Image quality evaluation method, device and storage medium

By constructing an image quality evaluation method, using the support vector regression model to extract the structural features and texture features of the image, the problem of image distortion affecting the accuracy of object detection is solved, and the rapid and accurate evaluation of the image quality of transmission line inspection is achieved, and the improvement of the Accuracy of diagnostic results.

CN112116568BActive Publication Date: 2024-11-08GUANGZHOU POWER SUPPLY BUREAU GUANGDONG POWER GRID CO LTD +1
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202010950511.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-09-10
Publication Date
2024-11-08
Estimated Expiration
2040-09-10

AI Technical Summary

Technical Problem

Images are affected by distortion during acquisition and transmission, resulting in a decline in image quality and affecting the accuracy of later target detection and defect recognition. Especially during transmission line inspection, severe distortion will reduce the diagnostic results of line defects.

Method used

The image quality evaluation method is constructed, and by acquiring standard images and applying distortion processing, structural features, texture features and exposure are extracted, and image quality evaluation model is established using support vector regression to quickly and accurately evaluate image quality.

Benefits of technology

It realizes rapid and accurate evaluation of image quality, improves the diagnostic accuracy of transmission line patrol images, reduces errors, and meets the needs of smart grid construction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN112116568B_ABST
    Figure CN112116568B_ABST
Patent Text Reader

Abstract

The present invention discloses an image quality evaluation method, device and storage medium, the method comprising: obtaining multiple standard images, applying distortion processing to the standard images, and obtaining multiple distorted images; using a full reference image quality evaluation method to evaluate each distorted image, obtaining an evaluation score for each distorted image, and obtaining a scoring vector space based on the evaluation scores of all distorted images; extracting structural features, texture features and exposure of each distorted image, obtaining a feature vector of each image, and obtaining an image feature vector space based on the feature vectors of all distorted images; using support vector regression to establish a functional relationship between the scoring vector space and the image feature vector space, and obtaining an image quality evaluation model; inputting an image to be evaluated, and using the image quality evaluation model to obtain a quality score of the image to be evaluated. The present invention can quickly and accurately evaluate image quality.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of image evaluation, and in particular to an image quality evaluation method, device and storage medium. Background Art

[0002] In the process of image acquisition and transmission, images are inevitably affected by various distortions, such as compression, defocus blur, motion blur, noise, abnormal exposure, etc. The introduction of distortion changes the correlation between pixels and loses the original image information to a certain extent, which will affect the accuracy of subsequent target detection and defect recognition. In order to enable the detection and recognition algorithms to identify the target more accurately and reduce errors, it is necessary to predict the quality of the image to be detected.

[0003] Especially for the inspection images of power transmission lines, most of them use robots or drones to shoot the transmission lines, and then target detection, target recognition and other processing algorithms are used to diagnose line defects based on the inspection images. If the distortion of the inspection images is too serious, it will greatly reduce the diagnosis results of line defects. Summary of the invention

[0004] The purpose of the embodiments of the present invention is to provide an image quality evaluation method, device and storage medium, which first builds a proprietary image database, then extracts the structural features, texture features and exposure of the image, and then obtains the image quality evaluation model through support vector regression, so as to quickly and accurately evaluate the image quality.

[0005] To achieve the above object, an embodiment of the present invention provides an image quality evaluation method, comprising the following steps:

[0006] Obtain n standard images, and apply distortion processing to the standard images to obtain m distorted images; wherein m≥n≥1;

[0007] Using a full-reference image quality evaluation method to evaluate each of the distorted images, obtaining an evaluation score for each of the distorted images, and obtaining a scoring vector space based on the evaluation scores of all of the distorted images;

[0008] Extracting structural features, texture features and exposure of each of the distorted images to obtain a feature vector of each image, and obtaining an image feature vector space based on the feature vectors of all the distorted images;

[0009] Using support vector regression to establish a functional relationship between the score vector space and the image feature vector space to obtain an image quality evaluation model;

[0010] An image to be evaluated is input, and the quality score of the image to be evaluated is obtained using the image quality evaluation model.

