Low-light face super-resolution method and system based on Retinex and diffusion model

Abstract

The invention discloses a low-light face super-division method and system based on Retinex and a diffusion model, and the method comprises the steps: constructing a low-light face super-division network model based on Retinex and the diffusion model, the model comprises a noise image estimation module and a denoising diffusion module, and carrying out the bicubic interpolation of an input face image, and obtaining an up-sampling image; a face image with noise is obtained through decomposition of a noise image estimation module; the face image with noise is fused with pure Gaussian noise, and a noise-eliminated result is obtained through a denoising diffusion module; by minimizing diffusion loss, training of a low-light face super-resolution model is completed; and testing the acquired face image test sample by using the trained low-light face super-resolution network model to obtain a super-resolution result, and evaluating the super-resolution result. According to the method, a low-light low-resolution face image can be better recovered, the effectiveness of the method is shown through wide experiments, and the method is applied to a low-light face super-resolution task.

Description

Technical Field

[0001] The present invention relates to a low-light face super-resolution technology, and in particular to a low-light face super-resolution method and system based on Retinex and diffusion models. Background Art

[0002] Face super-resolution is a long-standing and highly anticipated topic in the field of computer vision. In recent years, with the help of the leap in deep learning technology and the vigorous development of large-scale data sets, it has achieved remarkable results and shined in many practical application scenarios. However, in the specific field of face super-resolution, the problem of face super-resolution in low-light environments, as an emerging topic, has not been fully studied. At present, solutions for low-light face super-resolution are mainly divided into two categories: algorithms based on task splitting and algorithms designed specifically for this task. The former decomposes low-light face super-resolution into two independent steps, low-light enhancement and face super-resolution, which are executed sequentially to restore the image; the latter is committed to building a new model directly for low-light face super-resolution.

[0003] Unfortunately, both existing methods face challenges when dealing with low-light face super-resolution. The algorithm based on task splitting decomposes the original problem into two sub-tasks, which often leads to the problem of mutual constraints between the two, affecting the overall effect; while the algorithms designed specifically for low-light face super-resolution generally face the dilemma of high model complexity and high training difficulty. In view of this, there is an urgent need to develop a new low-light face super-resolution method that is both specially customized and easy to train to more effectively address this challenge. Summary of the invention

[0004] Purpose of the invention: One purpose of the present invention is to provide a low-light face super-resolution method based on Retinex and diffusion model, which can better restore low-light and low-resolution face images.

[0005] Another object of the present invention is to provide a low-light face super-resolution system based on Retinex and diffusion model.

[0006] Technical solution: The low-light face super-resolution method based on Retinex and diffusion model of the present invention comprises the following steps:

[0007] Construct a low-light face super-resolution network model based on Retinex and diffusion models, and use the face image dataset for training; Use the trained low-light face super-resolution network model to test the collected face image test samples, obtain the super-resolution results, and use evaluation indicators to evaluate;

[0008] The low-light face super-resolution network model includes: a noise image estimation module and a denoising diffusion module. The denoising diffusion module is a U-Net architecture, including a coding layer, a bottleneck layer, and a decoding layer. The noise image estimation module performs a Retinex transform on the model input image to obtain a noisy face image R noise ; R obtained by the noise image estimation module noise After being concatenated with pure Gaussian noise, it is processed through three consecutive coding layers to obtain the coding layer output image I E ; Coding layer output image I E Further information integration is performed at the bottleneck layer, and the integrated information is sequentially passed through three decoding layers for information integration and image restoration; then the convolution layer outputs a noise-free high-resolution face output image I HH ; The entire network model is trained using diffusion loss.

[0009] Furthermore, before using the face image dataset to train the low-light face super-resolution network model, it is necessary to normalize the face image data in the face image dataset and obtain the upsampled image I by cubic interpolation. LH .

