Intelligent image optimization processing system based on big data

Abstract

The invention discloses an intelligent image optimization processing system based on big data, and relates to the technical field of image optimization, the system comprises an intelligent terminal, and the intelligent terminal is in communication connection with an image acquisition module, an image processing module, an image analysis module and an image optimization module. A plurality of to-be-corrected images are acquired through an image acquisition module, and image denoising, image dimension reduction, gray processing and feature parameter extraction are performed on the plurality of to-be-corrected images through an image processing module, so that a parameter set of each to-be-corrected image is generated; data analysis is carried out on the parameter set of the plurality of to-be-corrected images through the image analysis module, then the to-be-corrected images needing image optimization are positioned, and the corresponding to-be-corrected images are marked as to-be-optimized images to be transmitted to the intelligent terminal; and uploading the optimization parameter set corresponding to each to-be-optimized image to an intelligent terminal through an image optimization module so as to intelligently optimize each to-be-optimized image into a final image.

Description

Technical Field

[0001] The present invention relates to the technical field of image optimization, and in particular to an image intelligent optimization processing system based on big data. Background Art

[0002] Image optimization refers to the process of improving the visual quality, adaptability and efficiency of images through various technical means and methods. Its main goal is to make the image perform better in different applications and scenarios. It specifically includes the following aspects: visual quality improvement, file size optimization, adaptability enhancement, specific application requirements, and automated processing and intelligent technology. Image optimization is not limited to a single technical operation, but is a comprehensive process involving comprehensive consideration of image processing, algorithm optimization, visual perception and application requirements.

[0003] How to perform intelligent optimization of images more perfectly to obtain images that better meet the requirements, how to significantly improve the visual effects and application performance of images through reasonable optimization, and meet the needs and expectations in different scenarios. Now we use an image intelligent optimization processing system based on big data to process images. Summary of the invention

[0004] In order to solve the above problems, the purpose of the present invention is to provide an image intelligent optimization processing system based on big data.

[0005] The object of the present invention can be achieved by the following technical solutions: an image intelligent optimization processing system based on big data, comprising an intelligent terminal, wherein the intelligent terminal is communicatively connected with an image acquisition module, an image processing module, an image analysis module and an image optimization module;

[0006] The image acquisition module is used to acquire a number of images to be corrected;

[0007] The image processing module is used to perform image denoising, image dimension reduction, grayscale processing and feature parameter extraction on a plurality of images to be corrected, thereby generating a parameter set for each image to be corrected;

[0008] The image analysis module is used to perform data analysis on a parameter set of a plurality of images to be corrected, thereby locating the images to be corrected that need to be optimized, and marking the corresponding images to be corrected as images to be optimized and transmitting them to the intelligent terminal;

[0009] The image optimization module is used to upload the optimization parameter set corresponding to each image to be optimized to the intelligent terminal, and then intelligently optimize each image to be optimized into a final image.

[0010] Furthermore, the process of collecting a plurality of images to be corrected includes:

[0011] A number of images to be corrected are numbered and recorded as i, i = 1, 2, 3, ..., n, where n is a natural number greater than 0, and the image data volume corresponding to the image to be corrected numbered i is obtained, which is recorded as C[i]. The image acquisition module is provided with a number of acquisition nodes, and the number of acquisition nodes are numbered and recorded as j, where j = 1, 2, 3, ..., m, where m is a natural number greater than 0, and each acquisition node is provided with a uniform node capacity;

[0012] Construct a mapping relationship between the image to be corrected numbered i and the acquisition node numbered j, and record the mapping relationship as R, R =<i,j> , the node capacity of the collection node is C 采集 , when C[i]≤C 采集 When C[i]>C 采集 When , the capacity of the collection node is expanded. The capacity expansion is performed by associating the collection node with the data auxiliary node. The node expansion capacity of the data auxiliary node is recorded as C 扩充 , the data auxiliary node and the collection node are integrated, and the integrated collection node meets: C 采集 +C 扩充 ≥C[i].

