Deep Learning-Based Image Adaptive Enhancement Method and System

Abstract

This invention discloses a deep learning-based adaptive image enhancement method and system. The invention relates to the field of image processing and includes an image processing platform comprising an image acquisition module, an image processing module, an image analysis module, a model management module, and an image management module. The image acquisition module is used to input image data and image information. The image processing module is used to acquire image quality data of segmented image regions and corresponding pixels. The image analysis module is used to acquire comprehensive classification results of pixels corresponding to segmented image regions. The model management module is used to construct image processing links and corresponding image processing link models. The image management module is used to acquire region-enhanced images corresponding to segmented image regions based on the image processing links and corresponding image processing link models, and to acquire adaptively enhanced images of the image data. This invention improves the accuracy of the image enhancement process to a certain extent.

Description

Technical Field

[0001] This invention relates to the field of image processing, specifically to an image adaptive enhancement method and system based on deep learning. Background Technology

[0002] In today's digital age, images, as an important information carrier, are widely used in various fields. However, during the acquisition, transmission, and storage of images, they are often affected by various factors, leading to a decline in image quality. Traditional image enhancement methods mainly include techniques such as histogram equalization and filtering to improve image contrast and sharpness, but they still have some limitations. For example, histogram equalization can lead to over-enhancement of the image and loss of image detail information, while filtering methods may blur image edges, thereby affecting image resolution.

[0003] The image adaptive enhancement algorithm for real-time image enhancement disclosed in CN112991240B includes: acquiring a grayscale image to be processed in the current frame; automatically calculating the effective grayscale level threshold of the grayscale histogram corresponding to the grayscale image based on an iterative algorithm; performing a grayscale equal-interval density equalization transformation on the grayscale histogram according to the effective grayscale level threshold; during the transformation process, when calculating the mapped grayscale value corresponding to the grayscale level in the enhanced image, a threshold limit is applied to the spacing after the grayscale level is rearranged at equal intervals, and grayscale bias correction is performed on the mapped grayscale value; after the grayscale equal-interval density equalization transformation is completed, the enhanced image corresponding to the grayscale image is obtained.

[0004] However, simply analyzing and processing the grayscale values ​​of an image to achieve image enhancement cannot personalize the enhancement according to the user's needs for the corresponding image, and the accuracy of the image enhancement process is insufficient. With the rapid development of deep learning technology, it has achieved great success in image recognition, image processing and other fields, which has also provided new ideas for adaptive image enhancement.

[0005] Therefore, how to adaptively process different parts of an image based on deep learning methods, avoid over-enhancement at different locations within the image, and improve the accuracy and personalization of the image enhancement process is a problem we need to solve. To this end, we now provide an image adaptive enhancement method and system based on deep learning. Summary of the Invention

[0006] To address the aforementioned technical problems, the present invention aims to provide an image adaptive enhancement method and system based on deep learning.

[0007] The objective of this invention can be achieved through the following technical solution: a deep learning-based image adaptive enhancement system, comprising an image processing platform, wherein the image processing platform is provided with an image acquisition module, an image processing module, an image analysis module, a model management module, and an image management module;

[0008] The image acquisition module is used to input image data and corresponding image information, and generate image data packets to send to the image processing module.

[0009] The image processing module is used to classify the corresponding image into regions based on image data and image information, perform image segmentation based on the region classification results, and obtain image quality data of the segmented image regions and corresponding pixels.

[0010] The image analysis module is used to perform classification processing based on the image quality data corresponding to the pixels in the segmented image region, and to obtain the comprehensive classification result of the pixels corresponding to the segmented image region.

[0011] The model management module is used to set up image processing sub-models according to different comprehensive classification results, perform comprehensive sorting of image processing sub-models, and construct image processing links and corresponding image processing link models.

[0012] The image management module is used to input the obtained segmented image regions into the image processing link respectively, obtain the corresponding region enhancement images of the segmented image regions according to the image processing link model, merge the region enhancement images, and obtain the adaptive enhancement image corresponding to the image data.

