Dark channel prior defogging method based on atmospheric light value and transmissivity improvement

Abstract

The invention discloses a dark channel prior defogging method based on atmospheric light value and transmissivity improvement, and the method comprises the steps: inputting a haze image, and defining each color channel of the inputted foggy color image; calculating an atmospheric light matrix according to the input image, and calculating the optimized atmospheric light value; calculating a dark channel image according to the input image, and calculating a transmissivity image; optimizing the transmissivity image according to the optimized atmospheric light value; and inputting the original input haze image and the optimized transmissivity image into the atomization imaging model, adding a compensation factor, and finally obtaining a fogless image. According to the scheme, the quality of the defogged image is remarkably improved, the image brightness is improved, the image noise is reduced, the calculation complexity and the energy consumption are reduced, and the adaptability to various foggy environments is enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of digital image processing, and particularly relates to a dark channel prior dehazing method based on improved atmospheric light value and transmittance. BACKGROUND

[0002] The existing original dark channel prior dehazing technology has good effects on removing near scenes or uniform thin fog in images in actual application, but when there are distinct far scenes and near scenes in images, especially when there are a large number of sky backgrounds, the brightness of the dehazed image is overall low, the image noise is aggravated, and the sky area appears a "patch" phenomenon, the main reason is that the calculation of the atmospheric light value is in the whole image, which cannot meet the images with rich depth of field, the calculation of the original transmittance does not consider the scene adaptation problem, and the sky color distortion problem after dehazing is not compensated in the final dehazing model. Although there are many improved algorithms for this problem, the actual processing effect is limited, the algorithm complexity is also increased, and the problem is not solved from the algorithm itself. SUMMARY

[0003] In view of the above technical problems, the present application provides a dark channel prior dehazing method based on improved atmospheric light value and transmittance, which is especially suitable for image enhancement and restoration in foggy environment, and particularly relates to the improvement of the transmittance and atmospheric light value in the dark channel prior dehazing technology. The method can be applied to the fields of automatic driving, security monitoring, unmanned aerial vehicle remote sensing and the like.

[0004] The present application is implemented by using the following technical scheme: A dark channel prior dehazing method based on improved atmospheric light value and transmittance, comprising the following steps: Step S1: input a foggy image, and define each color channel of the input foggy color image; Step S2: calculate an atmospheric light matrix according to the input image, and calculate an optimized atmospheric light value; Step S3: calculate a dark channel image according to the input image, and calculate a transmittance image; Step S4: optimize the transmittance image according to the optimized atmospheric light value in step S2; Step S5: input the original input foggy image and the optimized transmittance image into a fogging imaging model, and add a compensation factor, and finally obtain a non-fog image.

[0005] Specifically, the foggy image in step S1 is defined as , and the red, green and blue color channels of the color image are respectively represented as , and .

[0006] Specifically, step S2 specifically comprises: Step S21: Calculate the maximum color component of each pixel position , denoted as: ; Step S22: Apply the mean filter function to the maximum color component , to obtain the atmospheric light matrix, denoted as: ; wherein, is the atmospheric light value, denotes the atmospheric light matrix; is the mean filter template size, in the interval [5, 11], and generally takes the experience value 5.

[0007] Specifically, the step S3 of calculating the dark channel image specifically comprises: Step S31: Calculate the minimum color component of each pixel position , denoted as: ; Step S32: Apply the mean filter function to the minimum color component , to obtain the dark channel image, denoted as: ; wherein, is the mean filter template size, in the interval [7, 33], and generally takes the experience value 7; Step S33: Calculate the minimum value of the filtered dark channel image and the minimum color component image corresponding to the pixel, to obtain the final dark channel image, denoted as: ; wherein, is the final dark channel image obtained.

[0008] Specifically, the step S3 of calculating the transmittance image specifically comprises: According to the calculated dark channel image and the atmospheric light matrix , calculate the transmittance image, denoted as: .

[0009] Specifically, the step S4 of optimizing the transmittance image comprises the following sub-steps: Step S41: Calculate the global statistical property mean and maximum of the transmittance image, the mean and the maximum are denoted as: ; ; Step S42: optimizing the transmittance image, denoted as: ; wherein, is a fusion coefficient, is the original transmittance image, is the preliminary optimized transmittance image; Step S43: limiting , denoted as: ; wherein, generally 0.1, is the final optimized transmittance image.

[0010] Specifically, the step S5 of processing the haze imaging model specifically comprises: the atmospheric light value and the final optimized transmittance image are brought into the haze removal equation for haze removal, denoted as: ; wherein, is the original haze image, is the final dark channel image calculated, is the optimized transmittance image, is the haze removal result.

