Zero-reference retina image enhancement method, system and device and medium

Abstract

The invention provides a zero-reference retina image enhancement method, system and device and a medium, and the method comprises the steps: obtaining an atmospheric light image component of a to-be-processed retina image in an atmospheric scattering model; adding a random noise map to the grey-scale map of the image to obtain a transmission image component in the model; a to-be-processed image is deblurred, and a first image is generated under the guidance of the transmission image generation module; meanwhile, a to-be-processed image V channel is extracted, the low-light problem is improved by learning an input low-light enhancement curve, and a second image is generated; fusing the first image and the second image to obtain an enhanced image; and finally, calculating a quality degraded image by using an atmospheric scattering model formula in combination with the atmospheric light image, the transmission image and the enhanced image, and finally realizing high-efficiency image enhancement by performing reconstruction loss constraint on the quality degraded image and the to-be-processed image. According to the method, the effect of optimizing various distortions at the same time can be achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, and specifically provides a zero-reference retinal image enhancement method, system, device and medium. BACKGROUND

[0002] Retinal imaging is widely used for early detection and monitoring of retinal diseases such as diabetic retinopathy, glaucoma and macular degeneration. However, due to device limitations, patient factors and environmental factors, the collected retinal images often have various distortions. Low-quality retinal images can lead to misdiagnosis, missed diagnosis and increased medical costs. Therefore, developing an efficient and practical retinal image enhancement method is of great significance for the diagnosis of ophthalmic diseases.

[0003] Traditional retinal image enhancement methods mainly use image transformation and manually labeled features to represent image distortion and repair to achieve quality improvement. However, due to the limited ability of such methods to deal with complex distortions, the enhancement effect is usually poor, which hinders clinical decision-making. In recent years, deep neural networks (DNNs) have attracted more and more attention in the field of image enhancement due to their powerful feature extraction and fusion capabilities. Existing methods can be roughly divided into reference-based methods and zero-reference methods. The former learns the mapping relationship from low-quality images to high-quality images; the latter transfers knowledge through source-target image conversion from low-quality domain to high-quality domain to achieve enhancement, thereby overcoming the dependence on strictly paired images, which leads to time-consuming and expensive clinical data collection. Although excellent in some scenarios, existing DNN-based retinal image enhancement methods still have limitations. First, most methods rely on low-high quality image pairs for model training. However, in clinical practice, it is challenging to collect paired retinal images due to the difficulty in controlling or reproducing complex image degradation. Second, most methods usually only enhance a single type of distortion, which makes it difficult to apply to real scenarios. Generally speaking, images collected in real scenarios often have multiple coupled distortions such as low light, blur and noise.

[0004] Therefore, there is an urgent need for an image enhancement scheme to solve the above problems. SUMMARY

[0005] The application provides a zero-reference retinal image enhancement method, including the following steps: obtaining an atmospheric light map component of a to-be-processed retinal image in an atmospheric scattering model; adding a random noise map to a grayscale map of the to-be-processed retinal image and obtaining a transmittance map component in the atmospheric scattering model; performing deblurring processing on the to-be-processed retinal image first, and then generating a first image under the guidance of a transmittance map generation module; improving the to-be-processed retinal image in a V channel through a low-light enhancement curve input by learning, to generate a second image; fusing the first image and the second image to obtain an enhanced image; inputting the atmospheric light map component of the to-be-processed retinal image in the atmospheric scattering model, the transmittance map component in the atmospheric scattering model and the enhanced image into the atmospheric scattering model to obtain a quality-degraded image. The quality-degraded image and the to-be-processed retinal image are constrained by using a reconstruction loss, so that image enhancement is realized.

