METHOD FOR ABSTRACTING MULTIPLE MOVING OBJECTS

• US 20030165193A1
• Filed: 02/21/2003
• Published: 09/04/2003
• Est. Priority Date: 03/01/2002
• Status: Active Grant

• Current Assignee: Huper Laboratories Co., Ltd.
• Original Assignee: Huper Laboratories Co., Ltd.
• Inventors: CHANG, CHOU-MING, CHEN, HSIAO-PING
• Granted Patent: US 7,265,777 B2
• Time in Patent Office: Days
• US Class Current: 375/240.08
• CPC Class Codes: G06T 9/004 Predictors, e.g. intraframe, interframe coding
• CPC Class Codes: G06T 9/20 Contour coding, e.g. using detection of edges

Abstract

A motion detection method is used to abstract multiple moving objects. The motion detection method uses a first frame and a second frame to locate the moving objects appearing in a fixed scene. The method includes calculating a difference of each pixel between the first frame and the second frame, comparing each difference with a first threshold value for generating a plurality of first contours, comparing each difference with a second threshold value for generating a plurality of second contours, picking any of the second contours which overlapped with the first contours for generating corresponding moving blocks, determining whether the pixels located outside the moving blocks are background pixels, and comparing the parameter of each background pixel located inside the moving blocks with its corresponding background value for determining pixels related to the moving objects.

Claims

1. A moving object segmentation method used by a detecting system, the detecting system comprising:

a frame generator for periodically picturing a scene according to a predetermined period to generate a plurality of image frames, each image frame comprising a plurality of pixels, each pixel comprising at least a reference parameter for describing characteristics of the pixel; and

a calculating device for detecting moving objects moving inside the scene according to a first image frame and a second image frame generated from the frame generator, the first image frame being generated prior to the second image frame by an interval corresponding to the predetermined period;

the moving object segmentation method comprising;

(a)comparing the reference parameter of one pixel in the first image frame with the reference parameter of the same pixel in the second image frame to generate a corresponding difference value;

(b)comparing each difference value with a first threshold value and a second threshold value, the first threshold value being greater than the second threshold value, generating a plurality of first contours according to the pixels which are adjacent to each other and each having the difference value greater than the first threshold value, and generating a plurality of second contours according to the pixels which are adjacent to each other and each having the difference value greater than the second threshold value;

(c)selecting second contours that overlap the first contours for establishing a corresponding first moving area, the first moving area comprising every pixel associated with the selected second contours;

(d)determining whether the pixels outside the first moving area in the second image frame are registered as background pixels; and

(e)comparing the reference parameter of the background pixel in the second image frame with a corresponding background value of the background pixel to determine whether the background pixel is related to the moving object.

2. The moving object segmentation method of claim 1 wherein the difference value is generated from a comparison of at least one reference parameter of the pixel located in the first and second image frames.

3. The moving object segmentation method of claim 2 wherein the reference parameter is a brightness value, a gray level value, or a color temperature value.

4. The moving object segmentation method of claim 1 wherein step (b) further comprises:

comparing a number of pixels included in the first contour with a third threshold value so as to make the first contour with the number of pixels greater than the third threshold value reserved; and

comparing a number of pixels included in the second contour with a third threshold value so as to make the second contour with the number of pixels greater than the third threshold value reserved.

5. The moving object segmentation method of claim 1 wherein the first moving area corresponds to a moving contour, and the moving contour is a rectangle with a smallest area that surrounds the selected second contours.

6. The moving object segmentation method of claim 5 wherein step (c) further comprises:

merging a first moving contour and a second moving contour into a third moving contour if the first moving contour overlaps the second moving contour in the second image frame.

7. The moving object segmentation method of claim 6 wherein the third moving contour is a rectangle with a smallest area that surrounds the first moving contour and the second moving contour.

8. The moving object segmentation method of claim 1 wherein each pixel further comprises a background parameter and a first counter parameter, and the first counter parameter is changed when a difference value between the reference parameter and the background parameter of the pixel is less than a background threshold value.

9. The moving object segmentation method of claim 8 wherein step (d) further comprises:

comparing the reference parameter of the pixel outside the first moving area with the corresponding background parameter of the pixel to calculate the difference value;

increasing the first counter parameter by 1; and

updating the background parameter of the pixel if the difference value is less than the background threshold value.

10. The moving object segmentation method of claim 9 wherein each pixel further comprises a second counter parameter, and the second counter parameter is changed when the difference value between the reference parameter and the background parameter of the pixel is greater than the background threshold value.

11. The moving object segmentation method of claim 10 wherein step (d) further comprises:

comparing the reference parameter of the pixel outside the first moving area with the corresponding background parameter of the pixel to calculate the difference value;

increasing the second counter parameter by 1; and

updating the background parameter of the pixel if the difference value is greater than the background threshold value.

12. The moving object segmentation method of claim 11 wherein when the difference value is greater than the background threshold value and the first counter parameter is equal to an initial counter value, the reference parameter of the pixel is used to update the corresponding background parameter.

13. The moving object segmentation method of claim 12 wherein when the difference value is greater than the background threshold value and the first counter parameter is not equal to the initial counter value, the background parameter of the pixel is maintained.

14. The moving object segmentation method of claim 13 wherein the background parameter after being updated is generated by multiplying the background parameter before being updated and either the first or second counter parameters before being updated together to generate a product first and then dividing the product by the corresponding first or second counter parameters after being updated.

15. The moving object segmentation method of claim 14 wherein when either of the first or second counter parameters is greater than a background registration value, the corresponding pixel is registered as a background pixel.

16. The moving object segmentation method of claim 15 wherein if the first counter parameter is greater than the background registration value, the background parameter after being updated is assigned to be a corresponding background value of the background pixel.

17. The moving object segmentation method of claim 15 wherein if the second counter parameter is greater than the background registration value, the reference parameter of the pixel is assigned to be a corresponding background value of the background pixel.

18. The moving object segmentation method of claim 1 wherein step (e) further comprises:

among the pixels registered as background pixels outside the first moving area, excluding the pixels whose reference parameters approach the background values for generating a non-moving region related to the moving object;

among the pixels registered as background pixels inside the first moving area, excluding the pixels whose reference parameters approach the background values for generating a moving region related to the moving object;

generating a second moving area which is a rectangle with a smallest area that surrounds pixels associated with the non-moving region and the moving region.

19. The moving object segmentation method of claim 1 wherein when all the pixels associated with the moving object inside the first moving area are registered as background pixels, a status parameter is assigned to the moving object, and a movement of the moving object between the first and second image frames accordingly changes the status parameter.

20. The moving object segmentation method of claim 19 wherein when the movement of the moving object between the first and second image frames is detected, the status parameter is reset to be zero, and when the movement of the moving object between the first and second image frames is not detected, the status parameter is increased by 1.

21. The moving object segmentation method of claim 20 wherein when the status parameter is greater than a status confirmation value, the reference parameter of each pixel associated with the moving object is used to update the corresponding background value.

22. The moving object segmentation method of claim 1 further comprising:

if the first image frame comprises pixels registered as background pixels, and no first contour and second contour related to the moving objects are detected after the second image frame is compared with the first image frame, calculating a difference value associated with the reference parameter of the background pixel between the first and second image frames; and

updating the corresponding background value by the reference parameter of the background pixel in the second image frame if the difference value is less than a refresh threshold value.

23. The moving object segmentation method of claim 22 wherein the reference parameter is a brightness value.