[0013] Furthermore, the process by which the image acquisition module records image data and corresponding image information includes:

[0014] Set up image input terminals and information input terminals;

[0015] The image input terminal is used to input image data and perform labeling processing on the image data; the information input terminal is used to input image information corresponding to the image data, and the image information includes image description information and image key information.

[0016] The obtained image data and image information are used to generate image data packets based on the labeling processing results.

[0017] Furthermore, the process by which the image processing module acquires image quality data of segmented image regions and corresponding pixels includes:

[0018] The pixel values ​​corresponding to the corresponding pixels are obtained from the image data. The grayscale image corresponding to the image data is obtained from the pixel values. Edge detection processing is performed on the grayscale image to determine the corresponding edge pixels. An edge image is generated based on the position marking of the edge pixels in the grayscale image.

[0019] A semantic analysis algorithm is set up to analyze and process image information, extract corresponding descriptive semantic information and feature semantic information, compare and analyze the descriptive semantic information and feature semantic information with the corresponding pixels in the image data, determine the position of the corresponding pixel, and mark the descriptive semantic information and feature semantic information according to the position of the pixel.

[0020] Image sub-regions are set according to the distribution of corresponding edge pixels in the edge image. Based on the distribution results of descriptive semantic information and feature semantic information in each image sub-region, it is determined whether to segment and extract the corresponding image sub-regions to obtain the segmented image regions after segmentation and extraction.

[0021] The segmented image region is marked as the original image. A mapping function is set, and the pixel values ​​of the corresponding pixels in the original image are scaled based on the mapping function to obtain the corresponding scaled image.

[0022] Image quality data is defined as the mean square error of pixel values ​​between the original image and the scaled image.

[0023] Furthermore, the process by which the image analysis module obtains the comprehensive classification result of the corresponding pixels in the segmented image region includes:

[0024] Obtain the distribution of pixels corresponding to the semantic information and feature semantic information within the segmented image region, and divide the corresponding pixels into key pixels, basic pixels and auxiliary pixels according to the labeling results of the corresponding pixels;

[0025] The image quality data and historical image data packets within the segmented image region are acquired. Based on the historical image data packets, pixel quality level ranges are preset, namely high quality pixel range, medium quality pixel range and low quality pixel range. The image quality data is compared and analyzed with the corresponding pixel quality level ranges. Based on the comparison and analysis results, the pixel quality level of the corresponding image quality data is determined.

[0026] Obtain the classification results corresponding to the two different classification methods mentioned above, and mark the obtained classification results as the comprehensive classification results of the corresponding pixels.

[0027] Furthermore, the process by which the model management module sets up image processing sub-models based on different comprehensive classification results includes:

[0028] The image processing procedure obtains the image data pixels corresponding to different comprehensive classification results and generates the image processing dataset corresponding to the corresponding comprehensive classification results;

[0029] Based on deep learning algorithms, the obtained image processing dataset is analyzed and trained to construct image processing sub-models corresponding to the comprehensive classification results of the corresponding types.

[0030] Furthermore, the process by which the model management module constructs the image processing chain and the corresponding image processing chain model includes:

[0031] Obtain the importance level of the pixels corresponding to each comprehensive classification result in the image processing process, sort the corresponding comprehensive classification results according to the importance level, set the corresponding image processing space according to the sorting result, and associate the image processing space with the corresponding image processing sub-model. Input the pixel region corresponding to the corresponding comprehensive classification result into the image processing space, and connect the various image processing spaces to generate an image processing link.

[0032] The image merging process of each image processing space within the image processing link is obtained, and a link dataset is generated. The constructed link dataset is analyzed and processed based on deep learning algorithms to build an image processing link model.