[0006] In one of the technical solutions of the zero-reference retinal image enhancement method, the process of adding a random noise map to the grayscale map of the to-be-processed retinal image and obtaining a transmittance map component in the atmospheric scattering model includes: adding the grayscale map of the to-be-processed retinal image and the random noise map to obtain an initial input map; using the initial input map and its 2-fold, 4-fold and 8-fold down-sampling maps as inputs, extracting features from each down-sampling map by using a CONVB module to generate feature maps; for the initial feature map in the feature maps, considering the lack of high-level features, no fusion and CONVB processing are performed; splicing all the feature maps except the initial feature map and the 2-fold down-sampling of adjacent high-level features along the channel direction, and inputting the spliced features into the CONVB module to obtain guided repair features

[0007] In one of the technical solutions of the zero-reference retinal image enhancement method, the process of adding a random noise map to the grayscale map of the to-be-processed retinal image and obtaining a transmittance map component in the atmospheric scattering model includes: adding the grayscale map of the to-be-processed retinal image and the random noise map to obtain an initial input map; using the initial input map and its 2-fold, 4-fold and 8-fold down-sampling maps as inputs, extracting features from each down-sampling map by using a CONVB module to generate feature maps; for the initial feature map in the feature maps, considering the lack of high-level features, no fusion and CONVB processing are performed; splicing all the feature maps except the initial feature map and the 2-fold down-sampling of adjacent high-level features along the channel direction, and inputting the spliced features into the CONVB module to obtain guided repair features performing 2-fold up-sampling and inputting the up-sampled features into the CONVB module, and obtaining features by multiplying the features obtained by the CONVB module and the initial feature map performing element-by-element multiplication, wherein Sigmoid(·) represents a Sigmoid function, performing 2-fold up-sampling on the product and passing the up-sampled features through the CONVB module, and obtaining features by multiplying the features obtained by the CONVB module and the initial feature map performing element-by-element multiplication, wherein Sigmoid(·) represents a Sigmoid function, performing 2-fold up-sampling on the product and passing the up-sampled features through the CONVB module, and obtaining features by multiplying the features obtained by the CONVB module and the initial feature map

[0008] In one technical solution of the aforementioned zero-reference retinal image enhancement method, the process of first deblurring the retinal image to be processed and then generating a first image under the guidance of the transillumination generation module includes: inputting the retinal image to be processed into the deblurring module for deblurring, the deblurring module comprising four encoding stages and four decoding stages; each encoding stage consists of a CONVB module, an attention-based feature integration module, and a downsampling operation, the attention-based feature integration module being used to guide the model to obtain detailed information from the guidance restoration features; each decoding stage contains four stages, each stage consisting of an upsampling operation, skip connection concatenation, and the CONVB module, in each decoder stage, the feature map of the previous stage is upsampled, concatenated with the features of the corresponding encoder stage, and then refined by the CONVB module, after four progressive decoder stages, generating a deblurred image E1.

[0009] In one technical solution of the aforementioned zero-reference retinal image enhancement method, the attention-based feature integration module includes: a spatial attention-based AFI module and a channel attention-based AFI module; the spatial attention-based AFI module integrates feature maps... A spatial attention map is generated by applying average and max pooling operations along the channel dimension. The resulting features are concatenated and then processed by a 7×7 convolution and a sigmoid function. The generated spatial attention map is then used for adaptive recalibration of the CONVB output results X1 / X2. The channel attention-based AFI module includes two parallel average and max pooling operations, followed by a multilayer perceptron to analyze the feature map. The multilayer perceptron consists of two 1×1 convolutions and a LeakyReLU activation function. The outputs of these two parallel branches are summed element-wise and then used by the Sigmoid function to generate a channel attention map. The multiplication operation is used to recalibrate the output results X3 / X4 of CONVB.

[0010] In one technical solution of the aforementioned zero-reference retinal image enhancement method, the V channel of the retinal image I(x) to be processed is used as the input of the low-light enhancement module. After passing through the first CONVB module, v1 is obtained. v1 then passes through the second CONVB module to obtain v2. Subsequently, v1 and v2 are concatenated along the channel direction and input into the third CONVB module for dimensionality reduction to obtain v3. At this point, v3 and the previous v2 are concatenated along the channel direction and then sent to the fourth CONVB module for dimensionality reduction to obtain v4. v4 is used as the pixel-by-pixel trainable parameter map of the higher-order curve. After n iterations, I is obtained. V (x) Enhanced version V e (x); V e The H and S channels of I(x) are concatenated with those of I(x), and then converted to RGB space to obtain the illumination-enhanced image E2.

[0011] In one of the technical solutions of the zero-reference retinal image enhancement method described above, the process of fusing the first image and the second image to obtain the enhanced image includes: splicing the deblurring image E1 and the illumination enhancement image E2 along the channel direction, and processing the spliced features through the CONVB module; the obtained features are processed in parallel through the average pooling operation and the maximum pooling operation; the obtained features are spliced along the channel direction and processed through the convolution layer; the obtained features are normalized using the Sigmoid function to generate weights; the weights are multiplied with E1 element by element to generate an enhanced image.