[0013] Furthermore, the process of performing image denoising and image dimension reduction on a plurality of images to be corrected includes:

[0014] Set the noise reduction period and dimension reduction period of the image processing module;

[0015] During the noise reduction period, image denoising is performed on a plurality of images to be corrected. The content of the image denoising is as follows: each image to be corrected is divided into a plurality of sub-regions to be corrected with equal pixel block sizes, the noise value of the pixel at the center of each sub-region to be corrected is obtained as the reference noise, the noise values ​​of all other pixels except the center of each sub-region to be corrected are unified as the reference noise, and the above operation is repeated until the unification of all noise values ​​corresponding to the plurality of sub-regions to be corrected is completed;

[0016] During the dimensionality reduction period, image dimensionality reduction is performed on several images to be corrected, and the images to be corrected are divided into several sub-regions of different dimensions. The actual dimension of each sub-region is compared with the preset expected dimension, and the sub-regions whose actual dimensions do not match the expected dimensions are marked as regions to be reduced in dimension, and the dimensionality reduction algorithm is used to perform dimensionality reduction on the regions to be reduced in dimension until the actual dimension of each region to be reduced in dimension meets the corresponding expected dimension.

[0017] Furthermore, the process of grayscale processing the image to be corrected includes:

[0018] Set the grayscale operation period of the image processing module;

[0019] During the grayscale operation period, the grayscale values ​​of several pixels corresponding to each image to be corrected are marked accordingly. The grayscale value is obtained by calculating the RGB information of each pixel, where RGB information =<R,G,B> , R represents the brightness corresponding to the red part of the pixel, G represents the brightness corresponding to the green part of the pixel, and B represents the brightness corresponding to the blue part of the pixel, where the value range of all brightness is 0-255;

[0020] Then the grayscale values ​​of several pixels corresponding to each image to be corrected are expressed as follows:

[0021]

[0022] Among them, τ1, τ2 and τ3 represent the number of R, G and B with the highest brightness respectively, and the sum of τ1, τ2 and τ3 represents the total number of all pixels in each image to be corrected.

[0023] Furthermore, the process of extracting feature parameters for a plurality of images to be corrected and then generating a parameter set for each image to be corrected includes:

[0024] Set up a feature extraction program, initialize the feature extraction program, and configure the program parameters of the feature extraction program. The program parameters include program computing power, behavior form, and program operation parameter set. The program computing power is the minimum computing power value to maintain the operation of the feature extraction program. The behavior form is used to limit all operation behaviors allowed by the feature extraction program. The program operation parameter set is used to indicate the corresponding operation parameters when the feature extraction program is running normally. The program operation parameter set includes program execution time and program correct execution rate;

[0025] When the program computing power meets the preset maintenance computing power, there is no unauthorized operation behavior in the behavior form, and the program execution time and program correct execution rate in the program operation parameter set are within their respective preset normal value ranges, the program initialization of the feature extraction program is completed. Otherwise, the program initialization of the feature extraction program continues until all the program parameters are met.

[0026] The image parameters corresponding to several images to be corrected are obtained through a feature extraction program. The image parameters include image texture features, image color features, image shape features and image basic statistical features. The image texture features, image color features, image shape features and image basic statistical features of each image to be corrected are summarized to generate a parameter set for each image to be corrected.

[0027] Furthermore, the process of performing data analysis on parameter sets of a plurality of images to be corrected and then locating the images to be corrected that need to be optimized includes:

[0028] The image texture features, image color features and image basic statistical features in the parameter set corresponding to each image to be corrected are respectively used as analysis items of data analysis, and the image basic statistical features, image texture features and image color features in the analysis items are analyzed in turn. The image basic statistical features include image size, image resolution, image contrast and image saturation. When the image size, image resolution, image contrast and image saturation all meet their respective preset conditions, it means that the image basic statistical features are qualified, otherwise, the image basic statistical features are unqualified;

[0029] The image texture feature includes a texture mean and a texture standard deviation. When the texture mean is less than a preset mean threshold, and the texture standard deviation is less than a preset standard deviation threshold, the image texture feature is qualified; otherwise, the image texture feature is unqualified.