[0010] Furthermore, the input image is retinex transformed through the noise image estimation module to obtain the noisy face image R noise , including: first, perform three-channel averaging on the input image to obtain a color mean image I mean ; Then the input image and the color mean image I mean Splicing is performed in the channel dimension, and the brightness inverse component is obtained by sequentially passing through the convolution layer, the DW convolution layer and the convolution layer. Finally, according to the Retinex theory, with I LH Multiply to get the noisy face image R noise ; expressed as:

[0011] I mean =F(I LH )

[0012]

[0013]

[0014] Among them, the function F means taking the average on three channels, the function Con means splicing in the channel dimension, Conv means ordinary convolution, and DWConv means DW convolution.

[0015] Furthermore, the encoding layer includes a face feature extraction block and a downsampling module, which is expressed as:

[0016]

[0017] in, Represents the input of the encoding layer. For the first encoding layer, Represents a noisy face image R noise , for the second and third coding layers, represents the output of the previous encoding layer, represents the output of the i-th encoding layer, FFER represents the face feature extraction block, and downSample represents the downsampling block.

[0018] The facial feature extraction block includes two normalization layers, a global facial structure information extraction unit, and a gated convolutional layer that replaces the fully connected layer. The global facial structure information extraction unit uses linear layers, convolutional layers, and hourglass networks to implement the attention algorithm to extract global facial structure information, which can be expressed as:

[0019]

[0020]

[0021]

[0022] Among them, Linear represents the linear layer, HourGlass represents the hourglass network, Sigmod represents the activation function, I att represents the calculated attention feature, I filter represents the calculated attention weight, I G It represents the output result of the global face structure information unit;

[0023] GFSR is used to represent the global face structure information extraction unit, and the calculation process of the entire face feature extraction block is simplified to:

[0024]

[0025]

[0026] Among them, GateConv represents gated convolution, I noise represents a pure Gaussian noise, and LN represents a normalization layer.

[0027] Furthermore, the bottleneck layer includes a global face structure information extraction unit, which is expressed as:

[0028] I B =GFSR(I E )

[0029] Among them, I B represents the integration result of the bottleneck layer, I ERepresents the output of the last encoding layer.

[0030] Furthermore, the decoding layer corresponds to the encoding layer. Each decoding layer includes an upsampling block and a face feature extraction block. The decoding layer not only receives the output from the previous decoding layer, but also receives the output of the encoding layer corresponding to its scale. The calculation process is:

[0031]

[0032] Among them, upSample represents the upsampling block, Represents the input of the decoding layer. For the first decoding layer, Represents the integration result of the bottleneck layer I B , for the second and third decoding layers, represents the output of the previous decoding layer, represents the output of the i-th decoding layer, represents the output of the i-th encoding layer.

[0033] Finally, a convolutional layer is used to transform the output of the last decoding layer into the final result, which is expressed as follows:

[0034] I HH =Conv(I D )

[0035] Among them, I HH represents the noise-free high-resolution face output image under normal lighting, I D Result graph representing the output of the last decoding layer.

[0036] Furthermore, when calculating the diffusion loss, the Gaussian noise added to the output result is estimated using the inverse diffusion transform, and then the L1 Loss is calculated with the true value of the Gaussian noise to constrain the model training, which is expressed as:

[0037] I ε =INV(I HH )

[0038] Loss diff =||I ε -I noise ||1

[0039] Where INV represents the inverse diffusion transformation, I ε Represents the estimated value of Gaussian noise, Loss diff represents diffusion loss.

[0040] Furthermore, the evaluation indicators include peak signal-to-noise ratio PSNR and structural similarity SSIM;

[0041] The peak signal-to-noise ratio (PSNR) is calculated as follows:

[0042]

[0043]

[0044] Where MSE stands for mean square error, n stands for the total number of test samples, and x i and i They respectively represent the face images I HH and the real image I GT Pixels, MAX I Indicates the maximum value of the color of an image point;

[0045] The calculation method of structural similarity SSIM is:

[0046] c1=(k1L) 2

[0047] c2=(k2L) 2

[0048]

[0049] in, isI HH The average value of isI GT The average value of isI HH The variance of isI GT variance, L is the dynamic range of pixel values, k1 and k2 are parameters, and c1 and c2 represent intermediate variables.