[0033] Furthermore, the process by which the image management module acquires the adaptively enhanced image corresponding to the image data includes:

[0034] The comprehensive classification result of the corresponding pixel region within the segmented image region is obtained and input into the idle image processing link. The corresponding image processing sub-models in each image processing space of the image processing link analyze and process the corresponding pixel region in turn until the image processing link completes the enhancement analysis of the corresponding segmented image region. The result is then input into the image processing link model for analysis and processing, resulting in the region enhancement image of the corresponding segmented image region.

[0035] Obtain the edge pixels corresponding to the region enhancement image, merge the region enhancement images according to the distribution of edge pixels of each region enhancement image, and obtain the adaptive enhancement image corresponding to the image data.

[0036] Compared with the prior art, the beneficial effects of the present invention are:

[0037] 1. By setting image information to mark key information in the image data as perceived by humans, the personalized intervention in the image processing process is improved. In addition, by obtaining the classification results of different pixels in the image through image information and image data, targeted enhancement processing is carried out based on the classification results, which improves the accuracy of the image data processing process to a certain extent.

[0038] 2. Based on the classification results of the image data, set the corresponding segmented image regions, distribute the image information in each segmented image region, and perform image enhancement processing on each segmented image region. This avoids, to a certain extent, the impact of image processing failures or errors on the entire image data.

[0039] 3. By setting up an image processing link, the pixels corresponding to different classification results in each segmented image region are classified and processed. This ensures that each pixel is enhanced, while also improving the accuracy of the image enhancement process to a certain extent. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of a deep learning-based image adaptive enhancement system according to an embodiment of this application.

[0041] Figure 2 This is a schematic diagram of the deep learning-based image adaptive enhancement method according to an embodiment of this application. Detailed Implementation

[0042] like Figure 1 As shown, the deep learning-based image adaptive enhancement system includes an image processing platform, which is equipped with an image acquisition module, an image processing module, an image analysis module, a model management module, and an image management module.

[0043] The image processing platform is used to acquire image data that needs to be analyzed and processed, and to analyze the acquired image data based on deep learning algorithms to achieve adaptive enhancement of the corresponding image data, thereby improving image quality. The platform also stores the processing process and updates the deep learning algorithm based on the processing process, thus better preparing for image data processing.

[0044] The image acquisition module is equipped with an image input terminal and an information input terminal;

[0045] The image input terminal is used to input image data, mark the obtained image data, and store the marked image data.

[0046] The information input terminal is used to input corresponding image information based on the marking results of the input image data. The image information includes image description information and image key information, wherein:

[0047] Image description information is the textual description information corresponding to the image data;

[0048] Key information in an image refers to the key feature information corresponding to the image data.

[0049] It should be further explained that, in the specific implementation process, the image acquisition module sets the corresponding image data packet to the marking result corresponding to the obtained image data, and associates the obtained other image information corresponding to the corresponding image data with data, stores the image data and image information in the corresponding image data packet, and processes and marks the obtained image data packet as three states: pending processing, processing, and processing completed, and stores them in the image processing platform.

[0050] The image processing module is used to classify the corresponding image regions based on image data and image information, perform image segmentation based on the region classification results, and obtain image quality data of the segmented image regions and corresponding pixels. The specific implementation process includes:

[0051] Set up an image recognition unit and an image preprocessing unit;

[0052] The image recognition unit acquires the image data and image information included in the image data packet marked as to be processed, and performs image recognition processing on the acquired image data and image information:

[0053] Obtain the pixel value data corresponding to the corresponding image data pixel, and calculate the weighted average of the obtained pixel value data based on the RGB three channels to obtain the gray value of the corresponding pixel.

[0054] The grayscale value of the corresponding pixel is labeled as gray, and the values ​​of each channel corresponding to each pixel are labeled as G, R, and B respectively, where:

[0055] gray=0.299×R+0.587×G+0.114×B;

[0056] The grayscale values ​​corresponding to each pixel in the image data are statistically processed to obtain the grayscale image corresponding to the image data.

