TRACKING SYSTEM WITH FUSED MOTION AND OBJECT DETECTION

• US 20080002856A1
• Filed: 11/21/2006
• Published: 01/03/2008
• Est. Priority Date: 06/14/2006
• Status: Active Grant

• Current Assignee: Honeywell International Inc.
• Original Assignee: Honeywell International Inc.
• Inventors: Cohen, Isaac, Ma, Yunqian, Yu, Qian
• Granted Patent: US 8,467,570 B2
• Time in Patent Office: Days
• US Class Current: 382/103
• CPC Class Codes: G06F 18/23 Clustering techniques
• CPC Class Codes: G06F 18/254 of classification results, e.g. of results related to same input data
• CPC Class Codes: G06V 10/462 Salient features, e.g. scale invariant feature transforms [SIFT]
• CPC Class Codes: G06V 10/762 using clustering, e.g. of similar faces in social networks
• CPC Class Codes: G06V 10/809 of classification results, e.g. where the classifiers operate on the same input data
• CPC Class Codes: G06V 20/54 of traffic, e.g. cars on the road, trains or boats
• CPC Class Codes: G06V 40/20 Movements or behaviour, e.g. gesture recognition recognition of facial expressions G06V40/16
• CPC Class Codes: H04N 7/18 Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast

Abstract

A detection system fusing motion detection and object detection for various applications such as tracking, identification, and so forth. In an application framework, model information may be developed and used to reduce the false alarm rate. With a background model, motion likelihood for each pixel, for instance, of a surveillance image, may be acquired. With a target model, object likelihood for each pixel of the image may also be acquired. By joining these two likelihood distributions, detection accuracy may be significantly improved over the use of just one likelihood distribution for detection in applications such as tracking.

Claims

1. A detection system comprising:

a video stream source;

a motion detection module connected to the video stream source;

an object detection module connected to the video stream source;

a motion likelihood image module connected to the motion detection module;

a model likelihood image module connected to the object detection module; and

a fusion module connected to the motion likelihood image module and the model likelihood image module.

2. The system of claim 1, further comprising a tracking module connected to the fusion module.

3. The system of claim 1, wherein:

the video stream source is for channeling images from various cameras;

the motion detection module is for subtracting background to obtain foreground of the images according to motion; and

the object detection module is for correlating features of a model with the images to detect an object of the images.

4. The system of claim 3, wherein:

the motion likelihood image module is for computing a likelihood of each pixel of the images belonging to foreground of the images; and

the model likelihood image module is for computing a likelihood of each pixel of the images belonging to an object of the images.

5. The system of claim 4, wherein the fusion module is for generating hypotheses based at least partially on the likelihood of each pixel of the images belonging to foreground of the images and the likelihood of each pixel of the images belonging to an object of the images.

6. The system of claim 4, wherein the fusion module is for generating hypotheses based at least partially on the likelihood of each pixel of the images belonging to foreground of the images or the likelihood of each pixel of the images belonging to an object of the images.

7. A method for fusing detection techniques, comprising:

acquiring images of an area of surveillance;

extracting background from the images to obtain foreground;

determining foreground according to motion of a blob;

compiling a model having features characteristic for a category of an object in the area of surveillance;

detecting an object from the images according to similarity of features of the object with the features of the model; and

fusing the motion of the blob with the detecting of the object to form a hypothesis of a target.

8. The method of claim 7, wherein a basis of the hypothesis of the target comprises:

the blob in the images; and

the object detected in the images.

9. The method of claim 7, wherein a basis of the hypothesis of the target comprises:

the blob in the images;

or

the object detected in the images.

10. The method of claim 7, further comprising tracking a target of the hypothesis.

11. A detector system comprising:

a video streamer;

a motion detector having an input connected to an output of the video streamer;

an object detector having an input connected to the output of the video streamer; and

a detection combiner having inputs connected to an output of the motion detector and to an output of the object detector, respectively.

12. The system of claim 11, further comprising an object tracker having an input connected to an output of the detection combiner.

13. The system of claim 11, wherein:

the output of the motion detector is for providing a motion likelihood image; and

the output of the object detector is for providing a model likelihood image.

14. The system of claim 13, wherein:

the video streamer is for providing at least one image comprising a plurality of pixels;

the motion likelihood image is for providing a likelihood of each pixel belonging to a foreground of the image; and

the motion likelihood image is for providing a likelihood of each pixel belonging to a target in the image.

15. The system of claim 14, wherein the detection combiner is for merging the likelihood of each pixel belonging to a foreground of the image and the likelihood of each pixel belonging to a target in the image.

16. The system of claim 15, wherein an output of the detector combiner is for generating hypotheses.

17. The system of claim 15, wherein if there is no likelihood of each pixel belonging to a foreground of the image or there is no likelihood of each pixel belonging to a target in the image, then an output of the detection combiner would comprise the likelihood of each pixel belonging to a target in the image or the likelihood of each pixel belonging to a foreground of the image, respectively.

18. The system of claim 17, wherein the output of the detection combiner is connected to a target tracking mechanism.

19. The system of claim 13, wherein the detection combiner comprises:

a compute saving for one hypothesis module connected to the outputs of the motion detector and of the object detector;

a compute saving for overlapping hypotheses module connected to the compute saving for one hypothesis; and

a find a combination of hypotheses to maximize a total saving module connected to the compute saving for overlapping hypotheses module.

20. The system of claim 14, wherein the motion likelihood image and the object likelihood image provide a basis for an object detection probability map.