[0030] The image color features include RGB mean, grayscale mean and color variance. When the RGB mean, grayscale mean and color variance all meet their respective preset limiting conditions, the image color features are qualified; otherwise, the image color features are unqualified.

[0031] An image to be corrected whose basic statistical features, image texture features and image color features are unqualified is positioned as an image to be corrected that needs to be optimized, and an image to be corrected whose other basic statistical features, image texture features and image color features are all qualified is positioned as a qualified image, and image optimization is not performed on the qualified image.

[0032] Furthermore, the process of marking the image to be corrected that needs to be optimized as the image to be optimized and transmitting it to the smart terminal includes:

[0033] Acquire all the images to be corrected that need to be optimized, mark the images to be corrected that need to be optimized as images to be optimized, perform data stream compression for each image to be optimized, and then generate an image data stream corresponding to each image to be optimized;

[0034] The image data stream is composed of several frame unit node streams. The frame number corresponding to the unit node stream is obtained and recorded as S. The transmission speed of each image data stream is set according to the frame number of the unit node stream corresponding to the image data stream, and the transmission speed is recorded as V. 传输 , then V 传输 =S×V 单位 , where V 单位 The unit transmission speed corresponding to each frame unit node flow is used, and each image to be optimized is transmitted to the intelligent terminal for data storage according to the corresponding transmission speed.

[0035] Furthermore, the process of uploading the optimization parameter set corresponding to each image to be optimized to the intelligent terminal and then intelligently optimizing each image to be optimized into a final image includes:

[0036] A number of images to be optimized are numbered and recorded as k, k = 1, 2, 3, ..., p, p is a natural number greater than 0, and an optimization parameter set corresponding to each image to be optimized is set, and the optimization parameter set of the image to be optimized numbered k is marked as σ[k], σ[k] = {η1, η2, η3, μ}, η1 = (η 11 , η 12 , η 13 , η 14 ), η 11 , η 12 , η 13 and η 14 represent the restoration parameters of the image size, image resolution, image contrast and image saturation of the image to be optimized, respectively, η2=(η 21 , η 22 ), η 21 and η 22 They respectively represent processing parameters for local statistical feature processing and frequency processing of the image texture feature to be optimized. Through the local statistical feature processing and frequency processing, the texture mean and texture standard deviation corresponding to the image texture feature are within the respective mean threshold and standard deviation threshold, thereby making the image texture feature qualified;

[0037] η3=(f(η 31 ), f(η 32 ), f(η 33 )), where η 31 , η 32 and η 33 Respectively represent the RGB value, gray value and color variance of the image to be optimized, f(η 31 ) represents the RGB correction function, f(η 32 ) represents the grayscale correction function, f(η 33 ) represents the color correction function;

[0038] μ is used to characterize the expectation degree after the image to be optimized is intelligently optimized by η1, η2 and η3, to judge whether the expectation degree is higher than the set expectation threshold, and to perform different operations corresponding to μ according to the judgment result;

[0039] The image optimization module transmits the optimization parameter set corresponding to each image to be optimized to the intelligent terminal, and the intelligent terminal performs intelligent optimization on the image to be optimized using the corresponding optimization parameter set. The expected threshold is denoted as μ 期望 , if μ ≥ μ 期望 , then directly generate the final image, if μ<μ期望 , then reconfigure the optimization parameter set of the corresponding image to be optimized until the corresponding final image is generated.

[0040] Compared with the prior art, the beneficial effects of the present invention are as follows: a plurality of images to be corrected are collected by an image acquisition module, and an image processing module performs image denoising, image dimension reduction, grayscale processing and feature parameter extraction on the plurality of images to be corrected, thereby generating a parameter set for each image to be corrected, and an image analysis module performs data analysis on the parameter sets of the plurality of images to be corrected, thereby locating the images to be corrected that need to be optimized, and marking the corresponding images to be corrected as images to be optimized and transmitting them to an intelligent terminal, and finally an image optimization module uploads the optimization parameter set corresponding to each image to be optimized to the intelligent terminal, thereby intelligently optimizing each image to be optimized into a final image, thereby achieving high-efficiency optimization of a large number of images and improving the image quality to meet the corresponding requirements of different scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a schematic diagram of the present invention. DETAILED DESCRIPTION