[0057] Obtain image information from the image data packet, set the corresponding semantic analysis algorithm, perform semantic analysis on the corresponding image information based on the semantic analysis algorithm, and extract the corresponding descriptive semantic information and feature semantic information;

[0058] Based on the obtained descriptive semantic information and feature semantic information, a semantic duplication screening analysis is performed, and duplicate semantic information is marked as feature semantic information.

[0059] Image processing nodes are set according to the semantic information of the description, key image processing nodes are set according to the semantic information of the features, and the image processing nodes and key image processing nodes are marked at the corresponding pixel positions of the image data based on the deep learning algorithm.

[0060] The grayscale values ​​of each adjacent pixel in a grayscale image are obtained, and edge detection processing is performed on the grayscale values ​​in the grayscale image. The specific implementation process includes:

[0061] Label the corresponding pixel coordinates in the grayscale image as i and j, where i is the horizontal coordinate in the grayscale image and j is the vertical coordinate in the grayscale image, i = 0, 1, 2, ..., n-1 and n; j = 0, 1, 2, ..., m-1 and m; and denote X and Y as the horizontal and vertical directions of the grayscale image, respectively.

[0062] Set an edge detection gradient template, wherein the edge detection gradient template is a corresponding 2×2, where 2 is the detection range corresponding to the edge detection gradient template;

[0063] The process involves obtaining the corresponding grayscale values ​​within the edge detection template, performing finite difference calculations on the corresponding grayscale values ​​in the horizontal and vertical directions to obtain the gradient images in these two directions. The specific implementation process includes:

[0064] G x (i,j)=I(i,j)-I(i+1,j+1);

[0065] G y (i,j)=I(i+1,j)-I(i,j+1);

[0066] Obtain the gradient magnitude of the pixel within the corresponding edge detection gradient template in the grayscale image, and label the gradient magnitude as TF, where TF = |G| x |+|G y |;

[0067] A preset gradient threshold is set and labeled as TY. The obtained gradient magnitude is then compared and analyzed with the preset gradient threshold.

[0068] When TF > TY, then the pixel is an edge pixel;

[0069] When TF≤TY, the pixel is not an edge pixel;

[0070] Mark the edge points within the grayscale image, and generate an edge image based on the marking results;

[0071] The image processing nodes and key image processing nodes within the image data are mapped to the edge image, and image recognition and segmentation are performed based on the mapping results. The specific implementation process includes:

[0072] Obtain the corresponding edge pixels in the edge image, set the image sub-regions according to the distribution of the edge pixels, and mark the corresponding image sub-regions. Obtain the marking symbols of the image sub-regions, and mark the number of image processing nodes and key image processing nodes in the corresponding image sub-regions according to the marking symbols, and obtain the corresponding number marking results.

[0073] The system presets weight coefficients for image processing nodes and key image processing nodes, and analyzes and processes the corresponding image sub-regions based on the weight coefficients and quantity labeling results. The specific implementation process includes:

[0074] The number of image processing nodes and key image processing nodes are labeled TS and GS, respectively. The weight coefficients are labeled α and β, respectively. The comprehensive weight data of the corresponding image sub-region is labeled ZQ, where:

[0075] ZQ = TS × α + GS × β;

[0076] The average weight data and fluctuation coefficient within the preset edge image are used to obtain the corresponding weight standard interval based on the product of the average weight data and the fluctuation coefficient.

[0077] The comprehensive weight data and weight standard intervals corresponding to each image sub-region are compared and analyzed:

[0078] If the comprehensive weighted data belongs to the weight standard range, the corresponding image sub-regions will be segmented and extracted, and marked as segmented image regions.

[0079] If the comprehensive weight data does not belong to the weight standard interval, the corresponding image sub-region will not be segmented and extracted. Instead, the image sub-region will be divided or merged according to the relationship between the comprehensive weight data and the weight standard interval. The segmentation and extraction will be performed based on the division or merging results to obtain the corresponding segmented image region.

