Blind image quality evaluation method based on complex value deep convolutional network and gradient information

Abstract

The invention discloses a blind image quality evaluation method based on a complex value deep convolutional network and gradient information. The blind image quality evaluation method comprises the steps of obtaining a to-be-evaluated distorted image; generating a corresponding gradient image by using a gradient detection operator; extracting multilevel features from the distorted image and the gradient image corresponding to the distorted image through complex-valued deep convolutional networks with the same structure; the complex-valued deep convolutional network firstly obtains high-frequency and low-frequency complex-valued responses of an input image by using dual-tree complex wavelet transform, and then extracts multi-layer complex-valued convolutional features for the high-frequency and low-frequency responses by using a three-layer complex-valued convolutional block. Corresponding layer complex value convolution features of the high-frequency shunt and the low-frequency shunt are fused by using a high-low frequency feature aggregation block, and the fused features of the high-low frequency feature aggregation block are transmitted to the next high-low frequency feature aggregation block to obtain multi-level features; and finally, the distorted image and the gradient image are cascaded through multilevel features output by the complex-valued deep convolutional network. According to the method, the structure information is better extracted.

Description

Technical Field

[0001] The invention relates to a blind image quality evaluation method based on a complex-valued deep convolutional network and gradient information, and belongs to the technical field of image quality evaluation. Background Art

[0002] Image quality assessment has important applications in image compression, image enhancement and other fields. Image quality assessment methods can be roughly divided into traditional methods based on manual feature extraction and methods based on deep learning. The current mainstream image quality assessment methods based on deep learning mainly use real-valued convolutional neural networks to extract features and then complete the quality assessment task. Studies have shown that complex-valued neural networks have more advantages in learning the phase and structure information of image signals.

[0003] Using complex-valued deep convolutional networks to extract multi-level features of distorted images and combining gradient information to effectively express structural features for quality evaluation is expected to achieve better image quality evaluation results. Summary of the invention

[0004] In order to overcome the defects of the prior art, the present invention proposes a blind image quality assessment method based on complex-valued deep convolutional network and gradient information. The method explicitly uses the gradient information of the distorted image, uses the complex-valued deep convolutional network with the same structure to extract multi-level features for the distorted image and the gradient image, and uses the high- and low-frequency feature aggregation block to fuse the corresponding layer complex-valued convolution features of the high-frequency branch and the low-frequency branch. The method better extracts the phase information and multi-scale structural information of the distorted image, thereby improving the accuracy of blind image quality assessment.

[0005] To achieve the above object, the present invention is implemented by adopting the following technical solutions:

[0006] The present invention provides a blind image quality evaluation method based on a complex-valued deep convolutional network and gradient information, the method comprising:

[0007] Obtain a distorted image to be evaluated;

[0008] Generate the corresponding gradient image using the gradient detection operator;

[0009] The distorted image and its corresponding gradient image are respectively passed through a complex-valued deep convolutional network with the same structure to extract multi-level features; the complex-valued deep convolutional network first uses a dual-tree complex wavelet transform to obtain the high-frequency and low-frequency complex-valued responses of the input image, and then uses three layers of complex-valued convolutional blocks to extract multi-layer complex-valued convolutional features for both high-frequency and low-frequency responses, and uses a high- and low-frequency feature aggregation block to fuse the corresponding layers of complex-valued convolutional features of the high-frequency branch and the low-frequency branch, and the fused features of the high- and low-frequency feature aggregation block are passed to the next layer of high- and low-frequency feature aggregation blocks to obtain multi-level characteristics;

[0010] Finally, the multi-level features of the distorted image and the gradient image output by the complex-valued deep convolutional network are cascaded and input into the quality prediction module to obtain the quality prediction score.

[0011] Furthermore, the step of generating a corresponding gradient image using a gradient detection operator to explicitly extract structural features of the distorted image includes:

[0012] The gradient image can be obtained using common gradient detection operators, such as the Sobel operator.