[0050] The system corresponding to the method includes:

[0051] Model building and training unit, used to build a low-light face super-resolution network model based on Retinex and diffusion models, and train it using face image datasets;

[0052] and a test evaluation unit, which is used to test the collected face image test samples using the trained low-light face super-resolution network model, obtain the super-resolution results, and evaluate them using evaluation indicators;

[0053] The low-light face super-resolution network model includes: a noise image estimation module and a denoising diffusion module. The denoising diffusion module is a U-Net architecture, including a coding layer, a bottleneck layer, and a decoding layer. The noise image estimation module performs a Retinex transform on the model input image to obtain a noisy face image R noise ; R obtained by the noise image estimation module noiseAfter being concatenated with pure Gaussian noise, the image is processed through three consecutive coding layers to obtain image I E ; Coding layer output image I E Further information integration is performed at the bottleneck layer, and the integrated information is sequentially passed through three decoding layers for information integration and image restoration; then the convolution layer outputs a high-resolution output image I without noise. HH ; The entire network model is trained using diffusion loss.

[0054] A device for storing and executing the method comprises a memory and a processor, wherein:

[0055] A memory for storing computer programs that can be run on the processor;

[0056] The processor is used to execute the steps of the low-light face super-resolution method based on Retinex and diffusion model when running the computer program.

[0057] Beneficial effect: Compared with the prior art, the significant technical effect of the present invention is as follows: by introducing the Retinex theory and the diffusion model, the performance of the face super-resolution algorithm in a low-light environment is significantly improved. Specifically, the noise image estimation module is first used to perform Retinex transformation on the input image, and then the denoising diffusion module based on the U-Net architecture is used to denoise the above-mentioned noisy face image; the present invention cleverly integrates the Retinex theory to decompose the face image, and combines the diffusion model to repair the face image, and a series of extensive experimental verifications have proved the effectiveness and practicality of the method; in particular, the test results on the real face super-resolution dataset show that the method proposed in the present invention exhibits excellent performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 is a flow chart of the method of the present invention;

[0059] Figure 2 Flowchart for training low-light face super-resolution network model;

[0060] Figure 3 This is the framework diagram of the low-light face super-resolution network based on Retinex and diffusion model;

[0061] Figure 4 It is a structural diagram of a noise image estimation module of the present invention;

[0062] Figure 5 It is a structural diagram of a face feature extraction block of the present invention;

[0063] Figure 6 It is a structural diagram of the global face structure information extraction unit of the present invention. DETAILED DESCRIPTION

[0064] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0065] The present invention provides a low-light face super-resolution method based on Retinex and diffusion model. Figure 1 As shown in the figure, a low-light face super-resolution network model based on Retinex and diffusion model is first constructed, and then trained using face image datasets; finally, the trained low-light face super-resolution network model is used to test the collected face image test samples to obtain super-resolution results, which are then evaluated using evaluation indicators. The low-light face super-resolution network model includes two main parts: a noise image estimation module and a denoising diffusion module. First, the noise image estimation module performs a Retinex transform on the input face image to obtain a noisy face image R. noise ; Secondly, the denoising diffusion module of the U-Net architecture is used to eliminate the noise and obtain a high-resolution face output image I without noise HH The entire low-light face super-resolution network model is trained using diffusion loss.

[0066] like Figures 1 to 3 As shown, the specific steps include:

[0067] S1. Use data normalization method to input face image I LL Perform range constraints and obtain the upsampled image I through bicubic interpolation LH ;

[0068] The details are as follows: First, the maximum and minimum normalization method is used to control the data range in the face image dataset between 0 and 1. Then, the input face image I is interpolated using bicubic interpolation. LL Enlarge to the target size to get the upsampled image I LH .