[0012] In a second aspect, the present application discloses a zero-reference retinal image enhancement system, comprising: a first acquisition module for acquiring an atmospheric light map component of a retinal image to be processed in an atmospheric scattering model; a second acquisition module for adding a random noise map to a grayscale map of the retinal image to be processed and acquiring a transmission map component in the atmospheric scattering model; a first processing module for deblurring the retinal image to be processed first, and then generating a first image under the guidance of a transmission map generation module; a second processing module for improving the image through a low-light enhancement curve input by learning the V channel of the retinal image to be processed, to generate a second image; a fusion module for fusing the first image and the second image to obtain an enhanced image; and a processing module for inputting the atmospheric light map component in the atmospheric scattering model, the transmission map component in the atmospheric scattering model, and the enhanced image into the atmospheric scattering model to obtain a quality-degraded image. The quality-degraded image and the retinal image to be processed are constrained by a reconstruction loss to achieve image enhancement.

[0013] In a third aspect, the present application discloses an electronic device comprising a processor and a storage device, the storage device being adapted to store a plurality of program codes, characterized in that the program codes are adapted to be loaded and run by the processor to execute the zero-reference retinal image enhancement method of any one of the above aspects.

[0014] In a fourth aspect, the present application discloses a computer-readable storage medium, wherein a plurality of program codes are stored, characterized in that the program codes are adapted to be loaded and run by the processor to execute the zero-reference retinal image enhancement method of any one of the above aspects.

[0015] The zero-reference retinal image enhancement method has the beneficial effects that: on one hand, the atmospheric light map and the transmission map components are accurately decomposed based on an atmospheric scattering model, a first image is generated by combining a deblurring process and a transmission map, the scattering interference can be effectively eliminated, the fine structure can be restored, a second image is generated through a learning mechanism of a V-channel low-light enhancement curve, the low-light area brightness is improved, and after the two are fused, the image detail recognition and contrast can be comprehensively improved; on one hand, the transmission map is obtained by introducing random noise, the robustness of the model to interference is improved, the adaptive strategy of low-light enhancement avoids overexposure or insufficient enhancement caused by fixed parameters, the algorithm can stably cope with complex scenes such as noise and uneven illumination, and the effects of simultaneously optimizing multiple distortions are realized. BRIEF DESCRIPTION OF DRAWINGS

[0016] The disclosure of the present application will become more apparent from the following description in conjunction with the accompanying drawings. It is readily understood by those skilled in the art that the drawings are merely for the purpose of illustration and are not intended to limit the scope of protection of the present application. In addition, similar numbers in the figures represent similar components, wherein:

[0017] Figure 1 is a network framework diagram according to an embodiment of the present application;

[0018] Figure 2 is a transmission map generation module structure diagram according to an embodiment of the present application;

[0019] Figure 3 is a structure diagram of a deblurring module according to an embodiment of the present application. DETAILED DESCRIPTION

[0020] Some embodiments of the present application will be described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present application, and are not intended to limit the scope of protection of the present application.

[0021] As shown in Figures 1-3 , the present application discloses a zero-reference retinal image enhancement method, comprising the following steps:

[0022] Step S1, obtaining an atmospheric light map component of a to-be-processed retinal image in an atmospheric scattering model;

[0023] Step S2, adding a random noise map to a gray map of the to-be-processed retinal image, and obtaining a transmission map component in the atmospheric scattering model;

[0024] Step S3, processing the to-be-processed retinal image through a deblurring process first, and then generating a first image under the guidance of a transmission map generation module;

[0025] Step S4: Improve the low-light performance of the V channel of the retinal image to be processed by learning the low-light enhancement curve, and generate a second image;

[0026] Step S5: Fuse the first image and the second image to obtain the enhanced image;

[0027] Step S6: Input the atmospheric light map component of the retinal image to be processed in the atmospheric scattering model, the transmission map component of the atmospheric scattering model, and the enhanced image into the atmospheric scattering model to obtain the image with degraded quality. Constrain the image with degraded quality and the retinal image to be processed using reconstruction loss to achieve image enhancement.

