KR20140139742A - Method and apparatus for processing image including for area - Google Patents

Abstract

The present invention relates to a method and an apparatus for processing an image including a foggy area. The image processing method according to an embodiment of the present invention comprises: a step of obtaining brightness of respective pixels in an input image; a step of generating a histogram representing a brightness distribution of the input image; a step of splitting the histogram into a plurality of sub-areas; a step of determining initial redistribution ratios for each of the sub-areas by using a ratio of the number of pixels included in the sub-areas to the number of all pixels in the input image; a step of applying a weight to the initial redistribution ratios by considering a foggy area; and a step of applying a minimum pixel value and a maximum pixel value of an output image to the initial redistribution ratios, to which the weight is applied, to determine a final redistribution ratio. According to the invention, a natural image can be outputted after a fog distortion compensation process without post-processing steps.

Description

METHOD AND APPARATUS FOR PROCESSING IMAGE INCLUDING FOR AREA [0002]

The present invention relates to an image processing method and apparatus, and more particularly, to a method and apparatus for processing an image including a fog area.

The degradation of the camera image due to atmospheric or climatic influences is caused by light scattering. New light generated by light scattering by atmospheric or fog particles is called air light. The atmospheric scattering light affects the brightness of the light incident on the sensor of the camera, which causes the image quality to deteriorate.

By estimating the atmospheric scattering light due to the atmospheric scattering phenomenon and using this result in reverse to estimate the brightness before scattering, the fog distortion caused by the atmospheric scattering light can be corrected. The atmospheric scattering light is composed of bright luminance and is characterized by its distance. By using this feature, local calculation is performed and the scattered light of the air is estimated by pixel evaluation. One of these evaluation methods is a dark prior method of estimating scattered light by evaluating pixels having low luminance in the image.

Another method of evaluation is the Retinex method, which is a reflection component estimation method. The Retinex method is a method of separating an image into a reflection component and an illumination component. Here, assuming that the atmospheric scattering light is an illumination component, the fog distortion correction image is obtained by appropriately adjusting the separated object information (reflection component) and atmospheric scattering light (illumination component).

On the other hand, image degradation due to atmospheric scattering light is similar to contrast degradation of an image, so distortion caused by the fog area can be corrected by improving the contrast of the image. One of these contrast enhancement methods is the histogram equalization technique. The histogram smoothing technique improves the contrast of images by appropriately distributing the deflected luminance components in the image. This technique has the merit of obtaining a simple and natural image compared to the scattered light estimation method.

Since conventional techniques for fog distortion correction mainly use local operations, the use of line memory in hardware configuration is essential. Also, since the object information in the fog area is excessively emphasized, the image is unnatural, and color distortion and the like are generated. Therefore, a post-processing function is further required.

In the case of the contrast improvement method, it is applicable when the degree of fog in the image is generally similar. However, when the degree of fog in the image is different, since the luminance expressive power is fixed from 0 to 255, the bright and dark portions of the image become saturated and unsaturated, respectively.

It is an object of the present invention to provide an image processing method and apparatus capable of solving the problems of adaptive application of existing contrast improvement methods.

Another object of the present invention is to provide an image processing method and apparatus capable of increasing hardware resource utilization efficiency by not using a line memory in hardware implementation for fog distortion correction.

It is another object of the present invention to provide an image processing method and apparatus capable of outputting a natural image without post-processing even after correction of fog distortion.

The objects of the present invention are not limited to the above-mentioned objects, and other objects and advantages of the present invention which are not mentioned can be understood by the following description and more clearly understood by the embodiments of the present invention. It will also be readily apparent that the objects and advantages of the invention may be realized and attained by means of the instrumentalities and combinations particularly pointed out in the appended claims.

According to an aspect of the present invention, there is provided an image processing method including the steps of: obtaining luminance of each pixel of an input image; generating a histogram representing a luminance distribution of the input image; dividing the histogram into a plurality of sub- Determining an initial redistribution ratio of each of the plurality of sub regions using the ratio of the number of pixels included in the plurality of sub regions to the total number of pixels of the input image; Applying a weight; And determining a final redistribution ratio by applying a minimum luminance value and a maximum luminance value of the output image to the initial redistribution ratio to which the weight is applied.