[0069] S2, using the noise image estimation module and Retinex theory to upsample the image I LH Decompose and get the noisy face image R noise ;

[0070] The details are as follows: In the noise image estimation module, the upsampled image I LH You need to first take the average value to get the color mean map I mean , and then concatenate the two and estimate the brightness inverse component through three layers of convolution Finally, the noisy face image R is obtained by dot multiplication. noise , the process is as follows:

[0071] I mean =F(I LH )

[0072]

[0073]

[0074] Among them, the function F means taking the average on three channels, the function Con means splicing in the channel dimension, Conv means ordinary convolution, and DWConv means DW convolution.

[0075] like Figure 3 and Figure 4 As shown, in this embodiment, the upsampled image I LH It will first be sent to the noise image estimation module, and then in the noise image estimation module, a color mean image I will be obtained by averaging the three channels. mean , then I LH and I mean Splicing is performed on the channel dimension, and the convolution layer, DW convolution layer and convolution layer are obtained in turn. Finally, according to the Retinex theory, with I LH Multiply to get the noisy face image R noise .

[0076] S3, denoising diffusion module for the noisy face image R in the previous step noise De-noising is performed and the final super-resolution result is obtained, that is, a high-resolution output image without noise.

[0077] The details are as follows: The denoising diffusion module is mainly composed of three parts: the encoding layer, the bottleneck layer and the decoding layer. noise It will first be concatenated with a pure Gaussian noise, and then processed by three consecutive coding layers to obtain the coding layer result I E Each encoding layer consists of a face feature extraction block and a downsampling module, which can be expressed as follows.

[0078]

[0079] in, Represents the input of the encoding layer. For the first encoding layer, Represents a noisy face image R noise , for the second and third coding layers, represents the output of the previous encoding layer, represents the output of the i-th encoding layer, FFER represents the face feature extraction block, and downSample represents the downsampling block. The face feature extraction block consists of two normalization layers, a global face structure information extraction unit, and a gated convolutional layer that replaces the fully connected layer. The global face structure information extraction unit mainly uses linear layers, convolutional layers, and hourglass networks to implement a simple attention algorithm to complete the extraction of global face structure information. The calculation process is shown below.

[0080]

[0081]

[0082]

[0083] Among them, Linear represents the linear layer, HourGlass represents the hourglass network, Sigmod represents the activation function, I att represents the calculated attention feature, I filter represents the calculated attention weight, I G It represents the output result of the global face structure information unit.

[0084] GFSR is used to represent the global face structure information extraction unit, and the calculation process of the entire face feature extraction block is simplified to:

[0085]

[0086]

[0087] Among them, GateConv represents gated convolution, I noise represents a pure Gaussian noise, and LN represents a normalization layer.

[0088] After being processed by the encoding layer, the feature information obtained will be further integrated in the bottleneck layer. The design of the bottleneck layer is relatively simple, and only contains a global face structure information extraction unit, which can be expressed as follows.

[0089] I B =GFSR(I E )

[0090] Among them, I B represents the integration result of the bottleneck layer, I E Represents the output of the last encoding layer.

[0091] After information integration, the results of the bottleneck layer will be sent to the decoding layer for information integration and image restoration. Corresponding to the encoding layer, there are also three decoding layers, but each is composed of an upsampling block and a face feature extraction block. In addition, in order to obtain richer information, the decoding layer will not only receive the output from the previous decoding layer, but also the output of the encoding layer corresponding to its scale. The calculation process is as follows.

[0092]

[0093] Among them, upSample represents the upsampling block, Represents the input of the decoding layer. For the first decoding layer, Represents the output of the bottleneck layer, that is, the integration result of the bottleneck layer I B , for the second and third decoding layers, represents the output of the previous decoding layer, represents the output of the i-th decoding layer.

[0094] Finally, a convolutional layer is used to transform the output of the last decoding layer into the final result, which can be expressed as follows.

[0095] I HH =Conv(I D )

[0096] Among them, I HH represents the high-resolution face output image without noise under normal illumination, that is, the output image, I D Represents the result graph of the last decoding layer.