[0013] Furthermore, the distorted image and its corresponding gradient image are respectively passed through a complex-valued deep convolutional network with the same structure to extract multi-level features, including:

[0014] The complex-valued deep convolutional network first uses the dual-tree complex wavelet transform to obtain the high-frequency and low-frequency complex-valued responses of the input image, and then uses three-layer complex-valued convolution blocks to extract multi-layer complex-valued convolution features for both high-frequency and low-frequency responses. The high- and low-frequency feature aggregation block is used to fuse the corresponding layers of complex-valued convolution features of the high-frequency branch and the low-frequency branch, and the fused features of the high- and low-frequency feature aggregation block are passed to the next layer of high- and low-frequency feature aggregation blocks to obtain multi-level characteristics.

[0015] Furthermore, the complex-valued deep convolutional network first obtains high-frequency and low-frequency complex-valued responses of the input image using a dual-tree complex wavelet transform, including:

[0016] The distorted image and gradient image are decomposed by dual-tree complex wavelet transform to obtain the complex value response of the distorted image (including high-frequency features F H With low frequency feature F L ) and the complex-valued response of the gradient image (including high-frequency features G H With low frequency feature G L ).

[0017] Furthermore, the three-layer complex-valued convolution block is used to extract multi-layer complex-valued convolution features for both high-frequency and low-frequency responses, including:

[0018] Three layers of complex-valued convolution blocks are used to extract multi-layer complex-valued convolution features for the high-frequency and low-frequency responses of the distorted image. and They represent the nth layer of complex-valued convolution features extracted from the high-frequency and low-frequency branches of the distorted image, and the corresponding feature dimensions are H and W represent the height and width of the input image, respectively. n Represents the number of channels for extracting features in the corresponding layer; similarly, three layers of complex-valued convolution blocks are used to extract multi-layer complex-valued convolution features for the high-frequency and low-frequency responses of the gradient image, respectively, which is expressed as and

[0019] Each layer of complex-valued convolution block consists of three complex-valued convolution operations (CV-conv3×3), the convolution kernel size is 3×3, and complex-valued average pooling (AP) is used between the second and third complex-valued convolutions to reduce the spatial dimension. Taking the complex-valued convolution feature extraction of the high-frequency and low-frequency branches of the distorted image as an example, the feature extraction expression is as follows:

[0020]

[0021] Where n represents the index of the complex-valued convolution block; Respectively represent the output features of the nth complex-valued convolutional block of the high-frequency and low-frequency branches, and the corresponding feature dimensions are H and W represent the height and width of the distorted image, respectively. n Indicates the number of channels for extracting features; CV-conv3×3 represents 3×3 complex-valued convolution; AP represents complex-valued average pooling.

[0022] Furthermore, the high- and low-frequency feature aggregation block is used to fuse the corresponding layer complex-valued convolution features of the high-frequency branch and the low-frequency branch, and the fused features of the high- and low-frequency feature aggregation block are passed to the next layer of high- and low-frequency feature aggregation blocks to obtain multi-level features, including:

[0023] The high- and low-frequency feature aggregation block is composed of complex-valued dilated convolutions with different dilation rates and channel attention modules. The input of the high- and low-frequency feature aggregation block (X n ) The high-frequency features extracted by the corresponding layer complex-valued convolutional block Low frequency characteristics And the output features of the high and low frequency feature aggregation block of the previous layer (F n-1 ) cascaded, input feature X n First, multi-scale context information is captured through complex-valued dilated convolutions with different receptive field sizes. Then, channel-enhanced features are obtained through the channel attention module. Finally, the features after channel enhancement at different scales are cascaded to obtain the output features of the current high- and low-frequency feature aggregation block. The calculation process of the high- and low-frequency feature aggregation block can be expressed by the following formula:

[0024]

[0025] Where AP represents complex average pooling; n represents the index of the high- and low-frequency feature aggregation block; r represents the expansion rate; Represents the input feature X n The weighted features after dilated convolution with dilation rate r and passing through the channel attention module; GAP represents complex-valued global average pooling; f c represents 1×1 complex-valued convolution; σ represents complex-valued Sigmoid activation; δ represents complex-valued ReLU activation; Indicates element-wise multiplication; fr represents a complex-valued dilated convolution with a dilation rate of r; represents channel dimension cascade; F n Represents the output features of the current high- and low-frequency feature aggregation block, the output F of the high- and low-frequency feature aggregation block of the third layer (index n=3) 3 As the multi-level features of the input distorted image output by the complex-valued deep convolutional network; similarly, the output G of the high- and low-frequency feature aggregation block of the third layer (index n=3) can be obtained 3 As the multi-level features output by the complex-valued deep convolutional network of the input gradient image.