[0011] Preferably, the step of acquiring n standard images and applying distortion processing to the standard images to obtain m distorted images specifically includes:

[0012] Obtain n standard images, apply preset multiple different degrees of Gaussian noise to each of the standard images, apply preset multiple different degrees of Gaussian blur to each of the standard images, apply preset multiple different degrees of compression processing to each of the standard images, and obtain m distorted images; wherein m>n≥1.

[0013] Preferably, the structural features of the distorted image include a shape parameter and a scale parameter of a Weibull distribution corresponding to the gradient amplitude of the distorted image.

[0014] Preferably, the texture features of the distorted image include local image texture features and global image texture features; wherein the local image texture features are obtained by counting the LBP values ​​of all pixels in the distorted image, and the global image texture features include the contrast, ASM energy and homogeneity of the gray level co-occurrence matrix.

[0015] Preferably, the gray level co-occurrence matrix is ​​obtained by:

[0016] Compressing and quantizing the grayscale range of the distorted image to a level of L to obtain a quantization matrix; wherein the input size of the distorted image is H×W, and L≥1;

[0017] Counting the number of occurrences of a pixel whose distance from a pixel whose pixel value is i in the quantization matrix is ​​d, whose direction is θ, and whose pixel value is j; wherein 1≤i≤L, 1≤j≤L, i≠j;

[0018] The quantization matrix is ​​traversed to obtain the gray-level co-occurrence matrix.

[0019] Preferably, the calculation formula of the contrast of the gray level co-occurrence matrix is: The calculation formula of the ASM energy of the gray level co-occurrence matrix is: The calculation formula of the homogeneity of the gray level co-occurrence matrix is ​​G H =P d,θ (i,j) / (1+|ij-2|); where, G C is the contrast of the gray-level co-occurrence matrix, P d,θ (i, j) is the element h with coordinates (i, j) in the gray-level co-occurrence matrix d,θ The probability corresponding to (i,j), G E is the ASM energy of the gray-level co-occurrence matrix, G H is the homogeneity of the gray level co-occurrence matrix.

[0020] Preferably, the exposure of the distorted image is quantified by information entropy.

[0021] Another embodiment of the present invention provides an image quality assessment device, the device comprising:

[0022] An image acquisition module, used for acquiring n standard images, and applying distortion processing to the standard images to obtain m distorted images; wherein m≥n≥1;

[0023] An image evaluation module, used for evaluating each of the distorted images by using a full reference image quality evaluation method to obtain an evaluation score for each of the distorted images, and obtaining a scoring vector space according to the evaluation scores of all the distorted images;

[0024] A feature extraction module, used to extract structural features, texture features and exposure of each distorted image to obtain a feature vector of each image, and to obtain an image feature vector space based on the feature vectors of all the distorted images;

[0025] A model building module, used to establish a functional relationship between the score vector space and the image feature vector space by using support vector regression to obtain an image quality evaluation model;

[0026] The testing module is used to input the image to be evaluated and obtain the quality score of the image to be evaluated using the image quality evaluation model.

[0027] Another embodiment of the present invention provides a device for using an image quality assessment method, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and when the processor executes the computer program, the image quality assessment method as described in any one of the above items is implemented.

[0028] Another embodiment of the present invention provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any of the image quality evaluation methods described above.

[0029] Compared with the prior art, the image quality assessment method, device and storage medium provided in the embodiments of the present invention first establish a dedicated database for image quality assessment, then construct an image feature vector by extracting the structural features, texture features and exposure features of the image, and then use a support vector machine to learn a mapping model, so that the quality assessment score of the image to be evaluated can be obtained according to the mapping model. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 is a flow chart of an embodiment of an image quality evaluation method provided by the present invention;

[0031] Figure 2 is a flow chart of another embodiment of an image quality evaluation method provided by the present invention;

[0032] Figure 3 is a structural schematic diagram of an embodiment of an image quality evaluation device provided by the present invention;

[0033] Figure 4 It is a structural schematic diagram of an embodiment of a device using an image quality evaluation method provided by the present invention. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0035] See also Figure 1 , is a flow chart of an embodiment of an image quality evaluation method provided by the present invention, the method comprising steps S1 to S5:

[0036] S1, obtaining n standard images, and applying distortion processing to the standard images to obtain m distorted images; wherein m≥n≥1;