[0097] like Figure 3 , Figure 5 and Figure 6 As shown, in this embodiment, the noisy image R obtained in the previous step S2 is noise It is sent to the denoising diffusion module of the U-shaped architecture. In the denoising diffusion module, R noise It will first be followed by a pure Gaussian noise I noise The image is then stitched together and processed by the encoding layer, bottleneck layer, and decoding layer in turn to fully extract information and use it to repair the image. Finally, the output image I is obtained through a convolutional layer. HH .

[0098] S4. Complete the training of low-light face super-resolution network model through minimum diffusion loss;

[0099] Specifically, the low-light face super-resolution network model is trained using diffusion loss. When calculating the diffusion loss, the Gaussian noise added to the output result is estimated using the inverse diffusion transform, and then the L1 Loss is calculated with the true value of the Gaussian noise to constrain the model training, which can be expressed as follows:

[0100] I ε =INV(I HH )

[0101] Loss diff =||I ε -I noise ||1

[0102] Where INV represents the inverse diffusion transformation, I ε Represents the estimated value of Gaussian noise, Loss diff represents diffusion loss.

[0103] S5. Use the trained low-light face super-resolution network model to predict the collected face image test samples, obtain the super-resolution results, and use the evaluation indicators to evaluate them;

[0104] Specifically, the test samples are predicted and the super-resolution results are compared with the real results. In order to evaluate the performance of low-light face super-resolution results, the peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) are selected as evaluation indicators.

[0105] The peak signal-to-noise ratio (PSNR) is calculated as follows:

[0106]

[0107]

[0108] Where MSE stands for mean square error, n stands for the total number of test samples, and x i and i They respectively represent the face images I HH and the real image I GT Pixels, MAX I Indicates the maximum value of the color of an image point, usually 255.

[0109] The calculation method of structural similarity SSIM is:

[0110] c1=(k1L) 2

[0111] c2=(k2L) 2

[0112]

[0113] in, isI HH The average value of isI GT The average value of isI HH The variance of isI GT The variance of L is the dynamic range of pixel values, k1 and k2 are parameters with values ​​of k1=0.01 and k2=0.03, and c1 and c2 represent intermediate variables.

[0114] The test data is tested to obtain the super-resolution results, which are evaluated using the above indicators; the effectiveness of this method is demonstrated through extensive experiments and applied to the low-light face super-resolution task.

[0115] This method uses two indicators, PSNR and SSIM, to evaluate the super-resolution results. Two representative algorithms in the field of face super-resolution are selected: SR3 and IDM. The indicator results are shown in Table 1. The method of the present invention (Ours) has achieved the best results in both PSNR and SSIM indicators, indicating that the super-resolution results of the present invention are closer to the real results.

[0116] Table 1 Results of low-light face super-resolution indicators

[0117]

[0118] Based on the same concept, a low-light face super-resolution system based on Retinex and diffusion model of the present invention includes: a model construction and training unit, which constructs a low-light face super-resolution model based on Retinex and diffusion model, and uses a face image data set for training. A test evaluation module is used to use the trained low-light face super-resolution network model to test the collected face image test samples, obtain super-resolution results, and use evaluation indicators for evaluation.

[0119] The model building and training unit includes a data processing module, a noise image estimation module, a denoising diffusion module and a training module. The data processing module converts the input face image I LL Normalize and use bicubic interpolation to get the upsampled image I LH The noise image estimation module uses the Retinex theory to transform I LH Decomposed into a noisy face image R noise ; The denoising diffusion module will first convert R noise With a pure Gaussian noise I noise The image is then processed by the encoding layer, bottleneck layer, and decoding layer in turn to fully extract information and use it to repair the image. Finally, the output image I is obtained through a convolution layer. HHThe training module completes the training of the low-light face super-resolution network by minimizing the diffusion loss. The test evaluation unit calculates indicators based on the super-resolution results of the test samples and evaluates the results of the low-light face super-resolution.