According to another aspect of the present invention, there is provided an image processing apparatus including a histogram generator for obtaining a luminance of each pixel of an input image and generating a histogram representing a luminance distribution of the input image, a divider for dividing the histogram into a plurality of sub- A first calculation unit for determining an initial redistribution ratio of each of the plurality of sub regions using the ratio of the number of pixels included in the plurality of sub regions to the total number of pixels of the input image, A second arithmetic unit for applying the second arithmetic operation unit; And a third calculator for determining a final redistribution ratio by applying a minimum luminance value and a maximum luminance value of the output image to the initial redistribution ratio to which the weight is applied.

According to the present invention as described above, it is possible to solve the problem of the adaptive application of the existing contrast improvement method.

According to the present invention, in hardware implementation for fog distortion correction, there is an advantage that the use efficiency of hardware resources can be improved by not using a line memory.

In addition, according to the present invention, it is possible to output a natural image without post-processing even after correcting fog distortion.

1 is a flowchart of an image processing method according to an embodiment of the present invention;
FIG. 2 is a graph in which a histogram and a histogram of luminance values generated in an embodiment of the present invention are divided into a plurality of sub-areas. FIG.
3 is a graph illustrating a process of redistributing brightness values in an embodiment of the present invention.
4 is a configuration diagram of an image processing apparatus according to an embodiment of the present invention.
5 is an input image before the image processing method according to the present invention is applied.
6 is an output image after the image processing method according to the present invention is applied.

The above and other objects, features, and advantages of the present invention will become more apparent by describing in detail exemplary embodiments thereof with reference to the attached drawings, which are not intended to limit the scope of the present invention. In the following description, well-known functions or constructions are not described in detail since they would obscure the invention in unnecessary detail. Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the drawings, the same reference numerals are used to denote the same or similar elements.

Referring to FIG. 1, an image processing method according to an exemplary embodiment of the present invention includes a step S102 of obtaining luminance of each pixel of an input image, a step S104 of generating a histogram representing luminance distribution of the input image, Dividing the histogram into a plurality of sub regions (S106), determining an initial redistribution ratio of each of the plurality of sub regions by using the ratio of the number of pixels included in the plurality of sub regions to the total number of pixels of the input image A step S110 of applying a weight considering the fog area to an initial redistribution ratio, a step S112 of determining a final redistribution ratio by applying a minimum luminance value and a maximum luminance value of an output image to an initial redistribution ratio to which a weight is applied, , Redistributing the luminance value of the input image using the split position of the histogram and the final redistribution ratio (S114).

Hereinafter, each step shown in FIG. 1 will be described with reference to FIGS. 2 and 3. FIG.

(1) the brightness acquisition step (S102) and the histogram generation step (S104)

First, the luminance of each pixel included in the input image is acquired (S102). To this end, RGB (Red, Green, Blue) data of the input image may be converted into HSV (Hue, Saturation, Value) data. A value is obtained from the HSV data thus converted.

Next, a histogram representing the luminance distribution of the input image is generated using the luminance of each pixel obtained (S104). The histogram generated in the histogram generation step (S104) can be expressed as [Equation 1].

(Where n _i is the number of pixels having the luminance value i)

FIG. 2 shows an example 210 of the histogram generated by the histogram generation step (S104). The horizontal axis of the histogram 210 in FIG. 2 represents the luminance value i in Equation (1), and the vertical axis represents n _i .

(2) Histogram splitting step (S106)

Next, the generated histogram is divided into a plurality of sub-areas (S106). In one embodiment of the present invention, the histogram is divided using the histogram generated through step S104 and the average luminance value. At this time, only valid data is used for data validity.

In the histogram dividing step (S106), the dividing position of the histogram is determined using the average luminance value, the effective luminance minimum value, and the effective luminance maximum value of the input image, and the histogram is divided into a plurality of sub regions according to the determined division positions.

