A method and system for optimizing low-light image quality in tunnel environments

Abstract

This invention proposes a method and system for optimizing low-light image quality in tunnel environments. Addressing the issue of uneven light source distribution in curved tunnels, resulting in darker distant areas and brighter near areas, the method improves image quality through multi-step processing. Specifically, it includes: linear normalization and adaptive threshold segmentation of the tunnel image to divide it into near and far regions; constructing a channel mask matrix and reflection components for each region using RGB channel separation and discrete cosine transform; achieving color balance across regions through histogram equalization and channel merging; and finally fusing near and far region images through pixel coordinate mapping to generate an optimized enhanced tunnel image. This invention effectively improves upon the problem of overexposure in traditional methods for images with significant brightness differences, achieving better low-light image enhancement in tunnel environments.

Description

Technical Field

[0001] This invention belongs to the field of image processing, and in particular relates to a method and system for optimizing low-light image quality in tunnel environments. Background Technology

[0002] In tunnel environments, uneven lighting distribution often results in a "dark in the distance, bright in the foreground" problem. Existing traditional image processing methods only address specific issues and lack a comprehensive, systematic approach to improve overall image quality. They often fail to simultaneously enhance the brightness of distant objects while avoiding overexposure of nearby areas. To address this problem, this invention combines an adaptive thresholding method with image processing techniques to effectively adjust the brightness of tunnel images and improve visual effects. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention proposes a method and system for optimizing low-light image quality in tunnel environments, aiming to significantly improve the visual quality and detail representation of images.

[0004] This invention effectively improves image contrast and sharpness by decomposing and enhancing the illumination and reflection components of an image. The method steps are as follows:

[0005] Step 1: Input the tunnel image, and normalize each pixel in the tunnel image using a linear normalization method to obtain the normalized tunnel image;

[0006] Step 2: The normalized tunnel image is segmented using an adaptive thresholding method to obtain the near region image and the far region image of the normalized tunnel image.

[0007] Step 3: Perform image separation processing on the near region image of the normalized tunnel image using the RGB channel separation method to obtain a three-channel image of the near region image. Perform image separation processing on the far region image of the normalized tunnel image using the RGB channel separation method to obtain a three-channel image of the far region image.

[0008] Step 4: Perform discrete cosine transform on each channel image of the near region image to obtain the discrete cosine transform coefficient matrix of each channel image of the near region image in the frequency domain, construct the mask matrix of each channel image of the near region image, and construct the reflection component of each channel image of the near region image.

[0009] Step 5: Perform discrete cosine transform on each channel image of the distant region to obtain the discrete cosine transform coefficient matrix of each channel image of the distant region in the frequency domain, construct the mask matrix of each channel image of the distant region, and construct the reflection component of each channel image of the distant region.

[0010] Step 6: Obtain the near-field color-balanced image and the far-field color-balanced image by RGB channel merging and histogram equalization respectively. Then, fuse the near-field color-balanced image and the far-field color-balanced image by pixel coordinate mapping to obtain the tunnel-enhanced image.

[0011] Preferably, step 4 involves constructing the mask matrix for each channel of the near-field image, and the specific process is as follows:

[0012] The width and height of the tunnel image are used as the number of rows and columns of the mask matrix for each channel image of the near region image, respectively.

[0013] The values ​​of the elements of the mask matrix for each channel of the nearby region image are set by combining the discrete cosine transform coefficient matrix of each channel image in the frequency domain, as follows:

[0014] The threshold k for separating high-frequency and low-frequency information is determined through statistical analysis and threshold judgment.

[0015] When row index i in the mask matrix is ​​less than or equal to column index j plus a threshold k, the corresponding position in the mask matrix is ​​set to 1, and the remaining positions in the mask matrix are set to 0.

[0016] Step 4 involves constructing the reflection component of each channel image of the near-field region, as detailed below:

[0017] The low-frequency component matrix of each channel image of the near region is obtained by performing a matrix dot product between the mask matrix of each channel image of the near region and the discrete cosine transform coefficient matrix of each channel image of the near region.

[0018] The low-frequency component matrix of each channel image of the near region is obtained by inverse discrete cosine transform to obtain the illumination component of each channel image of the near region.

[0019] Further, the enhanced illumination components of each channel image of the near region are obtained through logarithmic transformation and Gamma correction.

