Method and device for correction restoration and analysis alarming of distorted image

1. A method for correction restoration and analysis alarming of a distorted image, comprising:

receiving an original distorted image acquired by a distortion lens, and acquiring original distorted coordinates (x d ′, y d ′) of each coordinate point in the original distorted image; and

acquiring pre-stored distortion parameters (k 1 , k 2 ) of the distortion lens, and determining restored coordinates (x u ′, y u ′) of the each coordinate point in the original distorted image to obtain a restored image;

wherein after the restored coordinates (x u ′, y u ′) of each coordinate point in the original distorted image are determined, the method further comprises:

acquiring direction angles (α, β, γ) of the distortion lens; and

performing spatial position changing on a distorted matrix to obtain restored coordinates (x u ″, y u ″) corresponding to the direction angles (α, β, γ) of the distortion lens according to the direction angles (α, β, γ) of the distortion lens and the restored coordinates (x u ′, y u ′).

2. The method for correction restoration and analysis alarming of the distorted image according to claim 1 , wherein calibrating the distortion parameters (k 1 , k 2 ) of the distortion lens comprises:

receiving a sample image acquired by the distortion lens and a sample image acquired by an ordinary lens;

acquiring distorted coordinates (x d , y d ) of a calibration characteristic in the sample image acquired by the distortion lens and undistorted coordinates (x u , y u ) of a calibration characteristic in the sample image acquired by the ordinary lens; and

substituting the distorted coordinates (x d , y d ) and the undistorted coordinates (x u , y u ) into formula:

x

di

=

x

u

1

+

k

1

⁢

r

i

-

1

2

+

k

2

⁢

r

i

-

1

4

,

⁢

y

di

=

y

u

1

+

k

1

⁢

r

i

-

1

2

+

k

2

⁢

r

i

-

1

4

,

and performing iteration for n times to obtain the distortion parameters (k 1 , k 2 ), wherein (x di , y di ) are distorted coordinates after ith iteration, r i-1 is a radius of the sample image after (i−1)th iteration, and an initial value is r 0 =√{square root over (x u 2 +y u 2 )}.

3. The method for correction restoration and analysis alarming of the distorted image according to claim 2 , wherein acquiring the distorted coordinates (x d , y d ) of the calibration characteristic in the sample image acquired by the distortion lens and the undistorted coordinates (x u , y u ) of the calibration characteristic in the sample image acquired by the ordinary lens comprises:

performing second-order image graying processing on the sample image acquired by the distortion lens and the sample image acquired by the ordinary lens to obtain a distorted sample grayscale image and an ordinary sample grayscale image respectively;

performing edge detection on the distorted sample grayscale image and the ordinary sample grayscale image to obtain a distorted sample profile diagram and an ordinary sample profile diagram by adopting an edge detection algorithm respectively; and

detecting the calibration characteristics in the distorted sample profile diagram and the ordinary sample profile diagram to obtain the distorted coordinates (x d , y d ) and the undistorted coordinates (x u , y u ) by adopting a Hough transform algorithm respectively.

4. The method for correction restoration and analysis alarming of the distorted image according to claim 1 , wherein the determining the restored coordinates (x u ′, y u ′) of each coordinate point in the original distorted image comprises:

substituting the original distorted coordinates (x d ′, y d ′) and the distortion parameters (k 1 , k 2 ) into formula:

x

di

′

=

x

u

′

1

+

k

1

⁢

r

i

-

1

′2

+

k

2

⁢

r

i

-

1

′4

,

⁢

y

di

′

=

y

u

1

+

k

1

⁢

r

i

-

1

′2

+

k

2

⁢

r

i

-

1

′4

,

and performing iteration for n′ times to obtain the restored coordinates (x u ′, y u ′), wherein (x di ′, y di ′) are original distorted coordinates after ith iteration, r i-1 ′ is a radius of the original distorted image after (i−1)th iteration, and an initial value is r 0 ′=√{square root over (x u ′2 +y u ′2 )}.

5. The method for correction restoration and analysis alarming of the distorted image according to claim 1 , wherein the performing the spatial position changing on the distorted matrix to obtain the restored coordinates (x u ″, y u ″) corresponding to the direction angles (α, β, γ) of the distortion lens according to the direction angles (α, β, γ) of the distortion lens and the restored coordinates (x u ′, y u ′) comprises:

substituting the restored coordinates (x u ′, y u ′) into formula:

[

x

u

′

y

u

′

]

=

f

z

c

⁡

[

x

c

y

c

⁢

]

to obtain lens coordinates (x c , y c , z c ), wherein f is an equivalent focal distance of the distortion lens; and

substituting the direction angles (α, β, γ) and lens coordinates (x c , y c , z c ) of the distortion lens into formula:

[

x

c

y

c

⁢

z

c

]

=

[

α

β

λ

]

⁡

[

k

1

k

2

k

3

k

4

k

5

k

6

k

7

k

8

k

9

]

⁡

[

x

u

″

y

u

″

z

u

″

]

+

[

t

x

t

y

t

z

]

to obtain the restored coordinates (x u ″, y u ″) corresponding to the direction angles (α, β, γ) of the distortion lens, wherein

[

k

1

k

2

k

3

k

4

k

5

k

6

k

7

k

8

k

9

]

is an optical characteristic parameter matrix of the distortion lens, and

[

t

x

t

y

t

z

]

is a translation vector.