[0028] In this embodiment, the following specific methods were adopted to implement the above method: Figure 1 The network framework shown mainly consists of a Transmission Map Generation (TMG) module, a Deblurring (DB) module, a Low-Light Enhancement (LLE) module, and a fusion module. This network architecture primarily includes three parallel branches. Assuming the retinal image to be processed is used as the network's input image, in the upper branch, it is input into the Dark Channel Prior (DCP) module to obtain the atmospheric light map component as part of the atmospheric scattering model. In the middle branch, the grayscale image of the retinal image I(x) to be processed is processed using a method with parameters 1-I... gray The form of I(x) is added to the random noise map and used as the input to the TMG to obtain the transmission map component t(x) in the atmospheric scattering model. In the next branch, I(x) is first processed by the deblurring module DB, and then the deblurring result E1 is generated under the guidance of the transmission map generation module TMG. At the same time, in order to achieve low-light enhancement of the retinal image, the V channel of I(x) (denoted as I) is... V (x) is used as input to LLE. The LLE module improves the low-light performance of the image by learning the low-light enhancement curve of the input, generating the illumination enhancement result E2. Finally, in order to make full use of the brightness information in the low-light enhanced image E2, E1 and E2 are combined by the fusion module to output the enhanced image E.

[0029] The atmospheric light map component A(x) of the retinal image to be processed in the atmospheric scattering model, the transmission map component t(x) of the atmospheric scattering model, and the enhanced image E are input into the atmospheric scattering model to obtain the image after quality degradation.

[0030] In one embodiment, the transmittance map generation module is used to generate a transmittance map. Since grayscale images are adept at capturing key structural details, this module uses the grayscale value of I(x). Figure 1 -I gray (x) is used as input. Furthermore, to enhance the network's learning ability and robustness to noise, a Gaussian random noise map is added to the grayscale image to generate I. gn .

[0031] like Figure 2 As shown, the TMG module contains four parallel branches, each with a corresponding I... gn The input is then sampled at 2x, 4x, and 8x, and the CONVB module is used to extract features from the input to generate feature maps g. k , k∈{1,2,3,4}. To achieve feature fusion and complementarity of different sizes, the feature map g is... k 2x downsampled output of adjacent high-rise features The features are stitched together along the channel direction, and the stitched features are input into the CONVB block to obtain...

[0032] A CONVB block consists of two 3×3 convolutions, a LeadReLU, and a residual connection implemented by a 1×1 convolution.

[0033] In the first parallel branch, g1 is directly taken as... Without going through the CONVB block operation. Afterwards, After performing a 2x upsampling, the input to the CONVB block yields features that are... Perform element-wise multiplication, where Sigmoid(·) represents the Sigmoid function. Upsample the product by a factor of 2 and pass it through a CONVB block. The resulting feature is then multiplied by... Multiplication. By performing cross-feature decoding operations step by step, refined features are finally obtained. These features are then input into the CONVB module, and a Sigmoid operation is performed again to generate the transmittance map t(x).

[0034] In one embodiment, the DB module is used to reconstruct the enhanced image after blur restoration, and its structure is as follows: Figure 3 As shown, the DB module adopts a U-shaped architecture, comprising four encoding stages and four decoding stages. Each encoding stage consists of a CONVB module, an attention-based feature integration (AFI) module, and a downsampling operation. The TMG mentioned above generates multi-scale features. Containing a wealth of contextual information, the AFI module can be used to guide the model from... Get more details.

[0035] Considering that shallow layers have a large spatial dimension while deep layers have a large channel dimension, the first two encoding stages use spatial attention-based AFI (referred to as AFI / S), while the latter two encoding stages use channel attention-based AFI (referred to as AFI / C). AFI / S works by analyzing the feature maps... The average and max pooling operations are applied along the channel dimension to generate the spatial attention map. The resulting feature is concatenated and processed by a 7x7 convolution and Sigmoid function to generate the spatial attention map, which is then used to adaptively recalibrate the CONVB output X1 / X2. The AFI / C block focuses on the inter-channel relationship, and two parallel average and max pooling operations are followed by a multi-layer perceptron (MLP) to more comprehensively analyze the feature map This part of the MLP consists of two 1x1 convolutions and LeakyReLU activation functions. The output results of the two parallel branches are summed element-wise and passed through a Sigmoid function to generate a channel attention map, which is used to recalibrate the output X3 / X4 of the CONVB through a multiplication operation.

