Image enhancement method

Abstract

The invention discloses an image enhancement method, which comprises the following steps of: inputting an image, identifying a pixel bad block at a specified position in the image, and recording a row and column index of a pixel corresponding to the bad block; bad block identification and restoration technology upgrading and multi-scale bad block detection and layered restoration are carried out; on the premise of not crossing boundaries, defining a 5 * 5 neighborhood by taking the row and column index as a center, and finding non-bad block pixels in the neighborhood; self-adaptive texture synthesis and repair are carried out; performing detail enhancement technology processing; carrying out interpolation processing on the non-bad block pixels, and calculating by using an average value or other interpolation algorithms to obtain a new pixel value; replacing the pixel value of the bad block with a value obtained by interpolation processing to complete bad block restoration; and calculating the mean value of each channel of the RGB of the image, and converting the image into a grey-scale map by using the mean value, thereby obtaining the basic brightness level of the image. According to the method, the bad blocks of the image are accurately repaired, the brightness is optimized, denoising and color correction are performed, and the image quality is improved.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and in particular to an image enhancement method. Background Art

[0002] In the field of image processing technology, especially in improving the quality of images shot under low-light conditions, although existing technologies and methods can achieve a certain degree of image enhancement, there are still great challenges in reducing noise and eliminating bad pixel blocks while maintaining picture details.

[0003] Traditional image enhancement methods usually rely on techniques such as global illumination compensation and histogram equalization, but they are not effective in processing pixel-level details, especially in repairing pixel patches.

[0004] Current image enhancement technology often cannot effectively avoid overexposure while enhancing brightness, and the repair of bad blocks often leads to obvious artifacts in the surrounding areas, affecting the quality of the final image. In order to solve the problem of insufficient brightness, missing details and overall image quality degradation caused by bad pixel blocks when capturing images with a micro pinhole camera in a dark environment, an image enhancement method is proposed. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides an image enhancement method to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: an image enhancement method, the method comprising: Input an image, identify bad pixel blocks at specified positions in the image, and record the row and column indexes of the pixels corresponding to the bad blocks; Upgrade bad block identification and repair technology and multi-scale bad block detection and layered repair; Under the premise of not crossing the boundary, define a 5x5 neighborhood with the row and column index as the center, and find the non-bad block pixels in the neighborhood; Perform adaptive texture synthesis and restoration; Perform detail enhancement technology processing; Interpolate these non-bad block pixels and use an average value or other interpolation algorithm to calculate new pixel values; Replace the pixel value of the bad block with the value obtained by interpolation processing to complete the bad block repair; Calculate the mean of each RGB channel of the image, and use this mean to convert the image into a grayscale image to obtain the basic brightness level of the image; Set the brightness adjustment coefficient, adaptively enhance the brightness, and perform targeted enhancement processing according to different brightness levels. Specifically, there are three brightness enhancement strategies: low, medium, and high. By comparing the effects of different enhancement levels, comprehensively consider the balance between brightness improvement and image detail retention, select the best enhancement solution, and perform color correction and enhancement. In addition, it is necessary to detect the noise after image enhancement, use advanced filtering technology combined with color correction and enhancement and image clarity technology, and remove noise through filtering or threshold segmentation methods to prevent noise from interfering with subsequent color adjustments; Finally, the pixel color deviation is carefully corrected, the pixel RGB offset value is adjusted, and a comprehensive optimization strategy is implemented to make the color more natural and realistic; Output image; There are multiple miniature pinhole cameras installed, each equipped with a set of image enhancement methods. When it is nighttime or in dim light, the brightness of the images captured by the camera is obviously low. At this time, the image enhancement method starts to work. First, the bad block recognition module quickly locates several bad pixel blocks and records their exact position coordinates in the image. Next, the bad block repair module constructs a 5x5 neighborhood matrix based on the coordinate information without affecting the boundary conditions, and searches for valid pixels in it. Once the valid pixel set is determined, a more reasonable intermediate pixel value is calculated using the local weighted average method to replace the original bad block pixels. Then, the brightness enhancement module is started, and the average intensity of each color channel is obtained through statistical analysis of the RGB components of the entire image. On this basis, the overall brightness of the image is gradually improved according to the pre-set brightness adjustment curve. Taking into account the uneven brightness distribution in different areas, this module also has adaptive adjustment. The whole function can dynamically adjust the treatment strategy according to the local illumination changes monitored in real time. At the same time, the noise detection and removal modules run synchronously, and the enhanced image is denoised with the help of an edge-preserving filter, aiming to remove random noise points caused by excessive gain and ensure that the clarity of the picture is not damaged. Finally, the color adjustment module comes on stage. It finely adjusts the ratio of the three primary colors of red, green and blue for each pixel to correct the possible color deviation phenomenon, so that the final output image color is richer and fuller, close to the real effect of the human eye's intuitive perception. Furthermore, in order to solve the problem that traditional image enhancement methods are difficult to achieve the ideal brightness enhancement and detail retention effects in a dark environment, a multi-level image fusion strategy is adopted. By dividing the original image into different frequency bands, the high-frequency part is enhanced in detail, while the low-frequency part focuses on the overall outline. In this way, not only the brightness level is improved, but also the texture characteristics of the scene are well preserved, avoiding the distortion caused by over-sharpening.