[0020] The enhanced illumination component of each channel image of the near region is smoothed by Gaussian filtering to obtain the smoothed illumination component of each channel image of the near region.

[0021] The smoothed illumination component of each channel image of the near region is calculated by combining the retinal cortex theory to obtain the reflection component of each channel image of the near region.

[0022] Preferably, step 5 involves constructing the mask matrix for each channel of the distant region image, and the specific process is as follows:

[0023] The width and height of the tunnel image are used as the number of rows and columns of the mask matrix for each channel image of the distant region image, respectively.

[0024] The values ​​of the elements of the mask matrix for each channel of the distant region image are set by combining the discrete cosine transform coefficient matrix of each channel image in the frequency domain of the distant region image, as follows:

[0025] The threshold k for separating high-frequency and low-frequency information is determined through statistical analysis and threshold judgment. When the row index i in the mask matrix is ​​less than or equal to the column index j plus the threshold k, the corresponding position in the mask matrix is ​​set to 1, and the remaining positions in the mask matrix are set to 0.

[0026] Step 5 involves constructing the reflection components of each channel image of the distant region, as detailed below:

[0027] The low-frequency component matrix of each channel image of the distant region is obtained by performing a matrix dot product between the mask matrix of each channel image of the distant region and the discrete cosine transform coefficient matrix of each channel image of the distant region.

[0028] The low-frequency component matrix of each channel image of the distant region is obtained by performing an inverse discrete cosine transform to obtain the illumination component of each channel image of the distant region.

[0029] Further, the enhanced illumination components of each channel image of the distant region are obtained through logarithmic transformation and Gamma correction.

[0030] The enhanced illumination components of each channel image of the distant region are smoothed by Gaussian filtering to obtain the smoothed illumination components of each channel image of the distant region.

[0031] The smoothed illumination component of each channel image of the distant region is calculated by combining the retinal cortex theory to obtain the reflection component of each channel image of the distant region.

[0032] Preferably, step 6 involves sequentially performing RGB channel merging and histogram equalization to obtain the color-balanced images of the near and far regions, as detailed below:

[0033] The reflection components of each channel of the near-field image are merged into RGB channels to obtain the near-field enhanced image and the far-field enhanced image. The near-field enhanced image and the far-field enhanced image are then processed by histogram equalization to obtain the near-field color balanced image and the far-field color balanced image.

[0034] This invention also proposes a low-light image quality optimization system for tunnel environments, as detailed below:

[0035] The image acquisition and preprocessing module is used to input the tunnel image, normalize each pixel in the tunnel image using a linear normalization method, and obtain the normalized tunnel image.

[0036] The image segmentation module is used to segment the normalized tunnel image using an adaptive threshold segmentation method to obtain the near region image and the far region image of the normalized tunnel image.

[0037] The channel separation module is used to perform image separation processing on the near region image of the normalized tunnel image using the RGB channel separation method to obtain a three-channel image of the near region image, and to perform image separation processing on the far region image of the normalized tunnel image using the RGB channel separation method to obtain a three-channel image of the far region image.

[0038] The near-field region reflection component construction module is used to perform discrete cosine transform on each channel image of the near-field region image to obtain the discrete cosine transform coefficient matrix of each channel image of the near-field region image in the frequency domain, construct the mask matrix of each channel image of the near-field region image, and construct the reflection component of each channel image of the near-field region image.

[0039] The distant region reflection component construction module is used to perform discrete cosine transform on each channel image of the distant region image to obtain the discrete cosine transform coefficient matrix of each channel image of the distant region image in the frequency domain, construct the mask matrix of each channel image of the distant region image, and construct the reflection component of each channel image of the distant region image.

[0040] The tunnel enhancement image construction module is used to obtain the near-area color-balanced image and the far-area color-balanced image by RGB channel merging and histogram equalization. The near-area color-balanced image and the far-area color-balanced image are then fused by pixel coordinate mapping to obtain the tunnel enhancement image.

[0041] The specific beneficial effects of this invention are as follows:

[0042] An adaptive threshold method for tunnel near and far scene segmentation: By adaptively adjusting the threshold, this method solves the problem that traditional methods cannot handle in curved tunnels, where uneven light source distribution causes dark areas in the distance and bright areas in the foreground.