[0037] S2. Evaluate each of the distorted images using a full-reference image quality evaluation method to obtain an evaluation score for each of the distorted images, and obtain a scoring vector space based on the evaluation scores of all the distorted images;

[0038] S3, extracting structural features, texture features and exposure of each distorted image to obtain a feature vector of each image, and obtaining an image feature vector space based on the feature vectors of all the distorted images;

[0039] S4, using support vector regression to establish a functional relationship between the score vector space and the image feature vector space to obtain an image quality evaluation model;

[0040] S5. Input an image to be evaluated, and use the image quality evaluation model to obtain a quality score of the image to be evaluated.

[0041] Specifically, n standard images are obtained, and distortion processing is applied to the standard images to obtain m distorted images; wherein, m≥n≥1. The standard image refers to a high-definition, distortion-free original image. In order to establish a standard image library for transmission line inspection images, high-quality images can be manually selected from a large number of inspection images provided as standard images, i.e., reference images. Image distortion includes compression, defocus blur, motion blur, noise, abnormal exposure, etc. The purpose of applying distortion processing to the standard image is to obtain a distorted image and compare it with the reference image.

[0042] Each distorted image is evaluated using the full reference image quality evaluation method to obtain an evaluation score for each distorted image, which is used as the label of the distorted image. Based on the evaluation scores of all distorted images, a scoring vector space is obtained.

[0043] The structural features, texture features and exposure of each distorted image are extracted to obtain the feature vector of each image. According to the feature vectors of all distorted images, the image feature vector space is obtained.

[0044] Support vector regression is used to establish the functional relationship between the score vector space and the image feature vector space, that is, to map the image feature vector space to the score vector space. This mapping relationship needs to be continuously adjusted and improved. Once the mapping relationship is established, the image quality evaluation model is obtained.

[0045] Input the image to be evaluated, extract the structural features, texture features and exposure of the image to be evaluated, obtain the image feature vector space of the image to be evaluated, and use the mapping relationship of the image quality evaluation model to obtain the quality score of the image to be evaluated.

[0046] Embodiment 1 of the present invention provides an image quality assessment method, which first establishes a dedicated database for image quality assessment, then constructs an image feature vector by extracting the structural features, texture features and exposure features of the image, and then uses a support vector machine to learn a mapping model, so that the quality assessment score of the image to be evaluated can be obtained according to the mapping model.

[0047] As an improvement of the above solution, the step of acquiring n standard images and applying distortion processing to the standard images to obtain m distorted images specifically includes:

[0048] Obtain n standard images, apply preset multiple different degrees of Gaussian noise to each of the standard images, apply preset multiple different degrees of Gaussian blur to each of the standard images, apply preset multiple different degrees of compression processing to each of the standard images, and obtain m distorted images; wherein m>n≥1.

[0049] Specifically, n standard images are obtained, and preset multiple different degrees of Gaussian noise are applied to each standard image, preset multiple different degrees of Gaussian blur are applied to each standard image, and preset multiple different degrees of compression are applied to each standard image to obtain m distorted images; wherein, m>n≥1. In order to obtain as many distorted images as possible based on a certain number of standard images, different degrees of distortion processing can be applied to each standard image, for example, 30%, 50% and 80% of Gaussian noise are applied, so that three distorted images can be obtained. Different types of distortion can also be applied. In this embodiment, the types of distortion applied mainly include noise, blur, and compression, wherein compression includes JPEG2000 compression and JPEG compression. For example, there are 128 standard images, and four types of distortion, namely, five degrees of Gaussian blur, Gaussian noise, JPEG2000 compression and JPEG compression, are applied to each standard image, so that 20 distorted images can be obtained for each standard image, for a total of 2560 images.

[0050] As an improvement of the above solution, the structural features of the distorted image include the shape parameter and scale parameter of the Weibull distribution corresponding to the gradient amplitude of the distorted image.

[0051] Specifically, the structural features of the distorted image include the shape parameters and scale parameters of the Weibull distribution corresponding to the gradient amplitude of the distorted image. In other words, the structural features of the distorted image are characterized by using the shape parameters and scale parameters of the Weibull distribution corresponding to the gradient amplitude of the distorted image.