[0026] Furthermore, the multi-level feature cascade of the distorted image and the gradient image output by the complex-valued deep convolutional network is input into the quality prediction module to obtain the quality prediction score, including:

[0027] The quality prediction module uses global average pooling and a multi-layer perceptron (MLP) containing three fully connected layers to fuse the above features and predict the perceived quality. The predicted quality score expression is:

[0028]

[0029] Among them, Q i is the quality prediction score of the i-th distorted image, F 3 and G 3 The distorted image and the corresponding gradient image are respectively extracted through the multi-level features of the complex-valued deep convolutional network.

[0030] Furthermore, the training method of the overall network model is as follows:

[0031] Get training images;

[0032] Use the gradient detection operator to generate the gradient image corresponding to the training image;

[0033] The training image and the corresponding gradient image are respectively passed through a complex-valued deep convolutional network with the same structure to extract multi-level features; the multi-level features of the distorted image and the gradient image output by the complex-valued deep convolutional network are cascaded and input into the quality prediction module to obtain the quality prediction score;

[0034] According to the overall loss function of the complex-valued deep convolutional network image quality evaluation model, the parameters of the complex-valued deep convolutional network model are adjusted by gradient back propagation to obtain the trained complex-valued deep convolutional network image quality prediction model.

[0035] Furthermore, the overall loss function is the mean absolute error between the quality scores predicted by the model and the subjective quality scores of all training images. The overall loss function expression is as follows:

[0036]

[0037] Among them, Q i is the predicted quality score of the i-th image, s i is its corresponding subjective quality score, and N is the number of training images.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] 1. The present invention proposes a blind image quality evaluation method based on a complex-valued deep convolutional network and gradient information. The method better extracts structural information by adopting a dual-branch network structure of a distorted image and a gradient image.

[0040] 2. The present invention adopts a complex-valued deep convolutional network to better express the quality-related phase information compared to the real-valued convolutional network, and at the same time adopts high- and low-frequency feature aggregation blocks to better extract multi-scale semantic information, thereby achieving better image quality evaluation performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is an overall flow chart of a blind image quality assessment method based on a complex-valued deep convolutional network and gradient information according to an embodiment of the present invention.

[0042] Figure 2 It is a schematic diagram of the structure of a complex-valued deep convolutional network according to an embodiment of the present invention.

[0043] Figure 3 2 is a schematic diagram of the structure of a complex-valued convolution block according to an embodiment of the present invention.

[0044] Figure 4 It is a schematic diagram of the structure of the high and low frequency feature aggregation block of an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The present invention will be further described below in conjunction with the accompanying drawings and embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0046] The present invention provides a blind image quality evaluation method based on a complex-valued deep convolutional network and gradient information, comprising:

[0047] Obtain a distorted image to be evaluated;

[0048] Generate the corresponding gradient image using the gradient detection operator;

[0049] The distorted image and its corresponding gradient image are respectively passed through a complex-valued deep convolutional network with the same structure to extract multi-level features; the complex-valued deep convolutional network first uses a dual-tree complex wavelet transform to obtain the high-frequency and low-frequency complex-valued responses of the input image, and then uses three layers of complex-valued convolutional blocks to extract multi-layer complex-valued convolutional features for both high-frequency and low-frequency responses, and uses a high- and low-frequency feature aggregation block to fuse the corresponding layers of complex-valued convolutional features of the high-frequency branch and the low-frequency branch, and the fused features of the high- and low-frequency feature aggregation block are passed to the next layer of high- and low-frequency feature aggregation blocks to obtain multi-level characteristics;

[0050] Finally, the multi-level features of the distorted image and the gradient image output by the complex-valued deep convolutional network are cascaded and input into the quality prediction module to obtain the quality prediction score.

[0051] This embodiment provides a blind image quality assessment method based on a complex-valued deep convolutional network and gradient information, such as Figure 1 As shown, the application process includes the following steps:

[0052] Step 1) Obtain the distorted image to be evaluated.