[0043] Comprehensive Processing Flow: This invention provides a comprehensive processing flow that organically combines multiple image processing steps, such as image reading, image segmentation, channel separation, various transformation processing (e.g., DCT, logarithmic transformation, illumination enhancement, reflection component calculation, etc.), denoising, histogram equalization, and Retinex theory. This comprehensive and systematic process can improve image quality and solve various image quality problems that are difficult to address with traditional single-processing methods.

[0044] Flexible illumination enhancement and reflection component processing: During the processing, an illumination enhancement mechanism with selectable methods (such as logarithmic transformation, Gamma transformation, etc.) and a method for calculating the reflection component based on illumination and the original image are adopted. This allows for flexible adjustment of the illumination effect and reflection characteristics of the image according to the characteristics and needs of different images, further optimizing image quality. Attached Figure Description

[0045] Figure 1 : Flowchart of the method according to an embodiment of the present invention;

[0046] Figure 2 : Processing effect diagram of tunnel image with obvious bright and dark areas;

[0047] Figure 3 : Processing effect diagram of a non-tunnel image with obvious bright and dark areas. Detailed Implementation

[0048] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0049] The algorithm proposed in this invention is written in Python 3.8 under the Ubuntu 20.04 environment. Throughout the design and testing of the experiment, Intel-i7 and NVIDIA 1660Ti hardware devices were used.

[0050] The following is combined Figures 1 to 3 The present invention specifically describes a method and system for optimizing low-light image quality in tunnel environments.

[0051] like Figure 1 The diagram shown is a flowchart of a method according to an embodiment of the present invention. The specific implementation is a method for optimizing low-light image quality in a tunnel environment, as detailed below:

[0052] Step 1: Input the tunnel image, and normalize each pixel in the tunnel image using a linear normalization method to obtain the normalized tunnel image;

[0053] Step 2: The normalized tunnel image is segmented using an adaptive thresholding method to obtain the near region image and the far region image of the normalized tunnel image.

[0054] Step 3: Perform image separation processing on the near region image of the normalized tunnel image using the RGB channel separation method to obtain a three-channel image of the near region image. Perform image separation processing on the far region image of the normalized tunnel image using the RGB channel separation method to obtain a three-channel image of the far region image.

[0055] Step 4: Perform discrete cosine transform on each channel image of the near region image to obtain the discrete cosine transform coefficient matrix of each channel image of the near region image in the frequency domain, and construct the mask matrix of each channel image of the near region image.

[0056] Step 4 involves constructing the mask matrix for each channel of the near-field image. The specific process is as follows:

[0057] The width and height of the tunnel image are used as the number of rows and columns of the mask matrix for each channel image of the near region image, respectively.

[0058] The values ​​of the elements of the mask matrix for each channel of the nearby region image are set by combining the discrete cosine transform coefficient matrix of each channel image in the frequency domain, as follows:

[0059] The threshold k for separating high-frequency and low-frequency information is determined through statistical analysis and threshold judgment.

[0060] When row index i in the mask matrix is ​​less than or equal to column index j plus a threshold k, the corresponding position in the mask matrix is ​​set to 1, and the remaining positions in the mask matrix are set to 0.

[0061] Step 4 involves constructing the reflection component of each channel image of the near-field region, as detailed below:

[0062] The low-frequency component matrix of each channel image of the near region is obtained by performing a matrix dot product between the mask matrix of each channel image of the near region and the discrete cosine transform coefficient matrix of each channel image of the near region.

[0063] The low-frequency component matrix of each channel image of the near region is obtained by inverse discrete cosine transform to obtain the illumination component of each channel image of the near region.

[0064] Further, the enhanced illumination components of each channel image of the near region are obtained through logarithmic transformation and Gamma correction.

[0065] The enhanced illumination component of each channel image of the near region is smoothed by Gaussian filtering to obtain the smoothed illumination component of each channel image of the near region.

[0066] The smoothed illumination component of each channel image of the near region is calculated by combining the retinal cortex theory to obtain the reflection component of each channel image of the near region.

[0067] Step 5: Perform discrete cosine transform on each channel image of the distant region to obtain the discrete cosine transform coefficient matrix of each channel image of the distant region in the frequency domain, construct the mask matrix of each channel image of the distant region, and construct the reflection component of each channel image of the distant region.

