, based on said detected object and said set of parameters; and

(ii) comparing the determined set of behaviour statistics with each behaviour mode in the set of behaviour modes associated with that block to determine an abnormality measure associated with said block; and

is (c) classifying the behaviour of the detected object based on the abnormality measures associated with the subset of said plurality of blocks associated with said detected object.

2. The method according to claim 1, wherein said classifying step comprises the steps of:

determining an average abnormality score for all blocks in said subset of blocks associated with said detected object, based on said abnormality measures;

comparing said average abnormality score with an abnormality threshold; and

triggering an abnormality event if said average abnormality score exceeds said abnormality threshold.

3. The method according to claim 1, wherein for each block in said subset of blocks associated with the detected object, the method comprises the further step of:

updating the behaviour model associated with said block, based on said set of behaviour statistics.

4. The method according to claim 1, comprising the further step of:

creating a new behaviour mode, substantially comprising the behaviour statistics of the detected object.

5. The method according to claim 1, comprising the further step of:

comparing said set of behaviour statistics against each behaviour mode in said set of behaviour modes

6. The method according to claim 1, wherein each of said blocks is associated with a to mode limit defining a maximum number of behaviour modes in said set of behaviour modes.

7. The method according to claim 1, wherein all of said plurality blocks are of equal size.

8. The method according to claim 1, wherein said set of parameters is selected from the group consisting of speed, object size, and object stability.

9. The method according to claim 1, wherein each of said behaviour modes includes a set of object parameters, wherein each of said object parameters is associated with a mean and standard deviation.

10. A camera system for classifying a behaviour of a detected object in a video frame, said camera system comprising:

a lens system for focussing on a scene including said detected object;

a camera module coupled to said first lens system to store said video frame;

an object detection module for detecting an object in said video frame, wherein said video frame comprises a plurality of blocks and said detected object is associated with a subset of said plurality of blocks;

an object tracking module for associating a set of parameters with said detected object;

a storage device for storing a computer program; and

a processor for executing the program, said program comprising;

code for associating a behaviour model with each of said plurality of blocks,

wherein each behaviour model includes a set of behaviour modes;

code for performing the steps, for each block in said video frame associated with the detected object, of;

(i) determining a set of behaviour statistics associated with the detected object, based on said detected object and said set of parameters; and

(ii) comparing the determined set of behaviour statistics with each behaviour mode in the set of behaviour modes associated with that block to determine an abnormality measure associated with said block; and

code for classifying the behaviour of the detected object based on the abnormality measures associated with the subset of said plurality of blocks associated with said detected object.

11. A method of detecting an abnormal behaviour of a detected object in a video frame, said video frame comprising a plurality of blocks and the detected object being associated with a subset of the plurality of blocks, said method comprising the steps of:

associating at least one block in said subset of the plurality of blocks with a behaviour mode;

determining a set of behaviour statistics associated with the detected object, for each block in said subset of the plurality of blocks;

comparing, for each block in the subset of the plurality of blocks, the determined set of behaviour statistics associated with the detected object for that block with the behaviour mode associated with that block; and

detecting the abnormal behaviour of the detected object based on the comparing step.

12. A method of displaying a set of behaviour statistics associated with a detected object in a video frame, said video frame comprising a plurality of blocks, said method comprising the steps of:

determining the set of behaviour statistics associated with the detected object, the detected object being associated with a subset of the plurality of blocks; and

displaying the behaviour statistics of the detected object in the plurality of blocks associated with the detected object.

13. A method of classifying a behaviour of a detected object in a video frame, said video frame comprising a plurality of blocks, said method comprising the steps of:

associating at least one of the plurality of blocks with a first behaviour mode;

determining a behaviour statistics of the detected object, the detected object associated with a portion of the plurality of blocks;

comparing said determined behaviour statistics with a second behaviour mode;

calculating a difference between the first behaviour mode and the second behaviour mode;

merging the first behaviour mode and the second behaviour mode to produce a merged behaviour mode, based on said calculation;

comparing the determined behaviour statistics with the merged behaviour mode of the associated portion of the plurality of blocks; and

classifying the behaviour of the detected object based on the comparing step.

14. A computer readable storage medium having recorded thereon a computer program for directing a processor to execute a method of classifying a behaviour of a detected object in a video frame, said video frame comprising a plurality of blocks and said detected object being associated with a subset of said plurality of blocks and a set of parameters, said computer program comprising code for performing the steps of:

(a) associating a behaviour model with each of said plurality of blocks, wherein each behaviour model includes a set of behaviour modes;

(b) for each block in said video frame associated with the detected object;

(i) determining a set of behaviour statistics associated with the detected object, based on said detected object and said set of parameters; and

(ii) comparing the determined set of behaviour statistics with each behaviour mode in the set of behaviour modes associated with that block to determine an abnormality measure associated with said block; and

(c) classifying the behaviour of the detected object based on the abnormality measures associated with the subset of said plurality of blocks associated with said detected object.