Recognition method and apparatus for false detection of an abandoned object and image processing device

1. A recognition apparatus for false detection, comprising:

a memory; and

a processor coupled to the memory and configured to:

perform foreground detection on a current frame image, so as to obtain a foreground image of the current frame image, the foreground image including an abandoned object foreground image,

cluster contours in the abandoned object foreground image into a group, and take the group as a candidate region of an abandoned object, and

compare the candidate region of the abandoned object with a reference background image, and judge whether the candidate region of the abandoned object matches with the reference background image, wherein the abandoned object is deemed a false detection upon the candidate region of the abandoned object matching the reference background image,

wherein the judging by the processor comprises:

determining a matching range of a candidate region in the reference background image corresponding to the abandoned object,

calculating a matching degree between the candidate region of the abandoned object and the matching range, determining, when the matching degree is greater than a threshold, the candidate region of the abandoned object is matched with the matching range, and the abandoned object is false detection, and

determining, when the matching degree is less than the threshold, the abandoned object is correct detection.

2. The recognition apparatus according to claim 1 , wherein the current frame image is a video monitoring image.

3. The recognition apparatus according to claim 1 , wherein the processor is configured to:

update the reference background image according to an occupancy of the foreground image, the occupancy of the foreground image being a ratio of the foreground image to a region of interest in the current frame image.

4. The recognition apparatus according to claim 3 , wherein when the occupancy of the foreground image is less than a threshold, the processor deems no abandoned object exists in the current frame image, and updates the current frame image into the reference background image, and

when the occupancy of the foreground image is greater than or equal to the threshold, the processor deems the abandoned object exists in the current frame image, and does not update the reference background image.

5. The recognition apparatus according to claim 3 , wherein the processor calculates the occupancy of the foreground image by using the following formula:

occ

=

∑

y

=

y

u

y

d

⁢

N

y

L

y

⁡

(

y

d

-

y

u

)

;

where, Ny is a number of pixels of a y-th row of foreground pixel, N y =Σ I y :fgmask(I y )≠0 1, Ly is a number of pixels of a y-th row of region of interest, L y =Σ I y :roimask(I y )≠0 1, Iy is pixels in the y-th row, fgmask is a foreground mask, and roimask is a region of interest mask.

6. The recognition apparatus according to claim 1 , wherein a size of the matching range is larger than a size of the candidate region of the abandoned object.

7. The recognition apparatus according to claim 1 , wherein the processor is further configured to:

feed back information on the false detection, and adjust a background model.

8. An image processing device, comprising:

a memory that stores a plurality of instructions; and

a processor coupled to the memory and configured to execute the plurality of instructions to:

perform foreground detection on a current frame image, so as to obtain a foreground image of the current frame image, the foreground image including an abandoned object foreground image;

cluster contours in the abandoned object foreground image into a group, and take the group as a candidate region of an abandoned object; and

compare the candidate region of the abandoned object with a reference background image, and judge whether the candidate region of the abandoned object matches with the reference background image, wherein the abandoned object is deemed a false detection upon the candidate region of the abandoned object matching the reference background image,

wherein the judging by the processor comprises:

determining a matching range of a candidate region in the reference background image corresponding to the abandoned object,

calculating a matching degree between the candidate region of the abandoned object and the matching range, determining, when the matching degree is greater than a threshold, the candidate region of the abandoned object is matched with the matching range, and the abandoned object is false detection, and

determining, when the matching degree is less than the threshold, the abandoned object is correct detection.

9. A recognition method for false detection, wherein the method comprises:

performing foreground detection on a current frame image, so as to obtain a foreground image of the current frame image, the foreground image including an abandoned object foreground image;

clustering contours in the abandoned object foreground image into a group, and take the group as a candidate region of an abandoned object; and

comparing the candidate region of the abandoned object with a reference background image, and judging whether the candidate region of the abandoned object matches with the reference background image, wherein the abandoned object is deemed a false detection upon the candidate region of the abandoned object matching the reference background image,

wherein the method further comprises:

determining a matching range of a candidate region in the reference background image corresponding to the abandoned object,

calculating a matching degree between the candidate region of the abandoned object and the matching range,

determining, when the matching degree is greater than a threshold, the candidate region of the abandoned object is matched with the matching range, and the abandoned object is false detection, and

determining, when the matching degree is less than the threshold, the abandoned object is correct detection.

10. The recognition method according to claim 9 , wherein a size of the matching range is larger than a size of the candidate region of the abandoned object.

11. The recognition method according to claim 9 , wherein the method further comprises:

feeding back information on the false detection, and adjusting a background model.

12. The recognition method according to claim 9 , wherein the current frame image is a video monitoring image.

13. The recognition method according to claim 9 , wherein the method further comprises:

updating the reference background image according to an occupancy of the foreground image, the occupancy of the foreground image being a ratio of the foreground image to a region of interest in the current frame image.

14. The recognition method according to claim 13 , wherein,

when the occupancy of the foreground image is less than a threshold, determining no abandoned object exists in the current frame image, and updating the current frame image into the reference background image, and

when the occupancy of the foreground image is greater than or equal to the threshold, determining the abandoned object exists in the current frame image where the reference background image is not updated.

15. The recognition method according to claim 13 , wherein the method comprises: calculating the occupancy of the foreground image by using the following formula:

occ

=

∑

y

=

y

u

y

d

⁢

N

y

L

y

⁡

(

y

d

-

y

u

)

;

where, Ny is a number of pixels of a y-th row of foreground pixel, N y =Σ I y :fgmask(I y )≠0 1, Ly is a number of pixels of a y-th row of region of interest,

L y =Σ I y :roimask(I y )≠0 1, Iy is pixels in the y-th row, fgmask is a foreground mask, and roimask is a region of interest mask.