[0068] Step 5 involves constructing the mask matrix for each channel of the distant region image. The specific process is as follows:

[0069] The width and height of the tunnel image are used as the number of rows and columns of the mask matrix for each channel image of the distant region image, respectively.

[0070] The values ​​of the elements of the mask matrix for each channel of the distant region image are set by combining the discrete cosine transform coefficient matrix of each channel image in the frequency domain of the distant region image, as follows:

[0071] The threshold k for separating high-frequency and low-frequency information is determined through statistical analysis and threshold judgment. When the row index i in the mask matrix is ​​less than or equal to the column index j plus the threshold k, the corresponding position in the mask matrix is ​​set to 1, and the remaining positions in the mask matrix are set to 0.

[0072] Step 5 involves constructing the reflection components of each channel image of the distant region, as detailed below:

[0073] The low-frequency component matrix of each channel image of the distant region is obtained by performing a matrix dot product between the mask matrix of each channel image of the distant region and the discrete cosine transform coefficient matrix of each channel image of the distant region.

[0074] The low-frequency component matrix of each channel image of the distant region is obtained by performing an inverse discrete cosine transform to obtain the illumination component of each channel image of the distant region.

[0075] Further, the enhanced illumination components of each channel image of the distant region are obtained through logarithmic transformation and Gamma correction.

[0076] The enhanced illumination components of each channel image of the distant region are smoothed by Gaussian filtering to obtain the smoothed illumination components of each channel image of the distant region.

[0077] The smoothed illumination component of each channel image of the distant region is calculated by combining the retinal cortex theory to obtain the reflection component of each channel image of the distant region.

[0078] Step 6: Obtain the near-field color-balanced image and the far-field color-balanced image by RGB channel merging and histogram equalization respectively. Then, fuse the near-field color-balanced image and the far-field color-balanced image by pixel coordinate mapping to obtain the tunnel-enhanced image.

[0079] Step 6 involves RGB channel merging and histogram equalization to obtain the color-balanced images of the near and far regions, as detailed below:

[0080] The reflection components of each channel of the near-field image are merged into RGB channels to obtain the near-field enhanced image and the far-field enhanced image. The near-field enhanced image and the far-field enhanced image are then processed by histogram equalization to obtain the near-field color balanced image and the far-field color balanced image.

[0081] The specific implementation of the system of the present invention is a low-light image quality optimization system for tunnel environments, as detailed below:

[0082] The image acquisition and preprocessing module is used to input the tunnel image, normalize each pixel in the tunnel image using a linear normalization method, and obtain the normalized tunnel image.

[0083] The image segmentation module is used to segment the normalized tunnel image using an adaptive threshold segmentation method to obtain the near region image and the far region image of the normalized tunnel image.

[0084] The channel separation module is used to perform image separation processing on the near region image of the normalized tunnel image using the RGB channel separation method to obtain a three-channel image of the near region image, and to perform image separation processing on the far region image of the normalized tunnel image using the RGB channel separation method to obtain a three-channel image of the far region image.

[0085] The near-field region reflection component construction module is used to perform discrete cosine transform on each channel image of the near-field region image to obtain the discrete cosine transform coefficient matrix of each channel image of the near-field region image in the frequency domain, and to construct the mask matrix of each channel image of the near-field region image.

[0086] The distant region reflection component construction module is used to perform discrete cosine transform on each channel image of the distant region image to obtain the discrete cosine transform coefficient matrix of each channel image of the distant region image in the frequency domain, construct the mask matrix of each channel image of the distant region image, and construct the reflection component of each channel image of the distant region image.

[0087] The tunnel enhancement image construction module is used to obtain the near-area color-balanced image and the far-area color-balanced image by RGB channel merging and histogram equalization. The near-area color-balanced image and the far-area color-balanced image are then fused by pixel coordinate mapping to obtain the tunnel enhancement image.

[0088] Figure 2 and Figure 3 As shown, this invention eliminates the problem of overexposure in the bright areas of the original image when enhancing images with obvious brightness and darkness. It implements an improved low-light image enhancement optimization method to address this problem and can be better applied to low-light image enhancement in tunnel environments.

[0089] In specific implementation, the method proposed in the technical solution of this invention can be automatically executed by those skilled in the art using computer software technology. System devices for implementing the method, such as computer-readable storage media storing the corresponding computer program of the technical solution of this invention and computer equipment including the computer program running the corresponding computer program, should also be within the protection scope of this invention.