[0036] Taking the kth stage as an example, the CONVB output X k Before downsampling, the G k Fusion is performed in the attention block. In contrast to the encoder, the decoder also includes four stages, each consisting of an upsampling operation, a jump connection concatenation, and a CONVB block. In each decoder stage, the feature map of the previous stage is upsampled, concatenated with the corresponding encoder stage feature, and then refined by the CONVB block. After four progressive decoder stages, the final deblurred image E1 is generated.

[0037] In one embodiment, a low-light enhancement module (LLE) is used to address the problem of uneven illumination and insufficient brightness in the retinal image. This module only acts on the V channel in the HSV space, performs brightness correction, thereby avoiding the generation of artifacts such as color bias or over-saturation.

[0038] First, the V channel of the retinal image I(x) to be processed is taken as the input of the low-light enhancement module, and v1 is obtained after passing through the first CONVB module. v1 is then input into the second CONVB module to obtain v2. Subsequently, v1 and v2 are concatenated along the channel direction and input into the third CONVB module for dimension reduction to obtain v3. At this time, v3 and v2 are concatenated along the channel direction and input into the fourth CONVB module for dimension reduction to obtain v4. v4 is taken as the pixel-wise trainable parameter map of the high-order curve, and after n iterations, the enhanced version V V (x) of I(x) is obtained. e (x). Finally, V e (x) is concatenated with the H and S channels of I(x), and then converted to the RGB space to obtain the illumination-enhanced image E2.

[0039] In one embodiment, after obtaining two intermediate results E1 and E2, they are integrated by a fusion module to generate the final enhanced image E. Specifically, first, the fusion module splices the deblurring image E1 and the illumination enhanced image E2 along the channel direction, and processes the spliced features through a CONVB block to halve the channel dimension. Subsequently, the obtained features are simultaneously processed through average pooling operation and maximum pooling operation in parallel. The obtained features are then spliced along the channel direction and processed through a 7x7 convolution layer. Thereafter, the obtained features are normalized using a Sigmoid function to generate weights. Finally, the weights are multiplied with E1 element by element to generate the enhanced image.

[0040] A plurality of loss functions are used for joint optimization, mainly composed of two parts of a loss function for implementing the low-light enhancement part and a loss function for implementing the deblurring part. The loss for low-light enhancement includes an illumination smoothing loss and an exposure control loss The loss for the deblurring part has a smoothing loss a spatial consistency loss a color constancy loss and a loss.

[0041] To verify the method, the first data set required for application was constructed by performing quality degradation operation on the retinal images in the large-scale public data set EyeQ, named EyeDe1, and then, 2000 images with quality rating of "good" were randomly selected from EyeQ as high-quality images, only using two degradation models of illumination distortion and blur distortion and their combination to synthesize 6,000 low-quality images, and changing the decay parameter value in the original degradation model to make the synthesized low-quality images visually present darker illumination and more serious blur, to obtain the second data set used EyeDe2. For convenience of experiment, the present application uniformly adjusts all pictures to 512x512 size in network training experiment, and the data in the above data sets EyeDe1 and EyeDe2 are divided into training set and test set in the ratio of 8:2 for model training and testing.

[0042] The network model of the present application is implemented under the PyTorch deep learning framework and runs on the Ubuntu 18.04 operating system. The network model uses the Adam optimizer with a learning rate of 1x10-4 and an epoch setting of 200.

[0043] Embodiment Two

[0044] The application further provides a zero-reference retinal image enhancement system, comprising: a first acquisition module, configured to acquire an atmospheric light map component of a to-be-processed retinal image in an atmospheric scattering model; a second acquisition module, configured to add a random noise map to a grayscale map of the to-be-processed retinal image and acquire a transmission map component in the atmospheric scattering model; a first processing module, configured to first perform deblurring processing on the to-be-processed retinal image, and then generate a first image under the guidance of a transmission map generation module; a second processing module, configured to improve the to-be-processed retinal image in a low-light environment by inputting a learned low-light enhancement curve to the image, and generate a second image; a fusion module, configured to fuse the first image and the second image to obtain an enhanced image; and a processing module, configured to input the atmospheric light map component of the to-be-processed retinal image in the atmospheric scattering model, the transmission map component in the atmospheric scattering model and the enhanced image into the atmospheric scattering model to obtain a quality-degraded image. The quality-degraded image and the to-be-processed retinal image are constrained by using a reconstruction loss, so that the image enhancement is realized.