[0011] Specifically, the compensation factor is added in the model, denoted as ; the final haze removal image obtained after adding the compensation factor is denoted as: .

[0012] The present application has the beneficial effects that: the atmospheric light value of the present application is based on local calculation, and finally a matrix of image size is obtained, and the brightness of the close-range part is obviously improved after haze removal; the original transmittance is increased by an adaptive parameter adjustment factor based on the statistical characteristics of the image, and the image noise after haze removal is obviously reduced; a compensation term (+0.1A) is introduced in the haze removal model to prevent sky color distortion; through the improvement of the three aspects, not only the quality of the haze removal image is significantly improved, the image brightness is improved, and the image noise is reduced, but also the calculation complexity and energy consumption are reduced, and the adaptability to various foggy environments is enhanced. BRIEF DESCRIPTION OF DRAWINGS

[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only show some of the embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from the structures shown in the drawings without any creative effort.

[0014] Figure 1 The flow chart of the dark channel prior dehazing process improved based on the atmospheric light value and transmittance in the embodiments of the present application. DETAILED DESCRIPTION

[0015] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the following will combine the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are some of the embodiments of the present application, not all the embodiments. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations.

[0016] It should be noted that: similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings.

[0017] The following will be combined with the drawings Figure 1 Some embodiments of the present application will be described in detail. The following embodiments and features in the embodiments can be combined with each other without conflict.

[0018] The present application proposes a dark channel prior dehazing method improved based on the atmospheric light value and transmittance, including the following steps: Step S1: input the haze image, and define each color channel of the input haze color image; Step S2: calculate the atmospheric light matrix according to the input image, and calculate the optimized atmospheric light value; Step S3: calculate the dark channel image according to the input image, and calculate the transmittance image; Step S4: optimize the transmittance image according to the optimized atmospheric light value in step S2; Step S5: input the original input haze image and the optimized transmittance image into the haze imaging model, and add the compensation factor, and finally obtain the haze-free image.

[0019] The following will be described in detail.

[0020] 1. Haze image: define the input haze color image as: , , , R, G, B represent red, green, blue color channel respectively.

[0021] 2. Calculate atmospheric light value A: (1) Calculate the maximum color component of each pixel position , denoted as: ; (2) Apply the well-known mean filter function to , get: ; where is the mean filter template size, the interval is [5, 11], generally take the experience value 5.

[0022] 3. Dark channel image: (1) Calculate the minimum color component of each pixel position , denoted as: ; (2) Apply the well-known mean filter function to , get dark channel image: ; where is the mean filter template size. The interval is [7, 33], generally take the experience value 7; (3) The minimum value of the filtered image and the above image corresponding to the pixel is obtained, and the dark channel image is obtained: ; where, is the obtained dark channel image.

[0023] 4. Transmittance map: According to the dark channel image and the atmospheric light matrix calculated above, the transmittance map calculation expression is: .

[0024] 5. Transmission map optimization: (1) Calculate the global statistical characteristics mean and maximum value of the transmittance image: Mean: ; Maximum: ; (2) Optimize the transmittance image: ; wherein, is a fusion coefficient, is an original transmittance image, is an optimized transmittance image. In practical applications, generally the fusion coefficient is 0.5, as follows:

[0025] Therefore, the final transmittance image expression is:

[0026] In order to prevent excessive defogging, it is necessary to limit : ; In applications generally 0.1 is taken.

[0027] 6. Atomization imaging model: According to the atmospheric light value and the optimized transmittance image , the defogging equation is brought in to defog: ; wherein is an original foggy image, is a calculated dark channel image, is an optimized transmittance image, is a defogging result.

[0028] In order to prevent color distortion in a large area of sky and the appearance of "patch" phenomenon, the compensation term is introduced in the defogging model of the algorithm to prevent sky color distortion. Therefore, the final defogging image expression is: .

[0029] In the present scheme, the following improvements are mainly made: Atmospheric light value calculation: the atmospheric light value of the original algorithm is calculated in the global, and finally a value is obtained. The brightness of the near scene part of the defogged image is low. The atmospheric light value of the present scheme is calculated based on the local, and finally a matrix of image size is obtained. The problem of image darkening caused by a value can be avoided.

[0030] Transmittance optimization: the calculation of the transmittance of the original algorithm will aggravate the image noise in scenes with large depth of field. In the present scheme, based on the statistical characteristics of the image, the original transmittance is increased by an adaptive parameter adjustment factor, which obviously reduces the image noise after defogging. The new transmittance formula is different from the original formula; The original transmittance formula is as follows: ; The new transmittance formula of the present solution is as follows: .