[0090] It should be understood that any parts not described in detail in this specification belong to the prior art.

[0091] It should be understood that the above description of the embodiments is quite detailed, but it should not be considered as a limitation on the scope of protection of this invention. Those skilled in the art can make substitutions or modifications under the guidance of this invention without departing from the scope of protection of the claims of this invention, and all such substitutions or modifications fall within the scope of protection of this invention. The scope of protection of this invention should be determined by the appended claims.

Claims

1. A method for optimizing low-light image quality in tunnel environments, characterized in that, Includes the following steps: Step 1: Input the tunnel image, and normalize each pixel in the tunnel image using a linear normalization method to obtain the normalized tunnel image; Step 2: The normalized tunnel image is segmented using an adaptive thresholding method to obtain the near region image and the far region image of the normalized tunnel image. Step 3: Perform image separation processing on the near region image of the normalized tunnel image using the RGB channel separation method to obtain a three-channel image of the near region image. Perform image separation processing on the far region image of the normalized tunnel image using the RGB channel separation method to obtain a three-channel image of the far region image. Step 4: Perform discrete cosine transform on each channel image of the near region image to obtain the discrete cosine transform coefficient matrix of each channel image of the near region image in the frequency domain, construct the mask matrix of each channel image of the near region image, and construct the reflection component of each channel image of the near region image. Step 5: Perform discrete cosine transform on each channel image of the distant region to obtain the discrete cosine transform coefficient matrix of each channel image of the distant region in the frequency domain, construct the mask matrix of each channel image of the distant region, and construct the reflection component of each channel image of the distant region. Step 6: Obtain the near-field color-balanced image and the far-field color-balanced image by RGB channel merging and histogram equalization respectively. Then, fuse the near-field color-balanced image and the far-field color-balanced image by pixel coordinate mapping to obtain the tunnel-enhanced image. The reflection component of each channel image of the near-field region is constructed as follows: The low-frequency component matrix of each channel image of the near region is obtained by performing a matrix dot product between the mask matrix of each channel image of the near region and the discrete cosine transform coefficient matrix of each channel image of the near region. The low-frequency component matrix of each channel image of the near region is obtained by inverse discrete cosine transform to obtain the illumination component of each channel image of the near region. Further, the enhanced illumination components of each channel image of the near region are obtained through logarithmic transformation and Gamma correction. The enhanced illumination component of each channel image of the near region is smoothed by Gaussian filtering to obtain the smoothed illumination component of each channel image of the near region. The smoothed illumination component of each channel image of the near region is calculated by combining the retinal cortex theory to obtain the reflection component of each channel image of the near region. The reflection components of each channel image of the distant region are constructed as follows: The low-frequency component matrix of each channel image of the distant region is obtained by performing a matrix dot product between the mask matrix of each channel image of the distant region and the discrete cosine transform coefficient matrix of each channel image of the distant region. The low-frequency component matrix of each channel image of the distant region is obtained by performing an inverse discrete cosine transform to obtain the illumination component of each channel image of the distant region. Further, the enhanced illumination components of each channel image of the distant region are obtained through logarithmic transformation and Gamma correction. The enhanced illumination components of each channel image of the distant region are smoothed by Gaussian filtering to obtain the smoothed illumination components of each channel image of the distant region. The smoothed illumination component of each channel image of the distant region is calculated by combining the retinal cortex theory, thus obtaining the reflection component of each channel image of the distant region.

2. The method for optimizing low-light image quality in tunnel environments according to claim 1, characterized in that: Step 4 involves constructing the mask matrix for each channel of the near-field image. The specific process is as follows: The width and height of the tunnel image are used as the number of rows and columns of the mask matrix for each channel image of the near region image, respectively. The values ​​of the elements of the mask matrix for each channel of the nearby region image are set by combining the discrete cosine transform coefficient matrix of each channel image in the frequency domain, as follows: The threshold k for separating high-frequency and low-frequency information is determined through statistical analysis and threshold judgment. When row index i in the mask matrix is ​​less than or equal to column index j plus a threshold k, the corresponding position in the mask matrix is ​​set to 1, and the remaining positions in the mask matrix are set to 0.