[0045] Embodiment three

[0046] The application further provides an electronic device. In an embodiment of the device according to the application, the device comprises a processor and a storage device, the storage device can be configured to store a program for implementing the zero-reference retinal image enhancement method of the above-mentioned method embodiments, and the processor can be configured to execute the program in the storage device, which includes but is not limited to the program for implementing the zero-reference retinal image enhancement method of the above-mentioned method embodiments. For the convenience of illustration, only the parts related to the embodiments of the application are shown, and the specific technical details not disclosed are referred to the method part of the embodiments of the application. The control device can be a control device formed by various electronic devices.

[0047] Embodiment four

[0048] The application further provides a computer-readable storage medium. In an embodiment of the computer-readable storage medium according to the application, the computer-readable storage medium can be configured to store a program for implementing the zero-reference retinal image enhancement method of the above-mentioned method embodiments, which can be loaded and run by a processor to implement the above-mentioned zero-reference retinal image enhancement method. For the convenience of illustration, only the parts related to the embodiments of the application are shown, and the specific technical details not disclosed are referred to the method part of the embodiments of the application. The computer-readable storage medium can be a storage device formed by various electronic devices, and optionally, the computer-readable storage medium in the embodiments of the application is a non-transitory computer-readable storage medium.

[0049] The technical scheme of the present application has been described in combination with the preferred embodiments shown in the drawings, but it is easily understood by those skilled in the art that the protection scope of the present application is obviously not limited to these specific embodiments. Those skilled in the art can make equivalent changes or replacements to the original technical features without departing from the principles of the present application, and the technical schemes after these changes or replacements will all fall within the protection scope of the present application.

Claims

1. A zero-reference retinal image enhancement method, characterized in that, Includes the following steps: Obtain the atmospheric light map components of the retinal image to be processed in the atmospheric scattering model; Random noise is added to the grayscale image of the retinal image to be processed, and the transmission map component in the atmospheric scattering model is obtained. The retinal image to be processed is first deblurred, and then the first image is generated under the guidance of the transmission image generation module. The V channel of the retinal image to be processed is used to improve the low light of the image by learning the low light enhancement curve of the input, and a second image is generated. The first image and the second image are fused together to obtain the enhanced image; The atmospheric light map component of the retinal image to be processed in the atmospheric scattering model, the transmission map component of the atmospheric scattering model, and the enhanced image are input into the atmospheric scattering model to obtain the image after quality degradation. Image enhancement is achieved by constraining the degraded image and the retinal image to be processed using a reconstruction loss.

2. The method according to claim 1, characterized in that, The process of adding a random noise map to the grayscale image of the retinal image to be processed and obtaining the transmittance components in the atmospheric scattering model includes: The grayscale image of the retinal image to be processed is added to the random noise image to obtain the initial input image; The initial input image and its 2x, 4x, and 8x downsampled images are used as inputs, and the CONVB module is used to extract features from each downsampled image to generate a feature map. For the initial feature map, considering the lack of high-level features, no fusion or CONVB processing is performed. Instead, all feature maps except the initial feature map are concatenated along the channel direction with a 2x downsampled version of the adjacent high-level features. The concatenated features are then input into the CONVB module to obtain features for guidance in feature repair.

3. The method according to claim 2, characterized in that, The process of adding a random noise map to the grayscale image of the retinal image to be processed and obtaining the transmittance component in the atmospheric scattering model further includes: After performing a 2x upsampling, the input is given to the CONVB module, and the resulting features are compared with... Element-wise multiplication is performed, where Sigmoid(·) represents the Sigmoid function. The product result is upsampled by a factor of 2 and passed through the CONVB module. The resulting feature is then multiplied by... Multiplication is performed step by step across feature decoding operations to obtain refined features. The refined features are then input into the CONVB module and a Sigmoid operation is performed again to generate a transmittance map.