[0053] Step 2) Generate the corresponding gradient image using the gradient detection operator.

[0054] In the implementation of the present invention, Sobel horizontal and vertical gradient operators are used to perform template operations on pixel areas at corresponding positions of the distorted image to calculate the horizontal and vertical gradients of each pixel position of the distorted image. The Sobel horizontal and vertical gradient operators (g x ,g y )as follows:

[0055]

[0056] The horizontal and vertical gradients calculated at the pixel coordinate (i, j) of the distorted image I are and The gradient amplitude (G) of the distorted image at position (i, j) is calculated according to the following formula: i，j ), the entire gradient image corresponding to the distorted image is denoted as G.

[0057]

[0058] Step 3) The distorted image and its corresponding gradient image are respectively passed through a complex-valued deep convolutional network with the same structure to extract multi-level features, such as Figure 2 First, the distorted image I and the gradient image G are decomposed by dual-tree complex wavelet transform to obtain the complex value response of the distorted image (including the high-frequency feature F H With low frequency feature F L ) and the complex-valued response of the gradient image (including high-frequency features GH With low frequency feature G L ).

[0059] Step 4) Use three layers of complex-valued convolution blocks to extract multiple complex-valued convolution features from the high-frequency and low-frequency responses of the distorted image and gradient image. Each layer of complex-valued convolution blocks consists of three complex-valued convolution operations (CV-conv3×3), and the convolution kernel size is 3×3. Complex-valued average pooling (AP) is used between the second and third complex-valued convolutions to reduce the spatial dimension, such as Figure 3 As shown. Taking the complex-valued convolution feature extraction of the high-frequency and low-frequency branches of the distorted image as an example, the expression of feature extraction is as follows:

[0060]

[0061] Where n represents the index of the complex-valued convolution block; They represent the output features of the nth complex-valued convolutional block of the high-frequency and low-frequency branches of the distorted image, respectively. The corresponding feature dimensions are H and W represent the height and width of the distorted image, respectively. n Indicates the number of channels for extracting features, and its values ​​are {C n |64, 256, 512}; CV-conv3×3 represents 3×3 complex-valued convolution; AP represents complex-valued average pooling.

[0062] Similarly, three layers of complex-valued convolution blocks are used to extract multi-layer complex-valued convolution features for the high-frequency and low-frequency responses of the gradient image, respectively, which can be expressed as and

[0063] Step 5) Use the high- and low-frequency feature aggregation block to fuse the corresponding layer complex-valued convolution features of the high-frequency branch and the low-frequency branch, such as Figure 4 As shown, the fused features of the high- and low-frequency feature aggregation blocks are passed to the next layer of high- and low-frequency feature aggregation blocks to obtain multi-level features.

[0064] The high- and low-frequency feature aggregation block consists of complex-valued dilated convolutions with different dilation rates and a channel attention module. The input of the high- and low-frequency feature aggregation block (X n ) The high-frequency features extracted by the corresponding layer complex-valued convolutional block Low frequency characteristics And the output features of the high and low frequency feature aggregation block of the previous layer (F n-1 ) cascaded, input feature X n First, multi-scale context information is captured through complex-valued dilated convolutions with different receptive field sizes. Then, channel-enhanced features are obtained through the channel attention module. Finally, the features after channel enhancement at different scales are cascaded to obtain the output features of the current high- and low-frequency feature aggregation block. The calculation process of the high- and low-frequency feature aggregation block can be expressed by the following formula:

[0065]

[0066]

[0067] Where AP represents complex average pooling; n represents the index of the high- and low-frequency feature aggregation block; r represents the expansion rate; Represents the input feature X n The weighted features after dilated convolution with dilation rate r and passing through the channel attention module; GAP represents complex-valued global average pooling; f c represents 1×1 complex-valued convolution; σ represents complex-valued Sigmoid activation; δ represents complex-valued ReLU activation; Indicates element-wise multiplication; f r represents a complex-valued dilated convolution with a dilation rate of r; represents channel dimension cascade; F n Represents the output features of the current high- and low-frequency feature aggregation block, the output F of the high- and low-frequency feature aggregation block of the third layer (index n=3) 3 As the multi-level features of the input distorted image output by the complex-valued deep convolutional network; similarly, the output G of the high- and low-frequency feature aggregation block of the third layer (index n=3) can be obtained 3 As the multi-level features output by the complex-valued deep convolutional network of the input gradient image.