(Where, DPq is the division position, Sp is the effective luminance minimum value of the input image, Ep is the maximum effective luminance value of the input image, Mp is the average luminance value of the input image,

Equation (2) is a formula for determining a position (DPq) for dividing the histogram. In the histogram 210 of FIG. 2, when the number of regions to be divided between the minimum luminance value 0 and the maximum luminance value 255 is k, k + 2 division positions are determined as shown in Equation (2). At this time, the division position is changed by the average luminance value (M _p ) of the input image. The division positions between the effective minimum brightness values (S _p ) and M _p of the input image, the effective maximum brightness values (E _p ) and M _p of the input image are determined at an equal ratio by the number of divisions (k) and M _p .

By dividing the histogram 210 of FIG. 2 according to the determined division position DPq, the graph 220 of FIG. 2 can be obtained. Referring to the graph 220 of FIG. 2, the histogram 210 is divided into a plurality of sub-areas SDRq according to the division position DPq.

(3) Initial redistribution ratio determination step (S108)

FIG. 3 illustrates a process of dividing a histogram of an input image into a plurality of sub-regions, and redistributing each sub-region according to a redistribution ratio.

Next, the initial redistribution ratio of each of the plurality of sub regions is determined using the ratio of the number of pixels included in the plurality of sub regions to the total number of pixels of the input image (S108). To do this, the number of pixels included in each sub-region must be obtained as in Equation (3).

(Where SHst is the number of pixels contained in each sub-region, and H (j) is the number of pixels of the jth luminance value)

When the number of pixels included in each sub-region is obtained, the ratio of the number of pixels in each sub-region to the total number of pixels of the input image, i.e., the initial redistribution ratio is calculated as shown in Equation (4).

(Where RHst? Is the initial redistribution ratio and N is the total number of pixels in the image)

(4) Weight application step considering the fog area (S110)

Redistribution of each sub-region using the initial redistribution ratio calculated in step S108 can provide a contrast improvement effect, but it is difficult to expect a better effect than the conventional contrast improvement method. Therefore, in the present invention, a weight considering the fog area is applied to the initial redistribution ratio calculated in step S108 (S110). When such a weight is applied, the contrast of the sub region corresponding to the fog region becomes higher, so that a higher distortion correction result can be obtained when only the initial redistribution ratio is applied.

In the weight application step (S110), a kth sub-region corresponding to the fog region is first selected among a plurality of sub-regions. When a fog occurs, the light is scattered by the fog when the light is transmitted, and when the scattered light is superimposed on the image processing device, a bright luminance value appears in the fog area. Therefore, an upper region among the divided sub regions, that is, a region having a high luminance value, can be determined as a fog region.

The redistribution ratio of the selected k-th sub-region is expanded, and the ratio of the remaining sub-regions is reduced.

(Where RHst is the initial redistribution ratio, RHst 'is the weighted initial redistribution ratio, and w is the weight)

In Equation (5), the weight (?) Is given to the ratio of the k-th region since the upper sub-region (k-th region of the sub-region) for which the ratio is to be expanded is a relatively bright portion in the input image. In the case of the other sub-regions except the k-th region, the sum of the total ratios is held at 1 by subtracting the product of the ratio of the expanded upper sub-region and the initial redistribution ratio at the initial redistribution ratio.

(5) Final redistribution ratio determination step (S112)

When the input image has a narrow expressive power, if the histogram is distributed with the maximum expressive power, the luminance value of the dark part is expressed as 0, and the luminance value of the bright part is expressed as 255, and the image is unnatural. Accordingly, in the present invention, the final redistribution ratio is determined by applying the minimum luminance value and the maximum luminance value of the output image to the initial redistribution ratio to which the weight is applied (S112). Thus, when the minimum luminance value and the maximum luminance value of the output image are applied, it is possible to prevent the luminance expressive power from being extremely increased in the histogram redistribution process, thereby obtaining a natural output image.

The minimum effective luminance value and the maximum effective luminance value of the output image are determined as shown in Equation (6). That is, by calculating the deviation between the average luminance value of the input image and the luminance representation median (0.5) and the degree of deviation of the histogram, subtracting it from the effective luminance minimum value (S _p ) and the effective luminance maximum value (E _p ) The minimum effective luminance value and the maximum effective luminance value of the image are determined.