[0052] Image structural information plays an important role in the recognition and understanding of objects. Image distortion changes the correlation between image pixels, affecting the recognition and understanding of content. The present invention first extracts image structural features to capture image quality sensitive features. For the input image I, first extract its gradient amplitude G, whose gradient amplitude is in, is the convolution operator, p x and p y Respectively represent the filter kernel functions in the horizontal direction and the vertical direction. The present invention adopts the prewitt filter,

[0053] The gradient amplitude of a high-quality image satisfies the two-dimensional Weibull distribution, and the existence of distortion will destroy this structure. Therefore, the Weibull distribution parameters are used to quantify the image structural distortion. The probability function of the two-dimensional Weibull distribution can be expressed as In the formula, x represents the gradient amplitude of the input image, i.e. G; α is the shape parameter, which controls the peak value of the distribution; λ is the scale parameter, which determines the width of the distribution. These parameters reflect the correlation hidden in the Weibull distribution, so the present invention selects α and λ to quantify the degree of image distortion.

[0054] As an improvement of the above scheme, the texture features of the distorted image include local image texture features and global image texture features; wherein the local image texture features are obtained by counting the LBP values ​​of all pixels in the distorted image, and the global image texture features include the contrast, ASM energy and homogeneity of the grayscale co-occurrence matrix.

[0055] Specifically, the texture features of the distorted image include local image texture features and global image texture features; the local image texture features are obtained by statistically calculating the LBP values ​​of all pixels in the distorted image, and the global image texture features include the contrast, ASM energy and homogeneity of the gray-level co-occurrence matrix. The LBP value of the extracted image represents the local texture information of the image, and the contrast, energy and homogeneity of the extracted gray-level co-occurrence matrix represent the global texture features of the image.

[0056] The local texture features of an image can be obtained through the relationship between the central pixel of the image and the adjacent surrounding pixels. For the input image I, its central pixel n c With adjacent surrounding pixels n i The relationship between can be expressed as Among them, s(·,·) is the local binary pattern (LBP) threshold function, I(n c ) and I(n i ) represent the center pixel n c With adjacent surrounding pixels n i The grayscale value of the pixel at n c and its surrounding Q pixels n i After the calculation is completed, LBP is used for binary coding to obtain Further improving the formula, we get: Among them, riu2 represents the rotation invariant uniform mode when the uniform measurement μ is less than 2, and μ is used to calculate the number of changes in the comparison relationship between the central pixel and the surrounding pixels. The calculation formula is:

[0057] After the above processing steps, the comparison relationship between a pixel and its surrounding Q points can get Q+2 possibilities. The LBP of all pixels in the image is counted, and these Q+2 values ​​are extracted to represent the local texture features of the image.

[0058] As an improvement of the above solution, the gray level co-occurrence matrix is ​​obtained in the following way:

[0059] Compressing and quantizing the grayscale range of the distorted image to a level of L to obtain a quantization matrix; wherein the input size of the distorted image is H×W, and L≥1;

[0060] Counting the number of occurrences of a pixel whose distance from a pixel whose pixel value is i in the quantization matrix is ​​d, whose direction is θ, and whose pixel value is j; wherein 1≤i≤L, 1≤j≤L, i≠j;

[0061] The quantization matrix is ​​traversed to obtain the gray-level co-occurrence matrix.

[0062] Specifically, the grayscale range of the distorted image is compressed and quantized to level L, and the quantization matrix M is obtained. H×W ; Among them, the input size of the distorted image I is H×W, L≥1; H is the image height, and W is the image width.

[0063] Statistics and Quantization Matrix M H×W The number of times a pixel with a pixel value of i appears at a distance of d and a direction of θ, and with a pixel value of j; where 1≤i≤L, 1≤j≤L, i≠j;

[0064] Traverse the quantization matrix M H×W , and get the gray-level co-occurrence matrix H d,θ .