[0042] like Figure 1 As shown, an image intelligent optimization processing system based on big data includes an intelligent terminal, and the intelligent terminal is communicatively connected with an image acquisition module, an image processing module, an image analysis module and an image optimization module;

[0043] The image acquisition module is used to acquire a number of images to be corrected;

[0044] The image processing module is used to perform image denoising, image dimension reduction, grayscale processing and feature parameter extraction on a plurality of images to be corrected, thereby generating a parameter set for each image to be corrected;

[0045] The image analysis module is used to perform data analysis on a parameter set of a plurality of images to be corrected, thereby locating the images to be corrected that need to be optimized, and marking the corresponding images to be corrected as images to be optimized and transmitting them to the intelligent terminal;

[0046] The image optimization module is used to upload the optimization parameter set corresponding to each image to be optimized to the intelligent terminal, and then intelligently optimize each image to be optimized into a final image.

[0047] It should be further explained that, in a specific implementation process, the process of the image acquisition module acquiring a plurality of images to be corrected includes:

[0048] A number of images to be corrected are numbered and the number is recorded as i, then i=1, 2, 3, ..., n, where n is a natural number greater than 0, and the image data volume corresponding to the image to be corrected numbered i is obtained, which is recorded as C[i];

[0049] The image acquisition module is provided with a plurality of acquisition nodes, the plurality of acquisition nodes are numbered and the number is recorded as j, where j=1, 2, 3, ..., m, wherein m is a natural number greater than 0, and each acquisition node is provided with a uniform node capacity;

[0050] Construct a mapping relationship between the image to be corrected numbered i and the acquisition node numbered j, and record the mapping relationship as R, then R =<i,j> , the node capacity of the collection node is C 采集 , when C[i]≤C 采集 When C[i]>C 采集 When , the capacity of the collection node is expanded. The capacity expansion is performed by associating the collection node with the data auxiliary node. The node expansion capacity of the data auxiliary node is recorded as C 扩充 The data auxiliary node is integrated with the collection node, and the integrated collection node meets the following conditions: C 采集 +C 扩充 ≥C[i].

[0051] It should be further explained that, in a specific implementation process, the process of the image processing module performing image denoising, image dimension reduction and grayscale processing on a plurality of images to be corrected includes:

[0052] Set the noise reduction period, dimension reduction period and grayscale operation period of the image processing module, and record the noise reduction period, dimension reduction period and grayscale operation period as T 降噪 , T 降维 and T 灰度 , where T 降噪 =[t0, t1], T 降维 =(t1, t2], T 灰度 =(t2, t3], where t0 is the start time of the noise reduction period, t1 is the connection time between the noise reduction period and the dimension reduction period, t2 is the connection time between the dimension reduction period and the grayscale operation period, and t3 is the end time of the grayscale operation period;

[0053] During the noise reduction period, image denoising is performed on a plurality of images to be corrected. The content of the image denoising is as follows: each image to be corrected is divided into a plurality of sub-regions to be corrected with equal pixel block sizes, the noise value of the pixel at the center of each sub-region to be corrected is obtained as the reference noise, the noise values ​​of all other pixels except the center of each sub-region to be corrected are unified as the reference noise, and the above operation is repeated until the unification of all noise values ​​corresponding to the plurality of sub-regions to be corrected is completed;

[0054] During the dimension reduction period, image dimension reduction is performed on a number of images to be corrected, the images to be corrected are divided into a number of sub-regions of different dimensions, and the actual dimension of each sub-region is compared with the preset expected dimension, and the sub-regions whose actual dimensions do not match the expected dimensions are marked as regions to be reduced in dimension, and the regions to be reduced in dimension are reduced in dimension by using a dimension reduction algorithm until the actual dimension of each region to be reduced in dimension meets the corresponding expected dimension;