[0068] The specific process of the channel attention operation is as follows: First, the input feature X n Perform a dilated convolution with a dilation rate of r to obtain f r (X n ), then f r (X n ) Perform a complex-valued global average pooling operation along the channel dimension, perform a 1×1 complex-valued convolution on the result and use a complex-valued ReLU activation, perform a 1×1 complex-valued convolution and a complex-valued Sigmoid activation again to obtain the corresponding channel-dimensional attention weight, and f r (X n ) element-wise multiplication to obtain the channel attention optimized features.

[0069] Step 6) The multi-level features of the distorted image and the gradient image are cascaded through the complex-valued deep convolutional network output and input into the quality prediction module to obtain the quality prediction score. The quality prediction module uses global average pooling and a multi-layer perceptron (MLP) containing three fully connected layers to fuse the above features and predict the perceived quality. The predicted quality score expression is:

[0070]

[0071] Among them, Q iis the quality prediction score of the i-th distorted image, F 3 and G 3 The distorted image and the corresponding gradient image are respectively extracted through the multi-level features of the complex-valued deep convolutional network.

[0072] Compared with the general image quality assessment method based on real-valued deep network, the blind image quality assessment method based on complex-valued deep convolutional network and gradient information proposed in the present invention makes better use of quality-related phase information and structural information, thereby improving the accuracy of image quality assessment.

[0073] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0074] The present invention is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0075] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0076] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1The steps for the functions specified in one or more boxes.

[0077] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A blind image quality assessment method based on complex-valued deep convolutional network and gradient information, characterized in that: include: Obtain a distorted image to be evaluated; Generate the corresponding gradient image using the gradient detection operator; The distorted image and its corresponding gradient image are respectively passed through a complex-valued deep convolutional network with the same structure to extract multi-level features; Finally, the multi-level features of the distorted image and the gradient image output by the complex-valued deep convolutional network are cascaded and input into the quality prediction module to obtain the quality prediction score.

2. The blind image quality assessment method based on complex-valued deep convolutional network and gradient information according to claim 1, characterized in that: The method of generating a corresponding gradient image by using a gradient detection operator to explicitly extract structural features of the distorted image includes: The gradient image is obtained using a common gradient detection operator, namely the Sobel operator.

3. The blind image quality assessment method based on complex-valued deep convolutional network and gradient information according to claim 1, characterized in that: The distorted image and its corresponding gradient image are respectively passed through a complex-valued deep convolutional network with the same structure to extract multi-level features, including: The complex-valued deep convolutional network first uses the dual-tree complex wavelet transform to obtain the high-frequency and low-frequency complex-valued responses of the input image, and then uses three-layer complex-valued convolution blocks to extract multi-layer complex-valued convolution features for both high-frequency and low-frequency responses. The high- and low-frequency feature aggregation block is used to fuse the corresponding layers of complex-valued convolution features of the high-frequency branch and the low-frequency branch, and the fused features of the high- and low-frequency feature aggregation block are passed to the next layer of high- and low-frequency feature aggregation blocks to obtain multi-level characteristics.

4. The blind image quality assessment method based on complex-valued deep convolutional network and gradient information according to claim 3, characterized in that: The complex-valued deep convolutional network first obtains high-frequency and low-frequency complex-valued responses of the input image using dual-tree complex wavelet transform, including: The distorted image and gradient image are decomposed by dual-tree complex wavelet transform to obtain the complex value response of the distorted image (including high-frequency features F H With low frequency feature F L ) and the complex-valued response of the gradient image (including high-frequency features G H With low frequency feature G L ).