[0065] As an improvement of the above scheme, the calculation formula of the contrast of the gray level co-occurrence matrix is: The calculation formula of the ASM energy of the gray level co-occurrence matrix is: The calculation formula of the homogeneity of the gray level co-occurrence matrix is ​​G H =P d,θ (i,j) / (1+|ij-2|); where, G C is the contrast of the gray-level co-occurrence matrix, P d,θ (i, j) is the element h with coordinates (i, j) in the gray-level co-occurrence matrix d,θ The probability corresponding to (i,j), G E is the ASM energy of the gray-level co-occurrence matrix, G H is the homogeneity of the gray level co-occurrence matrix.

[0066] Specifically, the gray-level co-occurrence matrix H d,θ The contrast ratio is calculated as Gray-level co-occurrence matrix H d,θ The calculation formula of ASM energy is: Gray-level co-occurrence matrix H d,θ The formula for calculating the homogeneity of G is H =P d,θ (i,j) / (1+|ij-2|); where, G C is the gray-level co-occurrence matrix H d,θ The contrast ratio, P d,θ(i,j) is the gray level co-occurrence matrix H d,θ The element h with coordinates (i,j) d,θ The probability corresponding to (i,j), G E is the gray-level co-occurrence matrix H d,θ ASM energy, G H is the gray-level co-occurrence matrix H d,θ homogeneity.

[0067] As an improvement of the above solution, the exposure of the distorted image is quantified by information entropy.

[0068] Specifically, the exposure of the distorted image is quantified by information entropy. Abnormal image exposure will also affect the image quality, and then affect the image understanding process. When the image exposure is abnormal, the grayscale value of the image is concentrated on the bright and dark ends, and the detail information is seriously lost. To reflect this feature, the present invention uses information entropy to reflect the degree of detail information retention in the inspection image. The expression of information entropy is Where U represents the maximum possible grayscale value of all pixels in the distorted image. For common 8-bit images, U = 255. μ Represents the probability that the pixel value is μ.

[0069] In order to deepen the understanding of the image quality assessment model, this embodiment of the present invention also describes the support vector regression SVR, as follows:

[0070] Support vector regression SVR in machine learning is applied to map the image feature vector space F to the score vector space M. The objective function of SVR is defined as follows:

[0071]

[0072] swt T Φ(x i )-y i ≤ε+ξ i ,

[0073]

[0074]

[0075] Among them, x i Represents the feature vector of the i-th distorted image, y i Indicates the evaluation score corresponding to the distorted image. K(x i ,y i )≡φ(x i ) T φ(x i ) represents the kernel function used. The present invention selects the radial basis function RBF function as the kernel function, that is,

[0076] After the mapping relationship is established, the image to be evaluated is input to extract the corresponding image features, and the corresponding predicted quality score can be obtained by establishing a good image quality evaluation model.

[0077] After the trained image quality evaluation model is obtained through training, the effectiveness of the image quality evaluation model is verified through testing. In order to test the performance of the algorithm, the Pearson linear correlation coefficient PLCC, the Spearman rank correlation coefficient SRCC, the Kendall rank correlation coefficient KRCC and the root mean square error RMSE are used as evaluation criteria. The closer the PLCC, SRCC and KRCC are to 1, and the closer the RMSE is to 0, the better the algorithm performance. The present invention has been experimentally tested, and the experimental results show that in the constructed transmission line inspection image quality evaluation database, the algorithm performance indicators reach PLCC=0.9794, SRCC=0.9368, KRCC=0.8133, and RMSE=0.0718, respectively, indicating that the obtained image quality evaluation model has good quality prediction capabilities. See Figure 2 , is a flow chart of another embodiment of an image quality evaluation method provided by the present invention.

[0078] See also Figure 3 , is a schematic diagram of the structure of an embodiment of an image quality evaluation device provided by the present invention, the device comprising:

[0079] The image acquisition module 11 is used to acquire n standard images and perform distortion processing on the standard images to obtain m distorted images; wherein m≥n≥1;

[0080] An image evaluation module 12 is used to evaluate each of the distorted images using a full reference image quality evaluation method to obtain an evaluation score for each of the distorted images, and to obtain a scoring vector space based on the evaluation scores of all the distorted images;

[0081] A feature extraction module 13 is used to extract the structural features, texture features and exposure of each distorted image to obtain a feature vector of each image, and obtain an image feature vector space based on the feature vectors of all the distorted images;

[0082] A model building module 14 is used to establish a functional relationship between the score vector space and the image feature vector space by using support vector regression to obtain an image quality evaluation model;

[0083] The testing module 15 is used to input the image to be evaluated and obtain the quality score of the image to be evaluated using the image quality evaluation model.

