Image processing method and apparatus for correcting distortion caused by air particles as in fog
View Patent ↗Provided are a method and apparatus for processing an image. The method includes receiving a first luminance image of an image including airlight, which is a type of light that occurs in a foggy environment, and generating an airtight map based on a ratio between an average luminance of the first luminance image, and a standard deviation; and removing the airtight by subtracting the airtight map from the first luminance image and outputting a second luminance image. According to the present invention, airlight components may be effectively removed.
1. An image processing method comprising:
receiving a first luminance image of an image including airlight and generating an airlight map based on a ratio between an average luminance of the first luminance image, and a standard deviation, wherein the generating of the airlight map comprises:
segmenting the first luminance image into a predetermined number of regions;
defining a cost function by using the ratio between the average luminance of the first luminance image, and the standard deviation, with respect to each region, and calculating an airlight component of each region by using the cost function; and
generating the airlight map of the first luminance image by performing a least squares method on airlight components of the regions;
the method further comprising:
removing the airlight by subtracting the airlight map from the first luminance image and outputting a second luminance image.
2. The method of claim 1 , wherein the airlight map represents a degree of influence of the airlight on the image.
3. The method of claim 1 , wherein the segmenting of the first luminance image is adaptively performed based on a depth difference of the first luminance image.
4. The method of claim 1 , wherein the segmenting of the first luminance image comprises summing gradient values in row and column directions of the first luminance image and segmenting the first luminance image with reference to coordinates having maximum sums of the gradient values.
5. The method of claim 1 , further comprising detecting a sky region by using edge information of the first luminance image before segmenting the first luminance image.
6. The method of claim 5 , further comprising performing pre-processing on the first luminance image by extending a luminance representation range of a non-sky region obtained by excluding the sky region from the first luminance image, adjusting luminance of the non-sky region by using a histogram, and generating a mapping function which represents the luminance representation range, by accumulating values of the histogram.
7. The method of claim 6 , wherein the performing of pre-processing comprises adjusting the luminance according to Equation 7:
h new ( k )=( h ( k )+1) 1/n (7);
wherein
h(k) represents a histogram,
k represents a luminance representation range, and
n represents a constant of an exponent.
8. The method of claim 7 , wherein the mapping function maintains the luminance representation range of the sky region.
9. The method of claim 7 , further comprising performing post-processing by compensating for luminance reduction in the second luminance image.
10. The method of claim 9 , further comprising compensating for saturation reduction caused by a luminance variation of the image including the airlight by using the first luminance image and a first chrominance image of the image, and by using the second luminance image on which post-processing is performed.
11. The method of claim 1 , wherein the cost function is calculated based on the following Equation 2:
A
(
λ
)
=
[
STD
(
Y
′
(
i
,
j
)
-
λ
)
MEAN
(
Y
′
(
i
,
j
)
-
λ
)
]
B
(
λ
)
=
[
MEAN
(
Y
′
(
i
,
j
)
-
λ
)
MEAN
(
I
(
i
,
j
)
)
]
×
[
STD
(
I
(
i
,
j
)
)
MEAN
(
I
(
i
,
j
)
)
]
=
[
(
MEAN
(
Y
′
(
i
,
j
)
-
λ
)
]
×
[
STD
(
I
(
i
,
j
)
)
MEAN
(
I
(
i
,
j
)
)
2
]
(
2
)
wherein
Y′(i,j) represents a luminance component of an image including airlight,
I(i,j) represents a luminance component of an image not including airlight,
(i,j) is a coordinate of a pixel, and
λ is a degree of luminance to be subtracted from an image.
12. The method of claim 11 , wherein the airlight component is calculated according to the following Equation 3:
λ Y(i,j) =arg min(| A (λ)− B (λ)|) (3).
13. The method of claim 1 , wherein the airlight is a type of light that occurs in a foggy environment.
14. An image processing method comprising:
receiving a first luminance image of an image including airlight and generating an airlight map based on a ratio between an average luminance of the first luminance image, and a standard deviation;
removing the airlight by subtracting the airlight map from the first luminance image and outputting a second luminance image; and
performing edge enhancement on the second luminance image according to the following Equation 5:
Y out(i,j) =Y ″( i,j )± s×g ( i,j ) (5);
wherein
Y out(i,j) represents a luminance image on which fog distortion correction and edge enhancement are performed,
Y″ represents a second luminance image,
s represents a parameter for controlling a degree of edge enhancement, and
g(i,j) represents an edge component that is passed through a high pass filter.