[0055] During the grayscale operation period, the grayscale values ​​of several pixels corresponding to each image to be corrected are marked accordingly. The grayscale value is obtained by calculating the RGB information of each pixel, where RGB information =<R,G,B> , R represents the brightness corresponding to the red part of the pixel, G represents the brightness corresponding to the green part of the pixel, and B represents the brightness corresponding to the blue part of the pixel;

[0056] Among them, the value range of all brightness is 0-255;

[0057] Then the grayscale values ​​of several pixels corresponding to each image to be corrected are expressed as follows:

[0058]

[0059] Among them, τ1, τ2 and τ3 represent the number of R, G and B with the highest brightness respectively, and the sum of τ1, τ2 and τ3 represents the total number of all pixels in each image to be corrected.

[0060] It should be further explained that, in a specific implementation process, the process of extracting feature parameters for a plurality of images to be corrected and then generating a parameter set for each image to be corrected includes:

[0061] A feature extraction program for extracting feature parameters for the image to be corrected is set, and the feature extraction program is initialized. The program initialization content is: configuring program parameters of the feature extraction program, the program parameters include program computing power, behavior form and program operation parameter set, the program computing power is the minimum computing power value to maintain the operation of the feature extraction program, the behavior form is used to limit all operation behaviors allowed by the feature extraction program, and the program operation parameter set is used to represent the corresponding operation parameters when the feature extraction program is running normally, and the program operation parameter set includes program execution time and program correct execution rate;

[0062] When the program computing power meets the preset maintenance computing power, there is no unauthorized operation behavior in the behavior form, and the program execution time and program correct execution rate in the program operation parameter set are within their respective preset normal value ranges, the program initialization of the feature extraction program is completed. Otherwise, the program initialization of the feature extraction program continues until all the program parameters are met.

[0063] The image parameters corresponding to several images to be corrected are obtained through a feature extraction program, and the image parameters include image texture features, image color features, image shape features and image basic statistical features. The image texture features, image color features, image shape features and image basic statistical features of each image to be corrected are summarized to generate a parameter set for each image to be corrected.

[0064] It should be further explained that, in a specific implementation process, the image analysis module performs data analysis on a parameter set of a plurality of images to be corrected, and then locates the image to be corrected that needs to be optimized, including:

[0065] The image analysis module obtains parameter sets corresponding to a plurality of images to be corrected, and uses image texture features, image color features and image basic statistical features in the parameter sets as analysis items for data analysis;

[0066] The basic statistical features of the image, the image texture features and the image color features in the analysis project are analyzed in sequence, wherein the basic statistical features of the image include image size, image resolution, image contrast and image saturation. When the image size, image resolution, image contrast and image saturation all meet their respective preset conditions, it means that the basic statistical features of the image are qualified, otherwise, the basic statistical features of the image are unqualified;

[0067] The image texture feature includes a texture mean and a texture standard deviation. When the texture mean is less than a preset mean threshold, and the texture standard deviation is less than a preset standard deviation threshold, the image texture feature is qualified; otherwise, the image texture feature is unqualified.

[0068] The image color features include RGB mean, grayscale mean and color variance. When the RGB mean, grayscale mean and color variance all meet their respective preset limiting conditions, the image color features are qualified; otherwise, the image color features are unqualified.

[0069] An image to be corrected whose basic statistical features, image texture features and image color features are unqualified is positioned as an image to be corrected that needs to be optimized, and an image to be corrected whose other basic statistical features, image texture features and image color features are all qualified is positioned as a qualified image, and image optimization is not performed on the qualified image.

[0070] It should be further explained that, in a specific implementation process, the process of marking the image to be corrected that needs to be optimized as the image to be optimized and transmitting it to the smart terminal includes:

[0071] Acquire all the images to be corrected that need to be optimized, mark the images to be corrected that need to be optimized as images to be optimized, perform data stream compression for each image to be optimized, and then generate an image data stream corresponding to each image to be optimized;

[0072] The image data stream is composed of several frame unit node streams. The frame number corresponding to the unit node stream is obtained, and the frame number is recorded as S. The transmission speed of each image data stream is set according to the frame number of the unit node stream corresponding to the image data stream, and the transmission speed is recorded as V. 传输 , then V 传输 =S×V 单位 , where V 单位 For the unit transmission speed corresponding to each frame unit node flow, each image to be optimized is transmitted to the intelligent terminal at the corresponding transmission speed, and the data is stored at the intelligent terminal.