[0080] If the comprehensive weight data is less than the weight standard interval, then the adjacent image sub-regions are merged with the image sub-region, and the merged image sub-regions are segmented and extracted to obtain the corresponding segmented image regions.

[0081] If the total weight data is greater than the weight standard interval, then the image sub-region is divided, and the divided image sub-region is segmented and extracted to obtain the corresponding segmented image region.

[0082] The image preprocessing unit is used to obtain the segmented image regions corresponding to the corresponding image data, analyze and process the segmented image regions, and determine the image quality data of the corresponding segmented image regions. Its specific implementation process includes:

[0083] Obtain the pixel value of the corresponding pixel point within the corresponding segmented image region, and scale the corresponding segmented image region according to the pixel value of the pixel point, where:

[0084] Set a mapping function to mark the segmented image region as the original image, with the pixel coordinates corresponding to (x, y), and the scaled coordinates as (x, y). s y s Get the scaled coordinates (x) s y s The actual position coordinates (x, y) in the original image j y j );

[0085] Get the actual position coordinates (x) j y j The four pixels surrounding (x0, y0), (x0, y1), (x1, y0), and (x1, y1);

[0086] The values ​​of the scaled pixels are obtained based on the bilinear interpolation formula, and the four pixels are denoted as S. 11 S 12 S 21 and S 22 The pixel value of the interpolated pixel is labeled as P, where:

[0087] P = (x1 - x) s )(y1-y s )S 11 +(x s -x0)(y1-y s )S 21 +(x1-x s )(y2-

[0088] y s )S 12+ (x s -x1)(y2-y s )S 22 ;

[0089] Based on the pixel values ​​of the corresponding pixels in the scaled image segmentation region;

[0090] Obtain the number of rows and columns of pixels in the segmented image region, denoted as m and n respectively; label the pixel values ​​of the original image as I(i,j) and the pixel values ​​of the scaled segmented image region as S(i,j); obtain the mean square error (MSE) of pixel values ​​between the original and scaled images, where:

[0091]

[0092] Based on the obtained mean square error value, set the image quality data for the corresponding pixels of the segmented image region. The larger the image quality data, the worse the image quality.

[0093] It should be further explained that, in the specific implementation process, the image processing module stores the data information obtained in the image recognition unit and the image preprocessing unit into the image data packet corresponding to the image data, and sends the image data packet to the image analysis module.

[0094] The image analysis module is used to acquire image data packets sent by the image processing module, analyze and process the content of the image data packets, and classify the image data.

[0095] Set up type analysis unit and comprehensive analysis unit;

[0096] The type analysis unit is used to acquire image quality data and image description data within the image data packet, and to perform image analysis on the image data according to the data type.

[0097] Obtain the image processing nodes and key image processing nodes within the segmented image region corresponding to the image data, and mark the corresponding pixel positions within the image processing nodes and key image processing nodes:

[0098] When a pixel within a segmented image region corresponds to a key image processing node, it is marked as a key pixel.

[0099] When a corresponding pixel in a segmented image region corresponds to an image processing node, it is marked as a basic pixel.

[0100] When a pixel within a segmented image region does not correspond to a key image processing node or an image processing node, it is marked as an auxiliary pixel.

[0101] Obtain the image quality data of the corresponding pixels within the segmented image region corresponding to the image data, analyze and process the image quality data, and classify the corresponding pixels;

[0102] Obtain historical image data packets marked as processed within the image processing platform, and set pixel quality level ranges based on the image quality data within the historical image data packets, namely, high quality pixel range, medium quality pixel range, and low quality pixel range;

[0103] Among them, the image quality data corresponding to the high-quality pixel range is smaller than the image quality data corresponding to the medium-quality pixel range, and the image quality data corresponding to the medium-quality pixel range is smaller than the image quality data corresponding to the low-quality pixel range.

[0104] The image quality data corresponding to the segmented image region is compared and analyzed with the corresponding pixel quality level range. Based on the comparison results, the pixel quality level to which the corresponding image quality data belongs is determined.