[0084] An image quality assessment device provided by an embodiment of the present invention can implement all processes of the image quality assessment method described in any of the above embodiments. The functions of each module and unit in the device and the technical effects achieved are respectively the same as the functions and technical effects achieved by the image quality assessment method described in the above embodiments, and will not be repeated here.

[0085] See also Figure 4 , is a structural diagram of an embodiment of an apparatus for using an image quality assessment method provided by the present invention, wherein the apparatus for using an image quality assessment method comprises a processor 10, a memory 20, and a computer program stored in the memory 20 and configured to be executed by the processor 10, and the processor 10 implements the image quality assessment method described in any of the above embodiments when executing the computer program.

[0086] Exemplarily, the computer program may be divided into one or more modules / units, one or more modules / units are stored in the memory 20 and executed by the processor 10 to complete the present invention. One or more modules / units may be a series of computer program instruction segments capable of completing specific functions, and the instruction segments are used to describe the execution process of the computer program in an image quality assessment method. For example, the computer program may be divided into an image acquisition module, an image assessment module, a feature extraction module, a model building module, and a test module, and the specific functions of each module are as follows:

[0087] The image acquisition module 11 is used to acquire n standard images and perform distortion processing on the standard images to obtain m distorted images; wherein m≥n≥1;

[0088] An image evaluation module 12 is used to evaluate each of the distorted images using a full reference image quality evaluation method to obtain an evaluation score for each of the distorted images, and to obtain a scoring vector space based on the evaluation scores of all the distorted images;

[0089] A feature extraction module 13 is used to extract the structural features, texture features and exposure of each distorted image to obtain a feature vector of each image, and obtain an image feature vector space based on the feature vectors of all the distorted images;

[0090] A model building module 14 is used to establish a functional relationship between the score vector space and the image feature vector space by using support vector regression to obtain an image quality evaluation model;

[0091] The testing module 15 is used to input the image to be evaluated and obtain the quality score of the image to be evaluated using the image quality evaluation model.

[0092] The device using the image quality evaluation method may be a computing device such as a desktop computer, a notebook, a PDA, and a cloud server. The device using the image quality evaluation method may include, but is not limited to, a processor and a memory. Those skilled in the art will appreciate that the schematic diagram Figure 4 It is merely an example of an apparatus using an image quality assessment method and does not constitute a limitation of the apparatus using the image quality assessment method. It may include more or fewer components than those shown in the figure, or a combination of certain components, or different components. For example, the apparatus using the image quality assessment method may also include input and output devices, network access devices, buses, etc.

[0093] The processor 10 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor 10 may also be any conventional processor, etc. The processor 10 is the control center of the device using the image quality assessment method, and uses various interfaces and lines to connect the various parts of the entire device using the image quality assessment method.

[0094] The memory 20 can be used to store the computer program and / or module. The processor 10 implements various functions of the device using the image quality evaluation method by running or executing the computer program and / or module stored in the memory 20 and calling the data stored in the memory 20. The memory 20 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory 20 can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0095] Wherein, if the module integrated in the device using the image quality evaluation method is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The above-mentioned computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned method embodiments can be implemented. Wherein, the above-mentioned computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electric carrier signals and telecommunication signals.

[0096] An embodiment of the present invention further provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the image quality evaluation method described in any of the above embodiments.

[0097] In summary, the image quality evaluation method, device and storage medium provided by the embodiments of the present invention first establish a dedicated database for image quality evaluation, then construct an image feature vector by extracting the structural features, texture features and exposure features of the image, and then use the support vector machine to learn the mapping model, so that the quality evaluation score of the image to be evaluated can be obtained according to the mapping model. The present invention can provide a way to evaluate the quality of transmission line inspection images, so as to quickly and accurately predict the quality of inspection images and meet the needs of smart grid construction.