[0073] It should be further explained that, in a specific implementation process, the image optimization module uploads the optimization parameter set corresponding to each image to be optimized to the intelligent terminal, and then the process of intelligently optimizing each image to be optimized into a final image includes:

[0074] A number of images to be optimized are numbered and the number is recorded as k, then k = 1, 2, 3, ..., p, where p is a natural number greater than 0, and an optimization parameter set corresponding to each image to be optimized is set, and the optimization parameter set of the image to be optimized numbered k is marked as σ[k];

[0075] Among them, σ[k]={eta1, eta2, eta3, μ}, eta1=(n 11 , η 12 , η 13 , η 14 ), η11 , η 12 , η 13 and η 14 represent the restoration parameters of the image size, image resolution, image contrast and image saturation of the image to be optimized, respectively, η2=(η 21 , η 22 ), η 21 and η 22 They respectively represent processing parameters for local statistical feature processing and frequency processing of the image texture feature to be optimized. Through the local statistical feature processing and frequency processing, the texture mean and texture standard deviation corresponding to the image texture feature are within the respective mean threshold and standard deviation threshold, thereby making the image texture feature qualified;

[0076] η3=(f(η 31 ), f(η 32 ), f(η 33 )), where η 31 , η 32 and η 33 Respectively represent the RGB value, gray value and color variance of the image to be optimized, f(η 31 ) represents the RGB correction function, f(η 32 ) represents the grayscale correction function, f(η 33 ) represents the color correction function;

[0077] μ is used to characterize the expectation degree after the image to be optimized is intelligently optimized by η1, η2 and η3, to judge whether the expectation degree is higher than the set expectation threshold, and to perform different operations corresponding to μ according to the judgment result;

[0078] The image optimization module transmits the optimization parameter set corresponding to each image to be optimized to the intelligent terminal, and the intelligent terminal performs intelligent optimization on the image to be optimized using the corresponding optimization parameter set. The expected threshold is denoted as μ 期望 , if μ ≥ μ 期望 , then directly generate the final image, if μ<μ 期望 , then reconfigure the optimization parameter set of the corresponding image to be optimized until the corresponding final image is generated.

[0079] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An image intelligent optimization processing system based on big data, including an intelligent terminal, characterized in that: The intelligent terminal is communicatively connected with an image acquisition module, an image processing module, an image analysis module and an image optimization module; The image acquisition module is used to acquire a number of images to be corrected; The image processing module is used to perform image denoising, image dimension reduction, grayscale processing and feature parameter extraction on a plurality of images to be corrected, thereby generating a parameter set for each image to be corrected; The image analysis module is used to perform data analysis on a parameter set of a plurality of images to be corrected, thereby locating the images to be corrected that need to be optimized, and marking the corresponding images to be corrected as images to be optimized and transmitting them to the intelligent terminal; The image optimization module is used to upload the optimization parameter set corresponding to each image to be optimized to the intelligent terminal, and then intelligently optimize each image to be optimized into a final image.

2. According to the big data-based image intelligent optimization processing system of claim 1, it is characterized in that: The process of acquiring several images to be corrected includes: A number of images to be corrected are numbered and recorded as i, i = 1, 2, 3, ..., n, where n is a natural number greater than 0, and the image data volume corresponding to the image to be corrected numbered i is obtained, which is recorded as C[i]. The image acquisition module is provided with a number of acquisition nodes, and the number of acquisition nodes are numbered and recorded as j, where j = 1, 2, 3, ..., m, where m is a natural number greater than 0, and each acquisition node is provided with a uniform node capacity; Construct a mapping relationship between the image to be corrected numbered i and the acquisition node numbered j, and record the mapping relationship as R, R =<i,j> , the node capacity of the collection node is C 采集 , when C[i]≤C 采集 When C[i]>C 采集 When , the capacity of the collection node is expanded. The capacity expansion is performed by associating the collection node with the data auxiliary node. The node expansion capacity of the data auxiliary node is recorded as C 扩充 , the data auxiliary node and the collection node are integrated, and the integrated collection node meets: C 采集 +C 扩充 ≥C[i].