[0105] The comprehensive analysis unit obtains the pixel type to which the corresponding pixel points belong, as obtained by the type analysis unit.

[0106] Obtain the classification type corresponding to each pixel, and record key pixels, basic pixels, and basic pixels as one set of classification types, and record high quality pixel level, medium quality pixel level, and low quality pixel level as another set of classification types.

[0107] Perform cross-combinations within the two classification types, and set corresponding comprehensive classification results based on the cross-combination classification results;

[0108] The classification results of the corresponding pixels are matched with the comprehensive classification results. The classification type of the corresponding pixels is analyzed based on the matching results and uploaded into the image data packet. The obtained image data packet is then sent to the model management module.

[0109] The model management module is used to acquire the corresponding image data packets and set the image processing model according to the corresponding image data packets. Its specific implementation process includes:

[0110] The model management module includes a model training unit and a model management unit.

[0111] The model training unit is used to acquire image data packets marked as processed, classify the processed image data packets, extract the corresponding type of pixel enhancement process, and set the image enhancement process of the same type of pixel as the image processing dataset. The image processing datasets are respectively the corresponding training set and validation set.

[0112] The training set obtained is analyzed and trained based on deep learning algorithms to obtain the image processing sub-model corresponding to the corresponding classification type until the loss function in the image processing sub-model tends to be stable. The corresponding image processing sub-model is then output. The obtained validation set is input into the corresponding image processing sub-model for validation analysis. Based on the validation analysis results, it is determined whether the corresponding model needs to be retrained. If it meets the validation criteria, the corresponding image processing sub-model is output.

[0113] The model management unit obtains the image processing sub-models corresponding to different types of comprehensive classification results, and sets up the image processing link model based on the obtained image processing sub-models. The specific implementation process includes:

[0114] The link dataset is analyzed and processed, and importance levels are set for the corresponding comprehensive classification results. The comprehensive classification results are then sorted according to their importance levels.

[0115] Among them, importance weight factors are set for key pixels, basic pixels, basic pixels, high-quality pixel level, medium-quality pixel level and low-quality pixel level respectively;

[0116] Based on the cross-combination results, obtain the importance weight factor of the corresponding classification result within the comprehensive classification result, and obtain the importance level data of the corresponding comprehensive classification result based on the obtained importance weight factor;

[0117] Sort the data by importance level from lowest to highest size;

[0118] Based on the sorting results, the corresponding image processing spaces are set up sequentially. The image processing sub-model corresponding to the comprehensive classification result is input into the corresponding image processing space, and the pixel region corresponding to the comprehensive classification result is input into the image processing space. The image processing sub-model analyzes and processes the corresponding pixel region in the corresponding image processing space. After the processing is completed, it is input into the next image processing space according to the sorting results.

[0119] The image merging process of each image processing space within the image processing link is obtained, and a link dataset is generated. The constructed link dataset is analyzed and processed based on deep learning algorithms to build an image processing link model.

[0120] The image management module is used to acquire the corresponding image processing link model, analyze and process the image data packets to be processed, and complete the adaptive image enhancement process. Its specific implementation process includes:

[0121] The image management module replicates the image processing links set by the model management module, sets up multiple image processing links with the same function, which facilitates the image processing platform to process multiple image data at the same time and detects the working status of the corresponding image processing links.

[0122] The comprehensive classification result of the corresponding pixel region within the segmented image region is obtained and input into the idle image processing link. The corresponding image processing sub-models in each image processing space of the image processing link analyze and process the corresponding pixel region in turn until the image processing link completes the enhancement analysis of the corresponding segmented image region. The result is then input into the image processing link model for analysis and processing, resulting in the region enhancement image of the corresponding segmented image region.