6. The method for correction restoration and analysis alarming of the distorted image according to claim 1 , wherein after the pre-stored distortion parameters (k 1 , k 2 ) of the distortion lens are acquired and the restored coordinates (x u ′, y u ′) of each coordinate point in the original distorted image are determined to obtain the restored image, the method further comprises:

repairing hole coordinate points in the restored image by adopting an interpolation algorithm; and

restoring pixels of the original distorted coordinates to the restored coordinates according to corresponding relationships between the original distorted coordinates of the original distorted image and the restored coordinates of the restored image.

7. A device for correction restoration and analysis alarming of a distorted image, comprising: a hardware processor, configured to perform programming components stored in a memory, wherein the programming components comprise:

an acquisition component configured to receive an original distorted image acquired by a distortion lens, and acquire original distorted coordinates (x d ′, y d ′) of each coordinate point in the original distorted image; and

a correction component configured to acquire pre-stored distortion parameters (k 1 , k 2 ) of the distortion lens, and determine restored coordinates (x u ′, y u ′) of the each coordinate point in the original distorted image to obtain a restored image;

wherein the acquisition component is further configured to acquire direction angles (α, β, γ) of the distortion lens; and

the correction component is further configured to perform spatial position changing on a distorted matrix to obtain restored coordinates (x u ″, y u ″) corresponding to the direction angles (α,β, γ) of the distortion lens according to the direction angles (α, β, γ) of the distortion lens and the restored coordinates (x u ′, y u ′).

8. The device for correction restoration and analysis alarming of the distorted image according to claim 7 , further comprising a parameter calibration component configured to

receive a sample image acquired by the distortion lens and a sample image acquired by an ordinary lens;

acquire distorted coordinates (x d , y d ) of a calibration characteristic in the sample image acquired by the distortion lens and undistorted coordinates (x u , y u ) of a calibration characteristic in the sample image acquired by the ordinary lens; and

substitute the distorted coordinates (x d , y d ) and the undistorted coordinates (x u , y u ) into formula:

x

di

=

x

u

1

+

k

1

⁢

r

i

-

1

2

+

k

2

⁢

r

i

-

1

4

,

⁢

y

di

=

y

u

1

+

k

1

⁢

r

i

-

1

2

+

k

2

⁢

r

i

-

1

4

,

and perform iteration for n times to obtain the distortion parameters (k 1 , k 2 ), wherein (x di , y di ) are distorted coordinates after ith iteration, r i-1 is a radius of the sample image after (i−1)th iteration, and an initial value is r 0 =√{square root over (x u 2 +y u 2 )}.

9. The device for correction restoration and analysis alarming of the distorted image according to claim 8 , wherein the parameter calibration component is further configured to:

perform second-order image graying processing on the sample image acquired by the distortion lens and the sample image acquired by the ordinary lens to obtain a distorted sample grayscale image and an ordinary sample grayscale image respectively;

perform edge detection on the distorted sample grayscale image and the ordinary sample grayscale image to obtain a distorted sample profile diagram and an ordinary sample profile diagram by adopting an edge detection algorithm respectively; and

detect the calibration characteristics in the distorted sample profile diagram and the ordinary sample profile diagram to obtain the distorted coordinates (x d , y d ) and the undistorted coordinates (x u , y u ) by adopting a Hough transform algorithm respectively.

10. The device for correction restoration and analysis alarming of the distorted image according to claim 7 , wherein the correction component is further configured to:

substitute the original distorted coordinates (x d ′, y d ′) and the distortion parameters (k 1 , k 2 ) into formula:

x

di

′

=

x

u

′

1

+

k

1

⁢

r

i

-

1

′2

+

k

2

⁢

r

i

-

1

′4

,

⁢

y

di

′

=

y

u

′

1

+

k

1

⁢

r

i

-

1

′2

+

k

2

⁢

r

i

-

1

′4

,

and perform iteration for n′ times to obtain the restored coordinates (x u ′, y u ′), wherein (x di ′, y di ′) are original distorted coordinates after ith iteration, r i-1 ′ is a radius of the original distorted image after (i−1)th iteration, and an initial value is r 0 ′=√{square root over (x u ′2 +y u ′2 )}.