[0120] The low-light face super-resolution network model includes two main parts: the noise image estimation module and the denoising diffusion module. In the first step, the noise image estimation module performs a Retinex transform on the input image to obtain a noisy face image R noise ; In the second step, the denoising diffusion module of the U-Net architecture is used to eliminate the noise and obtain a high-resolution face output image I without noise HH The entire network is trained using diffusion loss.

[0121] Based on the same inventive concept, a low-light face super-resolution device based on Retinex and diffusion model of the present invention includes a memory and a processor, wherein:

[0122] A memory for storing computer programs that can be run on the processor;

[0123] The processor is used to execute the steps of the above-mentioned low-light face super-resolution device method based on Retinex and diffusion model when running the computer program.

[0124] Based on the same inventive concept, a storage medium of the present invention stores a computer program, and when the computer program is executed by at least one processor, the steps of the above-mentioned low-light face super-resolution method based on Retinex and diffusion model are implemented.

Claims

1. A low-light face super-resolution method based on Retinex and diffusion model, characterized in that: The following steps are involved: Construct a low-light face super-resolution network model based on Retinex and diffusion models, and use the face image dataset for training; Use the trained low-light face super-resolution network model to test the collected face image test samples, obtain the super-resolution results, and use evaluation indicators to evaluate; The low-light face super-resolution network model includes: a noise image estimation module and a denoising diffusion module. The denoising diffusion module is a U-Net architecture, including an encoding layer, a bottleneck layer, and a decoding layer. The model input image is retinex transformed through the noise image estimation module to obtain the noisy face image R noise ; R obtained by the noise image estimation module noise After being concatenated with pure Gaussian noise, it is processed through three consecutive coding layers to obtain the coding layer output image I E ; Coding layer output image I E Further information integration is performed at the bottleneck layer, and the integrated information is sequentially passed through three decoding layers for information integration and image restoration; then the convolution layer outputs a noise-free high-resolution face output image I HH ; The entire network model is trained using diffusion loss.

2. The low-light face super-resolution method based on Retinex and diffusion model according to claim 1, characterized in that: Before using the face image dataset to train the low-light face super-resolution network model, it is also necessary to normalize the face image data in the face image dataset and obtain the upsampled image I by cubic interpolation. LH .

3. The low-light face super-resolution method based on Retinex and diffusion model according to claim 1, characterized in that: The input image is retinexed through the noise image estimation module to obtain the noisy face image R noise , including: first, perform three-channel averaging on the input image to obtain a color mean image I mean ; Then the input image and the color mean image I mean Splicing is performed in the channel dimension, and the brightness inverse component is obtained by sequentially passing through the convolution layer, the DW convolution layer and the convolution layer. Finally, according to the Retinex theory, with I LH Multiply to get the noisy face image R noise ; expressed as: I mean =F(I LH ) Among them, the function F means taking the average on three channels, the function Con means splicing in the channel dimension, Conv means ordinary convolution, and DWConv means DW convolution.

4. The low-light face super-resolution method based on Retinex and diffusion model according to claim 1, characterized in that: The encoding layer consists of a face feature extraction block and a downsampling module, expressed as: in, Represents the input of the encoding layer. For the first encoding layer, Represents a noisy face image R noise , for the second and third coding layers, represents the output of the previous encoding layer, represents the output of the i-th encoding layer, FFER represents the face feature extraction block, and downSample represents the downsampling block. The facial feature extraction block includes two normalization layers, a global facial structure information extraction unit, and a gated convolutional layer that replaces the fully connected layer. The global facial structure information extraction unit uses linear layers, convolutional layers, and hourglass networks to implement the attention algorithm to extract global facial structure information, which can be expressed as: Among them, Linear represents the linear layer, HourGlass represents the hourglass network, Sigmod represents the activation function, I att represents the calculated attention feature, I filter represents the calculated attention weight, I G It represents the output result of the global face structure information unit; GFSR is used to represent the global face structure information extraction unit, and the calculation process of the entire face feature extraction block is simplified to: Among them, GateConv represents gated convolution, I noise represents a pure Gaussian noise, and LN represents a normalization layer.