[0123] The process involves obtaining edge pixels corresponding to the region enhancement images, merging the region enhancement images based on their distribution, and obtaining the corresponding adaptive enhancement images. The specific implementation process includes:

[0124] Obtain the edge pixels corresponding to the region enhancement image, perform weighted processing on the obtained pixels to obtain the weight data of the corresponding edge detection results, and label it as w. n (x), label the pixel value of the nth input image as v. n (x), the merged pixel value is marked as z fused (x), where:

[0125]

[0126] During the fusion process, pixels are weighted according to the edge detection results to ensure that the pixel values ​​of edge regions within the region-enhanced image remain unchanged or transition smoothly, thereby realizing the merging process of the region-enhanced images.

[0127] like Figure 2 As shown, the deep learning-based image adaptive enhancement method includes the following steps:

[0128] Step 1: Set up the image processing platform, input image data and corresponding image information, and generate an image data package based on the input image data and image information;

[0129] Step 2: Classify the corresponding image regions based on the image data and image information respectively, perform image segmentation based on the region classification results, and obtain the image quality data of the segmented image regions and corresponding pixels;

[0130] Step 3: Perform classification processing based on the image quality data of corresponding pixels within the segmented image region to obtain the comprehensive classification result of corresponding pixels within the segmented image region;

[0131] Step 4: Set up image processing sub-models based on different comprehensive classification results, sort the image processing sub-models comprehensively, and construct the image processing link and the corresponding image processing link model;

[0132] Step 5: Input the obtained segmented image regions into the image processing link respectively, obtain the corresponding region enhancement images of the segmented image regions according to the image processing link model, merge the region enhancement images, and obtain the adaptive enhancement image corresponding to the image data.

[0133] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A deep learning-based image adaptive enhancement system, comprising an image processing platform, characterized in that, The image processing platform includes an image acquisition module, an image processing module, an image analysis module, a model management module, and an image management module. The image acquisition module is used to input image data and corresponding image information, and generate image data packets to send to the image processing module. The image processing module is used to classify the corresponding image into regions based on image data and image information, perform image segmentation based on the region classification results, and obtain image quality data of the segmented image regions and corresponding pixels. The image analysis module is used to perform classification processing based on the image quality data corresponding to the pixels in the segmented image region, and to obtain the comprehensive classification result of the pixels corresponding to the segmented image region. The model management module is used to set up image processing sub-models according to different comprehensive classification results, perform comprehensive sorting of image processing sub-models, and construct image processing links and corresponding image processing link models. The image management module is used to input the obtained segmented image regions into the image processing link respectively, obtain the corresponding region enhancement images of the segmented image regions according to the image processing link model, merge the region enhancement images, and obtain the adaptive enhancement image corresponding to the image data. The process by which the model management module constructs the image processing chain and the corresponding image processing chain model includes: Obtain the importance level of the pixels corresponding to each comprehensive classification result in the image processing process, sort the corresponding comprehensive classification results according to the importance level, set the corresponding image processing space according to the sorting result, and associate the image processing space with the corresponding image processing sub-model. Input the pixel region corresponding to the corresponding comprehensive classification result into the image processing space, and connect the various image processing spaces to generate an image processing link. The image merging process of each image processing space within the image processing chain is obtained, a chain dataset is generated, and the constructed chain dataset is analyzed and processed based on deep learning algorithms to build an image processing chain model. The process by which the image management module acquires the adaptively enhanced image corresponding to the image data includes: The comprehensive classification result of the corresponding pixel region within the segmented image region is obtained and input into the image processing link. The corresponding image processing sub-models in each image processing space of the image processing link analyze and process the corresponding pixel region in turn until the image processing link completes the enhancement analysis of the corresponding segmented image region. The result is then input into the image processing link model for analysis and processing, resulting in the region enhancement image of the corresponding segmented image region. Obtain the edge pixels corresponding to the region enhancement image, merge the region enhancement images according to the distribution of edge pixels of each region enhancement image, and obtain the adaptive enhancement image corresponding to the image data.