(Where, Lp is the minimum luminance value of the output image, Hp is the maximum luminance value of the output image, Ep is the maximum effective luminance value of the input image, and Mp is the average luminance value of the input image)

The luminance expressive power of the output image is defined as (Hp-Lp), and the unnaturalness of the image in the histogram redistribution process is suppressed by this luminance expressive power.

)은 휘도 표현력 문턱값에 할당된 비율만큼 감소된 새로운 휘도 표현력 안에서 결정된다. [수학식 7]에서 분모는 정규화를 위한 항이다.
The final redistribution ratio is determined as in Equation (7) by applying the calculated effective luminance minimum value Lp and effective luminance maximum value Hp of the output image to the initial redistribution ratio to which the weight is applied (S112). In Equation (7), the ratio of the first redistribution ratio and the last redistribution ratio is assigned according to the calculated luminance power threshold value. Accordingly, the final redistribution ratio (

) Is determined in the new luminance expression power reduced by the ratio assigned to the luminance expression threshold value. In Equation (7), the denominator is a term for normalization.

는 최종 재분배 비율, RHst'Ω는 가중치가 적용된 초기 재분배 비율, Lp는 출력 영상의 최소 휘도 값, Hp는 출력 영상의 최대 휘도 값)
(here,

Lp is the minimum luminance value of the output image, and Hp is the maximum luminance value of the output image.

(6) Redistributing the luminance value of the input image (S114)

Finally, the histogram is redistributed according to the calculated final redistribution ratio (S114). In this step, the sub-region including the luminance value of the input image is determined, and then the histogram is redistributed using the calculated final redistribution ratio, the luminance expressive power of the sub-region, and the divided position.

(Where, y (n) is an output luminance value, and x (n) is an input luminance value)

FIG. 3 illustrates a process of dividing a histogram of an input image into a plurality of sub-regions, and redistributing each sub-region according to a redistribution ratio.

Referring to FIG. 3, first, it is checked which sub-region of the plurality of sub-regions the input brightness value x (n) corresponds to (310). Then, the input luminance value x (n) is moved in parallel by subtracting the starting luminance value DPq of the corresponding sub region from the input luminance value x (n) (320).

)을 곱하여 새로운 휘도 값을 계산한다(330). 마지막으로, 계산된 새로운 휘도 값에 재분배 되는 영역의 시작 휘도값(

)을 더해줌으로써 입력 휘도 값을 재분배한다(340).
Then, the input luminance value that is separated by the starting luminance value is added to the final redistribution ratio (

) To calculate a new luminance value (330). Finally, the starting luminance value of the region redistributed to the calculated new luminance value (

) To redistribute the input luminance value (340).

The luminance data V to which the redistribution is applied through the above-described processes (1) to (6) is converted into RGB data together with H (hue) and S (saturation) data, A corrected output image is generated.

4 is a block diagram of an image processing apparatus according to an embodiment of the present invention.

4, an image processing apparatus 402 according to an embodiment of the present invention includes a histogram generator 404, a divider 406, a first calculator 408, a second calculator 410, And an operation unit 412.

The histogram generator 404 obtains the luminance of each pixel of the input image and generates a histogram representing the luminance distribution of the input image.

The division unit 406 divides the generated histogram into a plurality of sub-regions.

The first calculation unit 408 determines an initial redistribution ratio of each of the plurality of sub regions using the ratio of the number of pixels included in the plurality of sub regions to the total number of pixels of the input image.

The second calculation unit 410 applies a weight considering the fog area to the calculated initial redistribution ratio.

The third computing unit 412 determines the final redistribution ratio by applying the minimum luminance value and the maximum luminance value of the output image to the initial redistribution ratio to which the weight is applied.

Although not shown in FIG. 4, the image processing apparatus 402 may further include a redistributing unit for redistributing the luminance value of the input image using the split position and the final redistribution ratio of the histogram.

FIG. 5 shows an input image before the image processing method according to the present invention is applied, and FIG. 6 shows an output image after applying the image processing method according to the present invention.

Referring to FIG. 5, the input image includes a mist region generated due to atmospheric scattering light. Therefore, since the contrast ratio is not high, the image is not clear and the boundary portion of the object is not natural, so that it is difficult to recognize the object in the image.