15. The method of claim 14 , further comprising:
converting a chrominance image of the image including the airlight, and a third luminance image obtained by performing edge enhancement on the second luminance image, into an RGB image; and
performing histogram stretching on the RGB image.
16. An image processing method comprising:
receiving a first luminance image of an image comprising a foggy image and segmenting the first luminance image into a predetermined number of regions;
calculating an airlight component of each region based on a ratio between an average luminance of the first luminance image, and a standard deviation;
generating an airlight map of the first luminance image by performing a least squares method on airlight components of the regions; and
removing an influence of the foggy image by subtracting the airlight map from the first luminance image and outputting a second luminance image.
17. The method of claim 16 , wherein the segmenting of the first luminance image is adaptively performed based on a depth difference of the first luminance image.
18. The method of claim 17 , before segmenting the first luminance image, further comprising:
detecting a sky region by using edge information of the first luminance image; and
performing pre-processing on the first luminance image by adjusting luminance of a non-sky region obtained by excluding the sky region from the first luminance image,
wherein the segmenting of the first luminance image is adaptively performed based on depth information of the first luminance image on which pre-processing is performed.
19. The method of claim 16 , wherein the airlight is a type of light that occurs in a foggy environment.
20. A non-transitory computer readable recording medium having recorded thereon a computer program for executing the method of claim 1 .
21. An image processing apparatus comprising:
an airlight map generator for receiving a first luminance image of an image including airlight and generating an airlight map based on a ratio between an average luminance of the first luminance image, and a standard deviation, wherein the airlight map represents a degree of influence of the airlight on the image and the airlight map generator comprises:
a region segmentor for segmenting the first luminance image into a predetermined number of regions;
an airlight calculator for defining a cost function by using the ratio between the average luminance of the first luminance image, and the standard deviation, with respect to each region, and calculating an airlight component of each region by using the cost function; and
a map generator for generating the airlight map of the first luminance image by performing a least squares method on airlight components of the regions;
the image processing apparatus further comprising:
a subtracter for removing the airlight by subtracting the airlight map from the first luminance image and outputting a second luminance image.
22. The apparatus of claim 21 , further comprising an edge enhancer for performing edge enhancement on the second luminance image output from the subtracter.
23. The apparatus of claim 22 , further comprising:
an RGB converter for converting a chrominance image of the image including the airlight, and a third luminance image output from the edge enhancer, into an RGB image; and
a post-processor for performing histogram stretching on the RGB image.
24. The apparatus of claim 21 , wherein the region segmentor adaptively segments the first luminance image based on a depth difference of the first luminance image.
25. The apparatus of claim 21 , further comprising a sky region detector for detecting a sky region by using edge information of the first luminance image.
26. The apparatus of claim 25 , further comprising a pre-processor for extending a luminance representation range of a non-sky region obtained by excluding the sky region from the first luminance image, adjusting luminance of the non-sky region by using a histogram, and generating a mapping function which represents the luminance representation range, by accumulating values of the histogram.
27. The apparatus of claim 26 , further comprising a post-processor for compensating the second luminance image for luminance reduction.
28. The apparatus of claim 27 , further comprising a chrominance compensator for compensating for saturation reduction caused by a luminance variation of the image including the airlight by using the first luminance image and a first chrominance image of the image, and the second luminance image on which post-processing is performed.
29. The apparatus of claim 21 , wherein the airlight is a type of light that occurs in a foggy environment.
30. An image processing apparatus comprising:
an airlight map generator for receiving a first luminance image of an image including airlight and generating an airlight map based on a ratio between an average luminance of the first luminance image, and a standard deviation, wherein the airlight map represents a degree of influence of the airlight on the image;
a subtracter for removing the airlight by subtracting the airlight map from the first luminance image and outputting a second luminance image;
a sky region detector for detecting a sky region by using edge information of the first luminance image; and
a pre-processor for extending a luminance representation range of a non-sky region obtained by excluding the sky region from the first luminance image, adjusting luminance of the non-sky region by using a histogram, and generating a mapping function which represents the luminance representation range, by accumulating values of the histogram.