2. The deep learning-based image adaptive enhancement system according to claim 1, characterized in that, The process of the image acquisition module recording image data and corresponding image information includes: Set up image input terminals and information input terminals; The image input terminal is used to input image data and perform labeling processing on the image data; the information input terminal is used to input image information corresponding to the image data, the image information including image description information and image key information; The obtained image data and image information are used to generate image data packets based on the labeling processing results.

3. The deep learning-based image adaptive enhancement system according to claim 2, characterized in that, The process by which the image processing module acquires image quality data of segmented image regions and corresponding pixels includes: The pixel values ​​corresponding to the corresponding pixels are obtained from the image data. The grayscale image corresponding to the image data is obtained from the pixel values. Edge detection processing is performed on the grayscale image to determine the corresponding edge pixels. An edge image is generated based on the position marking of the edge pixels in the grayscale image. A semantic analysis algorithm is set up to analyze and process image information, extract corresponding descriptive semantic information and feature semantic information, compare and analyze the descriptive semantic information and feature semantic information with the corresponding pixels in the image data, determine the position of the corresponding pixel, and mark the descriptive semantic information and feature semantic information according to the position of the pixel. Image sub-regions are set according to the distribution of corresponding edge pixels in the edge image. Based on the distribution results of descriptive semantic information and feature semantic information in each image sub-region, it is determined whether to segment and extract the corresponding image sub-regions to obtain the segmented image regions after segmentation and extraction. The segmented image region is marked as the original image. A mapping function is set, and the pixel values ​​of the corresponding pixels in the original image are scaled based on the mapping function to obtain the corresponding scaled image. Image quality data is defined as the mean square error of pixel values ​​between the original image and the scaled image.

4. The deep learning-based image adaptive enhancement system according to claim 3, characterized in that, The process by which the image analysis module obtains the comprehensive classification result of the corresponding pixels in the segmented image region includes: Obtain the distribution of pixels corresponding to the semantic information and feature semantic information within the segmented image region, and divide the corresponding pixels into key pixels, basic pixels and auxiliary pixels according to the labeling results of the corresponding pixels; The image quality data and historical image data packets within the segmented image region are acquired. Based on the historical image data packets, pixel quality level ranges are preset, namely high quality pixel range, medium quality pixel range and low quality pixel range. The image quality data is compared and analyzed with the corresponding pixel quality level ranges. Based on the comparison and analysis results, the pixel quality level of the corresponding image quality data is determined. Obtain the classification results corresponding to the two different classification methods mentioned above, and mark the obtained classification results as the comprehensive classification results of the corresponding pixels.

5. The deep learning-based image adaptive enhancement system according to claim 4, characterized in that, The process by which the model management module sets up image processing sub-models based on different comprehensive classification results includes: The image processing procedure obtains the image data pixels corresponding to different comprehensive classification results and generates the image processing dataset corresponding to the corresponding comprehensive classification results; Based on deep learning algorithms, the obtained image processing dataset is analyzed and trained to construct image processing sub-models corresponding to the comprehensive classification results of the corresponding types.

6. The deep learning-based image adaptive enhancement method corresponding to the deep learning-based image adaptive enhancement system according to any one of claims 1 to 5, characterized in that, The following steps are involved: Step 1: Set up the image processing platform, input image data and corresponding image information, and generate an image data package based on the input image data and image information; Step 2: Classify the corresponding image regions based on the image data and image information respectively, perform image segmentation based on the region classification results, and obtain the image quality data of the segmented image regions and corresponding pixels; Step 3: Perform classification processing based on the image quality data of corresponding pixels within the segmented image region to obtain the comprehensive classification result of corresponding pixels within the segmented image region; Step 4: Set up image processing sub-models based on different comprehensive classification results, sort the image processing sub-models comprehensively, and construct the image processing link and the corresponding image processing link model; Step 5: Input the obtained segmented image regions into the image processing link respectively, obtain the corresponding region enhancement images of the segmented image regions according to the image processing link model, merge the region enhancement images, and obtain the adaptive enhancement image corresponding to the image data.

Images