[0098] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for evaluating image quality, characterized in that: The following steps are involved: Obtain n standard images, and apply distortion processing to the standard images to obtain m distorted images; wherein m≥n≥1; Using a full-reference image quality evaluation method to evaluate each of the distorted images, obtaining an evaluation score for each of the distorted images, and obtaining a scoring vector space based on the evaluation scores of all of the distorted images; Extracting structural features, texture features and exposure of each of the distorted images to obtain a feature vector of each image, and obtaining an image feature vector space based on the feature vectors of all the distorted images; Using support vector regression to establish a functional relationship between the score vector space and the image feature vector space to obtain an image quality evaluation model; Input an image to be evaluated, and obtain a quality score of the image to be evaluated using the image quality evaluation model; Among them, the standard image and the image to be evaluated are transmission line inspection images; the structural features of the distorted image include the shape parameters and scale parameters of the Weibull distribution corresponding to the gradient amplitude of the distorted image; the texture features of the distorted image include local image texture features and global image texture features.

2. The image quality evaluation method according to claim 1, characterized in that: The step of acquiring n standard images and applying distortion processing to the standard images to obtain m distorted images specifically includes: Obtain n standard images, apply preset multiple different degrees of Gaussian noise to each of the standard images, apply preset multiple different degrees of Gaussian blur to each of the standard images, apply preset multiple different degrees of compression processing to each of the standard images, and obtain m distorted images; wherein m>n≥1.

3. The image quality evaluation method according to claim 1, wherein: The local image texture features are obtained by counting the LBP values ​​of all pixels in the distorted image, and the global image texture features include the contrast, ASM energy and homogeneity of the gray level co-occurrence matrix.

4. The image quality evaluation method according to claim 3, characterized in that: The gray level co-occurrence matrix is ​​obtained in the following way: Compressing and quantizing the grayscale range of the distorted image to a level of L to obtain a quantization matrix; wherein the input size of the distorted image is H×W, and L≥1; Counting the number of occurrences of a pixel whose distance from a pixel whose pixel value is i in the quantization matrix is ​​d, whose direction is θ, and whose pixel value is j; wherein 1≤i≤L, 1≤j≤L, i≠j; The quantization matrix is ​​traversed to obtain the gray-level co-occurrence matrix.

5. The image quality evaluation method according to claim 4, characterized in that: The calculation formula of the contrast of the gray level co-occurrence matrix is: The calculation formula of the ASM energy of the gray level co-occurrence matrix is: The calculation formula of the homogeneity of the gray level co-occurrence matrix is ​​G H =P d,θ (i,j) / (1+|ij-2|); where, G C is the contrast of the gray-level co-occurrence matrix, P d,θ (i, j) is the element h with coordinates (i, j) in the gray-level co-occurrence matrix d,θ The probability corresponding to (i,j), G E is the ASM energy of the gray-level co-occurrence matrix, G H is the homogeneity of the gray level co-occurrence matrix.

6. The image quality evaluation method according to claim 1, wherein: The exposure of the distorted image is quantified by information entropy.

7. An image quality evaluation device, characterized in that: include: An image acquisition module, used for acquiring n standard images, and applying distortion processing to the standard images to obtain m distorted images; wherein m≥n≥1; An image evaluation module, used for evaluating each of the distorted images by using a full reference image quality evaluation method to obtain an evaluation score for each of the distorted images, and obtaining a scoring vector space according to the evaluation scores of all the distorted images; A feature extraction module, used to extract structural features, texture features and exposure of each distorted image to obtain a feature vector of each image, and to obtain an image feature vector space based on the feature vectors of all the distorted images; A model building module, used to establish a functional relationship between the score vector space and the image feature vector space by using support vector regression to obtain an image quality evaluation model; A testing module, used for inputting an image to be evaluated, and obtaining a quality score of the image to be evaluated using the image quality evaluation model; Among them, the standard image and the image to be evaluated are transmission line inspection images; the structural features of the distorted image include the shape parameters and scale parameters of the Weibull distribution corresponding to the gradient amplitude of the distorted image; the texture features of the distorted image include local image texture features and global image texture features.

8. A device using an image quality assessment method, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements the image quality assessment method according to any one of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the image quality assessment method according to any one of claims 1 to 6.

Citation Information

Cited By

  • Context-aware image enhancement by an electronic device

    WO2026141777A1