5. The blind image quality assessment method based on complex-valued deep convolutional network and gradient information according to claim 3, characterized in that: The three-layer complex-valued convolution block is used to extract multi-layer complex-valued convolution features for both high-frequency and low-frequency responses, including: Three layers of complex-valued convolution blocks are used to extract multi-layer complex-valued convolution features for the high-frequency and low-frequency responses of the distorted image. and They represent the nth layer of complex-valued convolution features extracted from the high-frequency and low-frequency branches of the distorted image, and the corresponding feature dimensions are H and W represent the height and width of the input image, respectively. n Represents the number of channels for extracting features in the corresponding layer; similarly, three layers of complex-valued convolution blocks are used to extract multi-layer complex-valued convolution features for the high-frequency and low-frequency responses of the gradient image, respectively, which is expressed as and 6. The blind image quality assessment method based on complex-valued deep convolutional network and gradient information according to claim 3, characterized in that: The high- and low-frequency feature aggregation blocks are used to fuse the corresponding layer complex-valued convolution features of the high-frequency branch and the low-frequency branch, and the fused features of the high- and low-frequency feature aggregation blocks are passed to the next layer of high- and low-frequency feature aggregation blocks to obtain multi-level features, including: The high- and low-frequency feature aggregation block is composed of complex-valued dilated convolutions with different dilation rates and channel attention modules. The input of the high- and low-frequency feature aggregation block (X n ) The high-frequency features extracted by the corresponding layer complex-valued convolutional block Low frequency characteristics And the output features of the high and low frequency feature aggregation block of the previous layer (F n-1 ) cascaded, input feature X n First, multi-scale context information is captured through complex-valued dilated convolutions with different receptive field sizes. Then, channel-enhanced features are obtained through the channel attention module. Finally, the features after channel enhancement at different scales are cascaded to obtain the output features of the current high- and low-frequency feature aggregation block. The calculation process of the high- and low-frequency feature aggregation block can be expressed by the following formula: Where AP represents complex average pooling; n represents the index of the high- and low-frequency feature aggregation block; r represents the expansion rate; Represents the input feature X n The weighted features after dilated convolution with dilation rate r and passing through the channel attention module; GAP represents complex-valued global average pooling; f c represents 1×1 complex-valued convolution; σ represents complex-valued Sigmoid activation; δ represents complex-valued ReLU activation; Indicates element-wise multiplication; f r represents a complex-valued dilated convolution with a dilation rate of r; represents channel dimension cascade; F n Represents the output features of the current high- and low-frequency feature aggregation block, the output F of the high- and low-frequency feature aggregation block of the third layer (index n=3) 3 As the multi-level features of the input distorted image output by the complex-valued deep convolutional network; similarly, the output G of the high- and low-frequency feature aggregation block of the third layer (index n=3) can be obtained 3 As the multi-level features output by the complex-valued deep convolutional network of the input gradient image.

7. The blind image quality assessment method based on complex-valued deep convolutional network and gradient information according to claim 1, characterized in that: The multi-level feature cascade of the distorted image and the gradient image output by the complex-valued deep convolutional network is input into the quality prediction module to obtain the quality prediction score, including: The quality prediction module uses global average pooling and a multi-layer perceptron (MLP) containing three fully connected layers to fuse the above features and predict the perceived quality. The predicted quality score expression is: Among them, Q i is the quality prediction score of the i-th distorted image, F 3 and G 3 The distorted image and the corresponding gradient image are respectively extracted through the multi-level features of the complex-valued deep convolutional network.

8. The blind image quality assessment method based on complex-valued deep convolutional network and gradient information according to claim 1, characterized in that: The training method of the overall network model is as follows: Get training images; Use the gradient detection operator to generate the gradient image corresponding to the training image; The training image and the corresponding gradient image are passed through a complex-valued deep convolutional network with the same structure to extract multi-level features; The multi-level features of the distorted image and the gradient image are cascaded through the complex-valued deep convolutional network output and input into the quality prediction module to obtain the quality prediction score; According to the overall loss function of the complex-valued deep convolutional network image quality evaluation model, the parameters of the complex-valued deep convolutional network model are adjusted by gradient back propagation to obtain the trained complex-valued deep convolutional network image quality prediction model.

9. The blind image quality assessment method based on complex-valued deep convolutional network and gradient information according to claim 8, characterized in that: The overall loss function is the mean absolute error between the quality scores predicted by the model and the subjective quality scores of all training images. The overall loss function expression is as follows: Among them, Q i is the predicted quality score of the i-th image, s i is its corresponding subjective quality score, and N is the number of training images.

Images