5. The low-light face super-resolution method based on Retinex and diffusion model according to claim 1, characterized in that: The bottleneck layer includes a global face structure information extraction unit, which is expressed as: I B =GFSR(I E ) Among them, I B represents the integration result of the bottleneck layer, I E Represents the output of the last encoding layer.

6. The low-light face super-resolution method based on Retinex and diffusion model according to claim 1, characterized in that: The decoding layer corresponds to the encoding layer. Each decoding layer includes an upsampling block and a face feature extraction block. The decoding layer not only receives the output from the previous decoding layer, but also receives the output of the encoding layer corresponding to its scale. The calculation process is: Among them, upSample represents the upsampling block, Represents the input of the decoding layer. For the first decoding layer, Represents the integration result of the bottleneck layer I B , for the second and third decoding layers, represents the output of the previous decoding layer, represents the output of the i-th decoding layer, represents the output of the i-th encoding layer. Finally, a convolutional layer is used to transform the output of the last decoding layer into the final result, which is expressed as follows: I HH =Conv(I D ) Among them, I HH represents the noise-free high-resolution face output image under normal lighting, I D Result graph representing the output of the last decoding layer.

7. The low-light face super-resolution method based on Retinex and diffusion model according to claim 1, characterized in that: When calculating the diffusion loss, the Gaussian noise added to the output result is estimated using the inverse diffusion transform, and then the L1 Loss is calculated with the true value of the Gaussian noise to constrain the model training, which is expressed as: I ε =INV(I HH ) Loss diff =||I ε -I noise ||1 Where INV represents the inverse diffusion transformation, I ε Represents the estimated value of Gaussian noise, Loss diff represents diffusion loss.

8. The low-light face super-resolution method based on Retinex and diffusion model according to claim 1, characterized in that: Evaluation indicators include peak signal-to-noise ratio (PSNR) and structural similarity (SSIM); The peak signal-to-noise ratio (PSNR) is calculated as follows: Where MSE stands for mean square error, n stands for the total number of test samples, and x i and i They respectively represent the face images I HH and the real image I GT Pixels, MAX I Indicates the maximum value of the color of an image point; The calculation method of structural similarity SSIM is: c1=(k1L) 2 c2=(k2L) 2 in, isI HH The average value of isI GT The average value of isI HH The variance of isI GT variance, L is the dynamic range of pixel values, k1 and k2 are parameters, and c1 and c2 represent intermediate variables.

9. A low-light face super-resolution system based on Retinex and diffusion model, characterized in that: include: Model building and training unit, used to build a low-light face super-resolution network model based on Retinex and diffusion models, and train it using face image datasets; and a test evaluation unit, which is used to test the collected face image test samples using the trained low-light face super-resolution network model, obtain the super-resolution results, and evaluate them using evaluation indicators; The low-light face super-resolution network model includes: a noise image estimation module and a denoising diffusion module. The denoising diffusion module is a U-Net architecture, including an encoding layer, a bottleneck layer, and a decoding layer. The model input image is retinex transformed through the noise image estimation module to obtain the noisy face image R noise ; R obtained by the noise image estimation module noise After being concatenated with pure Gaussian noise, the image is processed through three consecutive coding layers to obtain image I E ; Coding layer output image I E Further information integration is performed at the bottleneck layer, and the integrated information is sequentially passed through three decoding layers for information integration and image restoration; then the convolution layer outputs a noise-free high-resolution face output image I HH ; The entire network model is trained using diffusion loss.

10. A low-light face super-resolution device based on Retinex and diffusion model, characterized in that: comprising a memory and a processor, wherein: A memory for storing computer programs that can be run on the processor; A processor is used to execute the steps of the low-light face super-resolution method based on Retinex and diffusion model as described in any one of claims 1 to 8 when running the computer program.

Images