4. The method according to claim 2, characterized in that, The process of first deblurring the retinal image to be processed, and then generating the first image under the guidance of the transillumination generation module, includes: The retinal image to be processed is input into the deblurring module for deblurring. The deblurring module includes four encoding stages and four decoding stages. Each encoding stage consists of a CONVB module, an attention-based feature integration module, and a downsampling operation. The attention-based feature integration module is used to guide the model to obtain detailed information from the insulated features. Each decoding stage contains four stages, each consisting of an upsampling operation, a skip connection concatenation, and a CONVB module. In each decoder stage, the feature map of the previous stage is upsampled, concatenated with the features of the corresponding encoder stage, and then refined through the CONVB module. After four progressive decoder stages, a deblurred image E1 is generated.

5. The method according to claim 3, characterized in that, The attention-based feature integration module includes: an AFI module based on spatial attention and an AFI module based on channel attention; The spatial attention-based AFI module uses the feature map to be processed. A spatial attention map is generated by applying average and max pooling operations along the channel dimension. The resulting features are concatenated and then processed by 7×7 convolution and the sigmoid function. The generated spatial attention map is then used to adaptively recalibrate the CONVB output results X1 / X2. The channel attention-based AFI module consists of two parallel average and max pooling operations, followed by a multilayer perceptron to analyze the feature map. The multilayer perceptron consists of two 1×1 convolutions and a LeakyReLU activation function. The outputs of these two parallel branches are summed element-wise and then used by the Sigmoid function to generate a channel attention map. The multiplication operation is used to recalibrate the output results X3 / X4 of CONVB.

6. The method according to claim 4, characterized in that, The process of improving the low-light performance of the retinal image by learning the low-light enhancement curve from the input image to generate a second image includes: The V channel of the retinal image I(x) to be processed is used as the input of the low light enhancement module. After passing through the first CONVB module, v1 is obtained. v1 is then passed through the second CONVB module to obtain v2. Then, v1 and v2 are stitched together along the channel direction and input into the third CONVB module for dimensionality reduction to obtain v3. At this time, v3 and the previous v2 are stitched together along the channel direction and then sent into the fourth CONVB module for dimensionality reduction to obtain v4. Using v4 as a pixel-by-pixel trainable parameter map of a higher-order curve, after n iterations, I is obtained. V (x) Enhanced version V e (x); V e The H and S channels of I(x) are concatenated with those of I(x), and then converted to RGB space to obtain the illumination-enhanced image E2.

7. The method according to claim 6, characterized in that, The process of fusing the first image and the second image to obtain the enhanced image includes: The deblurred image E1 and the illumination-enhanced image E2 are stitched together along the channel direction, and the features of the stitched image are processed by the CONVB module. The resulting features are processed in parallel by both average pooling and max pooling operations. The acquired features are concatenated along the channel direction and then processed through a convolutional layer. The obtained features are normalized using the Sigmoid function to generate weights; The weights are multiplied element-wise by E1 to generate an enhanced image.

8. A zero-reference retinal image enhancement system, characterized in that, include: The first acquisition module is used to acquire the atmospheric light map components of the retinal image to be processed in the atmospheric scattering model; The second acquisition module is used to add a random noise map to the grayscale map of the retinal image to be processed, and to acquire the transmission map component in the atmospheric scattering model; The first processing module is used to first deblur the retinal image to be processed, and then generate the first image under the guidance of the transmission image generation module; The second processing module is used to improve the low-light performance of the V channel of the retinal image to be processed by learning the low-light enhancement curve input, and generate a second image. The fusion module is used to fuse the first image and the second image to obtain an enhanced image; The processing module is used to input the atmospheric light map component of the retinal image to be processed in the atmospheric scattering model, the transmission map component of the atmospheric scattering model, and the enhanced image into the atmospheric scattering model to obtain the image after quality degradation; and to use the reconstruction loss to constrain the image after quality degradation and the retinal image to be processed in order to achieve image enhancement.

9. An electronic device comprising a processor and a storage device, the storage device being adapted to store multiple lines of program code, characterized in that, The program code is adapted to be loaded and run by a processor to perform the zero-reference retinal image enhancement method of any one of claims 1 to 7.

10. A computer-readable storage medium storing a plurality of program codes, characterized in that, The program code is adapted to be loaded and run by a processor to perform the zero-reference retinal image enhancement method of any one of claims 1 to 7.

Images