11. The device for correction restoration and analysis alarming of the distorted image according to claim 7 , wherein the correction component is further configured to

substitute the restored coordinates (x u ′, y u ′) into formula:

[

x

u

′

y

u

′

]

=

f

z

c

⁡

[

x

c

y

c

⁢

]

to obtain lens coordinates (x c , y c , z c ), wherein f is an equivalent focal distance of the distortion lens; and

substitute the direction angles (α, β, γ) and lens coordinates (x c , y c , z c ) of the distortion lens into formula:

[

x

c

y

c

⁢

z

c

]

=

[

α

β

λ

]

⁡

[

k

1

k

2

k

3

k

4

k

5

k

6

k

7

k

8

k

9

]

⁡

[

x

u

″

y

u

″

z

u

″

]

+

[

t

x

t

y

t

z

]

to obtain the restored coordinates (x u ″, y u ″) corresponding to the direction angles (α, β, γ) of the distortion lens, wherein

[

k

1

k

2

k

3

k

4

k

5

k

6

k

7

k

8

k

9

]

is an optical characteristic parameter matrix of the distortion lens, and

[

t

x

t

y

t

z

]

is a translation vector.

12. The device for correction restoration and analysis alarming of the distorted image according to claim 7 , wherein the programming components further comprise:

a repair component configured to repair hole coordinate points in the restored image by adopting an interpolation algorithm; and

a pixel restoration component configured to restore pixels of the original distorted coordinates to the restored coordinates according to corresponding relationships between the original distorted coordinates of the original distorted image and the restored coordinates of the restored image.

13. The method for correction restoration and analysis alarming of the distorted image according to claim 1 , wherein after the pre-stored distortion parameters (k 1 , k 2 ) of the distortion lens are acquired and the restored coordinates (x u ′, y u ′) of each coordinate point in the original distorted image are determined to obtain the restored image, the method further comprises:

performing adaptive background learning on multi-frame images restored to obtain a reference background;

performing segmentation processing on subsequent restored images according to the reference background by adopting an image segmentation algorithm, and extracting a different target characteristic;

performing filtering processing on the different target characteristic to exclude a restored image of which the different target characteristic is within a filtering range;

performing motion matching tracking of the different target characteristic on the subsequent restored images by adopting the different target characteristic out of the filtering range; and

when the number of restored images with the different target characteristic exceeds a preset frame number, recording the restored images as images with a sudden change, and giving an alarm.

14. The device for correction restoration and analysis alarming of the distorted image according to claim 7 , further comprising an alarming analysis component configured to

perform adaptive background learning on multi-frame images restored to obtain a reference background;

perform segmentation processing on subsequent restored images according to the reference background by adopting an image segmentation algorithm, and extract a different target characteristic;

perform filtering processing on the different target characteristic to exclude a restored image of which the different target characteristic is within a filtering range;

perform motion matching tracking of the different target characteristic on the subsequent restored images by adopting the different target characteristic out of the filtering range; and

when the number of restored images with the different target characteristic exceeds a preset frame number, records the restored images as images with a sudden change, and give an alarm.

15. The method for correction restoration and analysis alarming of the distorted image according to claim 2 , wherein after the pre-stored distortion parameters (k 1 , k 2 ) of the distortion lens are acquired and the restored coordinates (x u ′, y u ′) of each coordinate point in the original distorted image are determined to obtain the restored image, the method further comprises:

performing adaptive background learning on multi-frame images restored to obtain a reference background;

performing segmentation processing on subsequent restored images according to the reference background by adopting an image segmentation algorithm, and extracting a different target characteristic;

performing filtering processing on the different target characteristic to exclude a restored image of which the different target characteristic is within a filtering range;

performing motion matching tracking of the different target characteristic on the subsequent restored images by adopting the different target characteristic out of the filtering range; and

when the number of restored images with the different target characteristic exceeds a preset frame number, recording the restored images as images with a sudden change, and giving an alarm.

16. The method for correction restoration and analysis alarming of the distorted image according to claim 3 , wherein after the pre-stored distortion parameters (k 1 , k 2 ) of the distortion lens are acquired and the restored coordinates (x u ′, y u ′) of each coordinate point in the original distorted image are determined to obtain the restored image, the method further comprises:

performing adaptive background learning on multi-frame images restored to obtain a reference background;

performing segmentation processing on subsequent restored images according to the reference background by adopting an image segmentation algorithm, and extracting a different target characteristic;

performing filtering processing on the different target characteristic to exclude a restored image of which the different target characteristic is within a filtering range;

performing motion matching tracking of the different target characteristic on the subsequent restored images by adopting the different target characteristic out of the filtering range; and

when the number of restored images with the different target characteristic exceeds a preset frame number, recording the restored images as images with a sudden change, and giving an alarm.

17. The method for correction restoration and analysis alarming of the distorted image according to claim 4 , wherein after the pre-stored distortion parameters (k 1 , k 2 ) of the distortion lens are acquired and the restored coordinates (x u ′, y u ′) of each coordinate point in the original distorted image are determined to obtain the restored image, the method further comprises:

performing adaptive background learning on multi-frame images restored to obtain a reference background;

performing segmentation processing on subsequent restored images according to the reference background by adopting an image segmentation algorithm, and extracting a different target characteristic;

performing filtering processing on the different target characteristic to exclude a restored image of which the different target characteristic is within a filtering range;

performing motion matching tracking of the different target characteristic on the subsequent restored images by adopting the different target characteristic out of the filtering range; and

when the number of restored images with the different target characteristic exceeds a preset frame number, recording the restored images as images with a sudden change, and giving an alarm.