However, referring to FIG. 6, which is the result of processing the input image according to the image processing method of the present invention, the color and boundary representation of the object become more natural and the contrast ratio is further improved.

The fog distortion correction method according to the present invention as described above does not require a line memory by using a contrast improvement method using a histogram, which is advantageous in a hardware configuration.

Further, according to the present invention, it is possible to solve distortion of other luminance information by the deflection luminance information of the contrast improvement method. Also, the present invention can be applied to an image in which different fog areas are present. By using only luminance information, color distortion can be prevented and a natural image can be obtained without additional post-processing.

While the present invention has been described in connection with what is presently considered to be practical exemplary embodiments, it is to be understood that the invention is not limited to the disclosed embodiments, but, on the contrary, But the present invention is not limited thereto.

Claims (10)

Obtaining a luminance of each pixel of an input image;
Generating a histogram representing a luminance distribution of the input image;
Dividing the histogram into a plurality of sub-regions;
Determining an initial redistribution ratio of each of the plurality of sub regions using a ratio of the number of pixels included in the plurality of sub regions to the total number of pixels of the input image;
Applying a weight to the initial redistribution ratio in consideration of the fog area; And
Determining a final redistribution ratio by applying a minimum luminance value and a maximum luminance value of the output image to the initial redistribution ratio to which the weight is applied
An image processing method comprising:
The method according to claim 1,
The step of dividing the histogram into a plurality of sub-
Determining a split position of the histogram using an average luminance value, an effective luminance minimum value, and an effective luminance maximum value of the input image; And
Dividing the histogram into a plurality of sub-regions according to the division position
An image processing method comprising:
3. The method of claim 2,
The division position of the histogram is determined by the following equation (1)
Image processing method.

[Equation 1]

(Where, DPq is the division position, Sp is the effective luminance minimum value of the input image, Ep is the maximum effective luminance value of the input image, Mp is the average luminance value of the input image,
The method according to claim 1,
The initial redistribution ratio is determined by the following equation (2)
Image processing method.

&Quot; (2) "

(Where SHst is the number of pixels in each sub-region, H (j) is the number of pixels in the jth luminance value, RHst is the initial redistribution ratio, and N is the total number of pixels in the image)
The method according to claim 1,
The step of weighting the initial redistribution ratio
Selecting a kth sub-region corresponding to a fog region among the plurality of sub-regions; And
The step of assigning weights to the k < th > sub-region and the remaining sub-regions by < EMI ID =
An image processing method comprising:

&Quot; (3) "

(Where RHst is the initial redistribution ratio, RHst 'is the weighted initial redistribution ratio, and w is the weight)
The method according to claim 1,
The minimum effective luminance value and the maximum effective luminance value of the output image are respectively determined by the following equation (4)
Image processing method.

&Quot; (4) "

(Where, Lp is the minimum luminance value of the output image, Hp is the maximum luminance value of the output image, Ep is the maximum effective luminance value of the input image, and Mp is the average luminance value of the input image)
The method according to claim 1,
The final redistribution ratio is determined by the following equation (5)
Image processing method.

&Quot; (5) "

(here,

Lp is the minimum luminance value of the output image, and Hp is the maximum luminance value of the output image.
The method according to claim 1,
And redistributing the luminance value of the input image using the split position of the histogram and the final redistribution ratio
Further comprising a video processing method.
9. The method of claim 8,
The step of redistributing the luminance value of the input image is performed by Equation (6)
Image processing method.

&Quot; (6) "

(Where, y (n) is an output luminance value, and x (n) is an input luminance value)
A histogram generation unit for obtaining a luminance of each pixel of an input image and generating a histogram representing a luminance distribution of the input image;
A divider dividing the histogram into a plurality of sub-regions;
A first calculation unit for determining an initial redistribution ratio of each of the plurality of sub regions using a ratio of the number of pixels included in the plurality of sub regions to the total number of pixels of the input image;
A second calculator for applying a weight considering the fog area to the initial redistribution ratio; And
And a third calculation unit for determining a final redistribution ratio by applying the minimum luminance value and the maximum luminance value of the output image to the initial redistribution ratio to which the weight is applied
Included video processing unit.

Images