3. The image intelligent optimization processing system based on big data according to claim 2 is characterized in that: The process of image denoising and image dimensionality reduction for several images to be corrected includes: Set the noise reduction period and dimension reduction period of the image processing module; During the noise reduction period, image denoising is performed on a plurality of images to be corrected. The content of the image denoising is as follows: each image to be corrected is divided into a plurality of sub-regions to be corrected with equal pixel block sizes, the noise value of the pixel at the center of each sub-region to be corrected is obtained as the reference noise, the noise values ​​of all other pixels except the center of each sub-region to be corrected are unified as the reference noise, and the above operation is repeated until the unification of all noise values ​​corresponding to the plurality of sub-regions to be corrected is completed; During the dimensionality reduction period, image dimensionality reduction is performed on several images to be corrected, and the images to be corrected are divided into several sub-regions of different dimensions. The actual dimension of each sub-region is compared with the preset expected dimension, and the sub-regions whose actual dimensions do not match the expected dimensions are marked as regions to be reduced in dimension, and the dimensionality reduction algorithm is used to perform dimensionality reduction on the regions to be reduced in dimension until the actual dimension of each region to be reduced in dimension meets the corresponding expected dimension.

4. The image intelligent optimization processing system based on big data according to claim 3 is characterized in that: The process of grayscale processing of the image to be corrected includes: Set the grayscale operation period of the image processing module; During the grayscale operation period, the grayscale values ​​of several pixels corresponding to each image to be corrected are marked accordingly. The grayscale value is obtained by calculating the RGB information of each pixel, where RGB information =<R,G,B> , R represents the brightness corresponding to the red part of the pixel, G represents the brightness corresponding to the green part of the pixel, and B represents the brightness corresponding to the blue part of the pixel, where the value range of all brightness is 0-255; Then the grayscale values ​​of several pixels corresponding to each image to be corrected are expressed as follows: ； Among them, τ1, τ2 and τ3 represent the number of R, G and B with the highest brightness respectively, and the sum of τ1, τ2 and τ3 represents the total number of all pixels in each image to be corrected.

5. The image intelligent optimization processing system based on big data according to claim 4 is characterized in that: The process of extracting feature parameters for a plurality of images to be corrected and then generating a parameter set for each image to be corrected includes: Set up a feature extraction program, initialize the feature extraction program, and configure the program parameters of the feature extraction program. The program parameters include program computing power, behavior form, and program operation parameter set. The program computing power is the minimum computing power value to maintain the operation of the feature extraction program. The behavior form is used to limit all operation behaviors allowed by the feature extraction program. The program operation parameter set is used to indicate the corresponding operation parameters when the feature extraction program is running normally. The program operation parameter set includes program execution time and program correct execution rate; When the program computing power meets the preset maintenance computing power, there is no unauthorized operation behavior in the behavior form, and the program execution time and program correct execution rate in the program operation parameter set are within their respective preset normal value ranges, the program initialization of the feature extraction program is completed. Otherwise, the program initialization of the feature extraction program continues until all the program parameters are met. The image parameters corresponding to several images to be corrected are obtained through a feature extraction program. The image parameters include image texture features, image color features, image shape features and image basic statistical features. The image texture features, image color features, image shape features and image basic statistical features of each image to be corrected are summarized to generate a parameter set for each image to be corrected.

