Method and system for selecting image region that facilitates blur kernel estimation

Patent metadata

Field	Value
Publication number	US-2021192251-A1
Application number	US-201816079565-A
Country	US
Kind code	A1
Filing date	Jan 8, 2018
Priority date	Jul 14, 2017
Publication date	Jun 24, 2021
Grant date	—

Abstract

The present invention discloses a method and system for selecting an image region that facilitates blur kernel estimation, in which the method includes: calculating a relative total variation value of each pixel in a blurred image to obtain a relative total variation mapping image; setting a threshold value to determine whether respective pixel in the image is a boundary pixel or not; then sampling the blurred image and its relative total variation mapping image to obtain a series of image patches; and finally counting the number of boundary pixels in each mapping image patch and selecting out an image region that facilitates blur kernel estimation. According to the method and the system, the problems of excessive dependency on operator experience, low efficiency and the like in the existing region selection methods are effectively solved. The image region that facilitates blur kernel estimation is automatically selected out. And the method and the system are suitable for the application occasion of the blur kernel estimation in an image deblurring algorithm.

First claim

1 . A method for selecting an image region that facilitates blur kernel estimation, comprising:
(1) calculating a relative total variation value RTV(p) of each pixel P in a blurred image B to obtain a relative total variation mapping image B rtv with a same size as the blurred image; (2) determining that the pixel P is a boundary pixel if its relative total variation value RTV(p) is less than a threshold value; otherwise, determining that the pixel P is a non-boundary pixel; (3) sampling the blurred image B and its relative total variation mapping image B rtv to obtain image patches B i and mapping image patches B rtv i so that an image patch set P B ={B 1 ,B 2 , . . . ,B i } and a mapping image patch set P rtv ={B rtv 1 ,B rtv 2 , . . . ,B rtv i } are respectively obtained after these image patches are cropped; and (4) counting a number of boundary pixels in each mapping image patch B rtv i , and selecting out a mapping image patch B rtv i* with a largest number of boundary pixels in the mapping image patch set P rtv ={B rtv 1 ,B rtv 2 , . . . ,B rtv i }, an image patch B i* corresponding to the mapping image patch B rtv i* being an image region that facilitates blur kernel estimation, wherein the relative total variation value RTV(p) is:
RTV( p )=|RTV x ( p )|+|RTV y ( p )|or RTV( p )=√{square root over (|RTV x ( p )| 2 +|RTV y ( p )| 2 )},

wherein RTV x (p) represents a relative total variation value of the pixel P in a horizontal direction, and RTV y (p) represents a relative total variation value of the pixel P in a vertical direction.

2 . (canceled)

3 . The method for selecting an image region that facilitates blur kernel estimation according to claim 1 , wherein the relative total variation value of the pixel P in the horizontal direction is:

R

T


V
x



(
p
)

=

∑

q
∈

R


(
p
)



g

p
,
q

·



(

∂
x


B

)

q





∑

q
∈

R


(
p
)



g

p
,
q

·

(

∂
x


B

)

q



+
ɛ

,

wherein R(p) represents a neighborhood centered on the pixel P, q represents a pixel in the neighborhood, (∂ x B) q represents a partial derivative of the pixel q in the horizontal direction, ϵ represents an infinitesimal quantity which ensures that the denominator of the above equation is not zero, and g p,q represents a weight function, the value of which is inversely proportional to a distance between the pixel q and the pixel P; and
the relative total variation value of the pixel P in the vertical direction is:

R

T


V
y



(
p
)

=

∑

q
∈

R


(
p
)



g

p
,
q

·



(

∂
y


B

)

q





∑

Assignees

• Univ Huazhong Science Tech

Inventors

• Sang Nong
• Li Lerenhan
• Yan Hao
• Gao Changxin
• Yan Luxin
• Zhang Shiwei
• Pi Zhixiong

Classifications

• G06V10/22 Primary

  by selection of a specific region containing or referencing a pattern; Locating or processing of specific regions to guide the detection or recognition · CPC title

• G06V10/50

  by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis · CPC title

• G06V10/44

  Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components · CPC title

• G06T7/10

  Segmentation; Edge detection (motion-based segmentation G06T7/215) · CPC title

• G06K9/4642

  Physics · mapped topic