Image monitoring apparatus for estimating size of singleton, and method therefor

Patent metadata

Field	Value
Publication number	US-9478032-B2
Application number	US-201514682805-A
Country	US
Kind code	B2
Filing date	Apr 9, 2015
Priority date	Oct 9, 2012
Publication date	Oct 25, 2016
Grant date	Oct 25, 2016

Abstract

An image monitoring apparatus includes a foreground/background separator generating a foreground image by removing background pixels exhibiting no movement from pixels in an image and separating foreground pixels representing pixels of a moving object from the pixels; a foreground object information generator generating foreground object information containing a coordinate value of each of one or more foreground objects included in the foreground image and the number of pixels of the foreground objects; an clustering unit clustering the foreground object information into a plurality of groups; a singleton information detector determining at least one group from the plurality of groups as a group of singletons, and detecting singleton information on the singletons; and a singleton estimation calculator estimating a predetermined area calculation function using the singleton information and calculate at least one of estimates with respect to an area, a width and a height of each of the singletons.

First claim

What is claimed is:

1. An image monitoring apparatus, comprising:
a foreground/background separator configured to generate a foreground image by removing background pixels exhibiting no movement from pixels in an image and separating foreground pixels representing pixels of a moving object from the pixels in the image;
a foreground object information generator configured to generate foreground object information containing (i) a coordinate value of each of one or more foreground objects included in the foreground image and (ii) the number of pixels of the foreground objects;
a clustering unit configured to cluster the foreground object information into a plurality of groups;
a singleton information detector configured to
determine at least one group from the plurality of groups as a group of singletons and
detect singleton information on the singletons; and

a singleton estimation calculator configured to
estimate a predetermined area calculation function using the singleton information and
calculate at least one of estimates with respect to an area, a width and a height of each of the singletons.

2. The image monitoring apparatus of claim 1 , further comprising:
a denoiser configured to remove information identified as a noise from the foreground object information generated by the foreground object information generator.

3. The image monitoring apparatus of claim 2 , wherein the denoiser is configured to
generate a histogram for the number of pixels using the foreground object information, and
remove the information identified as the noise, wherein the number of pixels corresponding to the noise is less than or equal to a predetermined reference value.

4. The image monitoring apparatus of claim 3 , wherein the denoiser is configured to remove the information identified as the noise, wherein the number of pixels corresponding to the noise is within a preset range in the histogram.

5. The image monitoring apparatus of claim 2 , wherein the denoiser is configured to determine, as a noise, an object having at least one edge of upper, lower, left and right edges cropped partially out of the foreground image, based on the coordinate value contained in the foreground object information, and remove the determined object as the noise.

6. The image monitoring apparatus of claim 1 , wherein the foreground object information generator is configured to generate the foreground object information containing a coordinate of a centroid of a minimum bounding box for the foreground object or a coordinate of a point present at an edge of the minimum bounding box as the coordinate value.

7. The image monitoring apparatus of claim 1 , wherein the foreground object information generator is configured to measure the coordinate value and the number of pixels contained in the foreground object information until the foreground object information reaches a predetermined reference data amount,
wherein the coordinate value and the number of pixels are measured for a predetermined reference time.

8. The image monitoring apparatus of claim 1 , wherein the clustering unit is configured to cluster the foreground object information into two groups using at least one of (i) an aspect ratio calculated using the foreground object information (ii) a fill ratio of the foreground object information and (iii) a second moment,
wherein the two groups comprise a singleton group formed of the singletons and a plurality-of-objects group formed of multiple objects, and
wherein the clustering unit is configured to cluster the foreground object information using at least one of a k-means clustering method, a hierarchical clustering method, a non-hierarchical clustering method and a model-based method.

9. The image monitoring apparatus of claim 1 , wherein the singleton information detector is configured to determine at least one group from the plurality of groups as the group of singletons by using a value of at least one of image features including an aspect ratio calculated using the foreground object information, a fill ratio of the foreground object information and a second moment.

10. The image monitoring apparatus of claim 1 , wherein the singleton information detector is configured to
estimate a predetermined linear function for the foreground object information using a least-squares method, and
calculate an error for the foreground object computed by using the predetermined linear function to determine the group of singletons.

11. The image monitoring apparatus of claim 10 , wherein the singleton information detector is configured to
compute parameters for a y-intercept and a slope for the clustered foreground object information using the least-squares method, and
estimate the predetermined linear function proportional to y coordinates of predetermined foreground pixels.

12. The image monitoring apparatus of claim 11 , wherein the singleton information detector is configured to
measure the number of pixels of the foreground object using the estimated predetermined linear function defined by the following equation:
a=p 1 +p 2 ×y,

wherein ‘a’ is an estimated number of pixels of the foreground object, p 1 is an a-intercept of the predetermined linear function, p 2 is the slope of the predetermined linear function, and y is the y coordinates of the foreground pixels.

13. The image monitoring apparatus of claim 10 , wherein the error for the foreground object computed using the predetermined linear function is calculated using the following equation:

e
c

=

1

N
c

⁢

∑

i
=
1

N
c

⁢

{

a
i

-

(

p
1

+

p
2

×

y
i

)

}

2

,

wherein e c is the error for the foreground object in relation to the number of pixels of the foreground object, N c is the number of foreground objects in group c, a i is an actual number of pixels of a corr

Assignees

• Sk Telecom Co Ltd

Inventors

• Lee Gwang-Gook
• Lee Hee-Yul
• Kim Ki-Mun

Classifications

• G06T5/40

  using histogram techniques · CPC title

• G06T2207/30196

  Human being; Person · CPC title

• G06T7/0081

  Physics · mapped topic

• G06T7/004 Primary

  Physics · mapped topic

• G06T2207/20148

  Physics · mapped topic