6. The image intelligent optimization processing system based on big data according to claim 5 is characterized in that: The process of performing data analysis on parameter sets of several images to be corrected and then locating the images to be corrected that need image optimization includes: The image texture features, image color features and image basic statistical features in the parameter set corresponding to each image to be corrected are respectively used as analysis items of data analysis, and the image basic statistical features, image texture features and image color features in the analysis items are analyzed in turn. The image basic statistical features include image size, image resolution, image contrast and image saturation. When the image size, image resolution, image contrast and image saturation all meet their respective preset conditions, it means that the image basic statistical features are qualified, otherwise, the image basic statistical features are unqualified; The image texture feature includes a texture mean and a texture standard deviation. When the texture mean is less than a preset mean threshold, and the texture standard deviation is less than a preset standard deviation threshold, the image texture feature is qualified; otherwise, the image texture feature is unqualified. The image color features include RGB mean, grayscale mean and color variance. When the RGB mean, grayscale mean and color variance all meet their respective preset limiting conditions, the image color features are qualified; otherwise, the image color features are unqualified. An image to be corrected whose basic statistical features, image texture features and image color features are unqualified is positioned as an image to be corrected that needs to be optimized, and an image to be corrected whose other basic statistical features, image texture features and image color features are all qualified is positioned as a qualified image, and image optimization is not performed on the qualified image.

7. The image intelligent optimization processing system based on big data according to claim 6 is characterized in that: The process of marking the image to be corrected that needs to be optimized as the image to be optimized and transmitting it to the smart terminal includes: Acquire all the images to be corrected that need to be optimized, mark the images to be corrected that need to be optimized as images to be optimized, perform data stream compression for each image to be optimized, and then generate an image data stream corresponding to each image to be optimized; The image data stream is composed of several frame unit node streams. The frame number corresponding to the unit node stream is obtained and recorded as S. The transmission speed of each image data stream is set according to the frame number of the unit node stream corresponding to the image data stream, and the transmission speed is recorded as V. 传输 , then V 传输 =S×V 单位 , where V 单位 The unit transmission speed corresponding to each frame unit node flow is used, and each image to be optimized is transmitted to the intelligent terminal for data storage according to the corresponding transmission speed.

8. The image intelligent optimization processing system based on big data according to claim 7 is characterized in that: The process of uploading the optimization parameter set corresponding to each image to be optimized to the intelligent terminal and then intelligently optimizing each image to be optimized into a final image includes: A number of images to be optimized are numbered and recorded as k, k = 1, 2, 3, ..., p, p is a natural number greater than 0, and an optimization parameter set corresponding to each image to be optimized is set, and the optimization parameter set of the image to be optimized numbered k is marked as σ[k], σ[k] = {η1, η2, η3, μ}, η1 = (η 11 , η 12 , η 13 , η 14 ), η 11 , η 12 , η 13 and η 14 represent the restoration parameters of the image size, image resolution, image contrast and image saturation of the image to be optimized, respectively, η2=(η 21 , η 22 ), η 21 and η 22 They respectively represent processing parameters for local statistical feature processing and frequency processing of the image texture feature to be optimized. Through the local statistical feature processing and frequency processing, the texture mean and texture standard deviation corresponding to the image texture feature are within the respective mean threshold and standard deviation threshold, thereby making the image texture feature qualified; η3=(f(η 31 ), f(η 32 ), f(η 33 )), where η 31 , η 32 and η 33 Respectively represent the RGB value, gray value and color variance of the image to be optimized, f(η 31 ) represents the RGB correction function, f(η 32 ) represents the grayscale correction function, f(η 33 ) represents the color correction function; μ is used to characterize the expectation degree after the image to be optimized is intelligently optimized by η1, η2 and η3, to judge whether the expectation degree is higher than the set expectation threshold, and to perform different operations corresponding to μ according to the judgment result; The image optimization module transmits the optimization parameter set corresponding to each image to be optimized to the intelligent terminal, and the intelligent terminal performs intelligent optimization on the image to be optimized using the corresponding optimization parameter set. The expected threshold is denoted as μ 期望 , if μ ≥ μ 期望 , then directly generate the final image, if μ<μ 期望 , then reconfigure the optimization parameter set of the corresponding image to be optimized until the corresponding final image is generated.

Images