[0031] Fog removal model compensation: the original algorithm fog removal model, in a large area of sky will be partial color distortion, "patch" phenomenon, therefore, the present solution in the fog removal model is introduced into the compensation term (+0.1A) to prevent the sky color distortion. Improved fog removal formula is: ; The present solution increases the supplementary term , the increase of this term can reduce the pseudo color caused by the original algorithm sky processing, such as "patch" effect.

[0032] For the foregoing embodiments, in order to simply describe, they are all expressed as a series of action combinations, but those skilled in the art should know that the present application is not limited by the action sequence described, because according to the present application, some steps can be performed in other order or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification belong to preferred embodiments, and the actions involved are not necessarily necessary for the present application.

[0033] In the above embodiments, the basic principles and main features of the present application and the advantages of the present application are described. Those skilled in the art should understand that the present application is not limited by the above embodiments, and the above embodiments and the description in the specification are only to illustrate the principles of the present application. Any modification and change made by those skilled in the art without departing from the spirit and scope of the present application shall be within the protection scope of the claims of the present application.

Claims

1. A dark channel prior dehazing method based on atmospheric light value and transmittance improvement, characterized in that, Includes the following steps: Step S1: Input a haze image and define each color channel of the input hazy color image; Step S2: Calculate the atmospheric light matrix based on the input image, and calculate the optimized atmospheric light value; Step S3: Calculate the dark channel image and the transmittance image based on the input image; Step S4: Optimize the transmittance image based on the atmospheric light value optimization in step S2; Step S5: Input the original haze image and the optimized transmittance image into the haze imaging model, and add a compensation factor to finally obtain a haze-free image.

2. The dark channel prior dehazing method based on atmospheric light value and transmittance improvement as described in claim 1, characterized in that, The haze image in step S1 is defined as The red, green, and blue color channels of a color image are represented as follows: , and .

3. The dark channel prior dehazing method based on atmospheric light value and transmittance improvement as described in claim 2, characterized in that, Step S2 specifically includes: Step S21: Calculate the position of each pixel The largest color component is represented as: ； Step S22: For Apply mean filtering function The atmospheric light matrix is ​​represented as follows: ； in, Atmospheric light value, Represents the atmospheric light matrix; It is the size of the mean filter template, with an interval of [5, 11], and is generally taken as an empirical value of 5.

4. The dark channel prior dehazing method based on atmospheric light value and transmittance improvement as described in claim 3, characterized in that, The calculation of the dark channel image in step S3 specifically includes: Step S31: Calculate the position of each pixel The smallest color component is represented as: ； Step S32: For the smallest color component Apply mean filtering function The resulting dark channel image is represented as: ； in, It is the size of the mean template, with an interval of [7, 33], and is generally taken as an empirical value of 7; Step S33: Filtered dark channel image Image and minimum color component The final dark channel image is obtained by finding the minimum value of the corresponding pixels in the image, as shown below: ； in, This is the final dark channel image obtained.

5. The dark channel prior dehazing method based on atmospheric light value and transmittance improvement as described in claim 4, characterized in that, The calculation of the transmittance image in step S3 is specifically as follows: Based on the calculated dark channel image and atmospheric light matrix Calculate the transmittance image, which is represented as: 。 6. The dark channel prior dehazing method based on atmospheric light value and transmittance improvement as described in claim 5, characterized in that, The step S4, optimizing the transmittance image, includes the following sub-steps: Step S41: Calculate the mean and maximum values ​​of the global statistical characteristics of the transmittance image. and maximum value They are represented as follows: ； ； Step S42: Optimize the transmittance image, represented as: ； in, The fusion coefficient is... This is the original transmittance image. To initially optimize the transmittance image; Step S43: For To impose restrictions, it is represented as: ； in, Generally, 0.1 is used. To ultimately optimize the transmittance image.

7. The dark channel prior dehazing method based on atmospheric light value and transmittance improvement as described in claim 6, characterized in that, The processing of the fogging imaging model in step S5 specifically includes: atmospheric light value and the final optimized transmittance image Substituting the values ​​into the defogging equation, we can perform defogging as follows: ； in, The original haze image. The final dark channel image obtained from the calculation, To optimize the transmittance image, The result is for defogging.

8. The dark channel prior dehazing method based on atmospheric light value and transmittance improvement as described in claim 7, characterized in that, The compensation factor is added to the model and is represented as follows: The final dehazed image obtained after adding a compensation factor is represented as follows: 。

Images