3. The method for optimizing low-light image quality in tunnel environments according to claim 2, characterized in that: Step 5 involves constructing the mask matrix for each channel of the distant region image. The specific process is as follows: The width and height of the tunnel image are used as the number of rows and columns of the mask matrix for each channel image of the distant region image, respectively. The values ​​of the elements of the mask matrix for each channel of the distant region image are set by combining the discrete cosine transform coefficient matrix of each channel image in the frequency domain of the distant region image, as follows: The threshold k for separating high-frequency and low-frequency information is determined through statistical analysis and threshold judgment. When row index i in the mask matrix is ​​less than or equal to column index j plus a threshold k, the corresponding position in the mask matrix is ​​set to 1, and the remaining positions in the mask matrix are set to 0.

4. The method for optimizing low-light image quality in tunnel environments according to claim 3, characterized in that: Step 6 involves RGB channel merging and histogram equalization to obtain the color-balanced images of the near and far regions, as detailed below: The reflection components of each channel of the near-field image are merged into RGB channels to obtain the near-field enhanced image and the far-field enhanced image. The near-field enhanced image and the far-field enhanced image are then processed by histogram equalization to obtain the near-field color balanced image and the far-field color balanced image.

5. A system for optimizing low-light image quality in tunnel environments, characterized in that, include: The image acquisition and preprocessing module is used to input the tunnel image, normalize each pixel in the tunnel image using a linear normalization method, and obtain the normalized tunnel image. The image segmentation module is used to segment the normalized tunnel image using an adaptive threshold segmentation method to obtain the near region image and the far region image of the normalized tunnel image. The channel separation module is used to perform image separation processing on the near region image of the normalized tunnel image using the RGB channel separation method to obtain a three-channel image of the near region image, and to perform image separation processing on the far region image of the normalized tunnel image using the RGB channel separation method to obtain a three-channel image of the far region image. The near-field region reflection component construction module is used to perform discrete cosine transform on each channel image of the near-field region image to obtain the discrete cosine transform coefficient matrix of each channel image of the near-field region image in the frequency domain, construct the mask matrix of each channel image of the near-field region image, and construct the reflection component of each channel image of the near-field region image. The distant region reflection component construction module is used to perform discrete cosine transform on each channel image of the distant region image to obtain the discrete cosine transform coefficient matrix of each channel image of the distant region image in the frequency domain, construct the mask matrix of each channel image of the distant region image, and construct the reflection component of each channel image of the distant region image. The tunnel enhancement image construction module is used to obtain the near-area color-balanced image and the far-area color-balanced image by RGB channel merging and histogram equalization, respectively. The near-area color-balanced image and the far-area color-balanced image are then fused by pixel coordinate mapping to obtain the tunnel enhancement image. The reflection component of each channel image of the near-field region is constructed as follows: The low-frequency component matrix of each channel image of the near region is obtained by performing a matrix dot product between the mask matrix of each channel image of the near region and the discrete cosine transform coefficient matrix of each channel image of the near region. The low-frequency component matrix of each channel image of the near region is obtained by inverse discrete cosine transform to obtain the illumination component of each channel image of the near region. Further, the enhanced illumination components of each channel image of the near region are obtained through logarithmic transformation and Gamma correction. The enhanced illumination component of each channel image of the near region is smoothed by Gaussian filtering to obtain the smoothed illumination component of each channel image of the near region. The smoothed illumination component of each channel image of the near region is calculated by combining the retinal cortex theory to obtain the reflection component of each channel image of the near region. The reflection components of each channel image of the distant region are constructed as follows: The low-frequency component matrix of each channel image of the distant region is obtained by performing a matrix dot product between the mask matrix of each channel image of the distant region and the discrete cosine transform coefficient matrix of each channel image of the distant region. The low-frequency component matrix of each channel image of the distant region is obtained by performing an inverse discrete cosine transform to obtain the illumination component of each channel image of the distant region. Further, the enhanced illumination components of each channel image of the distant region are obtained through logarithmic transformation and Gamma correction. The enhanced illumination components of each channel image of the distant region are smoothed by Gaussian filtering to obtain the smoothed illumination components of each channel image of the distant region. The smoothed illumination component of each channel image of the distant region is calculated by combining the retinal cortex theory, thus obtaining the reflection component of each channel image of the distant region.

Images