[0007] Preferably, the bad block identification and repair technology upgrade specifically includes: Introducing machine learning algorithms and using deep learning models to automatically identify pixel modules in images to improve recognition accuracy and efficiency; Adaptive neighborhood selection: dynamically adjust the neighborhood size according to the distribution of pixels around the module to include more effective information for interpolation processing; Advanced interpolation algorithms use bilinear interpolation and bicubic interpolation or more advanced interpolation algorithms to improve the accuracy and naturalness of restored pixels.

[0008] Preferably, the adaptive texture synthesis and restoration specifically includes: Combining the local and global texture features of the image, an adaptive texture synthesis algorithm is used to generate a repair result that is highly consistent with the surrounding area based on the texture and brightness information of the pixels around the bad block, achieving a more natural repair effect.

[0009] Preferably, the multi-scale bad block detection and hierarchical repair specifically include: The bad blocks of the image are detected at different scales. According to the size and distribution characteristics of the bad blocks, a layered repair strategy is adopted to repair large-scale bad blocks first and then gradually refine them to small-scale bad blocks to achieve a more refined repair effect.

[0010] Preferably, the detail enhancement technology specifically includes: Edge enhancement, using the Sobel operator and Canny edge detection algorithm to enhance the edge details in the image and improve the clarity of the image; Frequency domain enhancement converts the image to the frequency domain through Fourier transform, enhances the high-frequency components, and then inversely transforms back to the spatial domain to achieve detail enhancement.

[0011] Preferably, the brightness adaptive enhancement specifically includes: Based on histogram equalization, according to the histogram distribution of the image, the brightness distribution is dynamically adjusted to enhance the contrast of the image; Application of Retinex theory, using Retinex theory to simulate the human visual system's perception of brightness and color to achieve more natural brightness enhancement.

[0012] Preferably, the color correction and enhancement specifically includes: Color space conversion: convert the image from RGB color space to color space such as HSV or Lab, and adjust the brightness, saturation and hue respectively to achieve more refined color correction; Color balance adjustment automatically adjusts the proportion of the three primary colors of red, green and blue according to the overall color distribution of the image to make the colors more balanced and natural.

[0013] Preferably, the advanced filtering technology specifically includes: Non-local mean filtering uses the information of similar pixel blocks in the image to filter, effectively removing noise while retaining image details; Guided filtering, combining edge-preserving filtering and guided images, achieves smooth denoising while preserving edge details.

[0014] Preferably, the image clearing technology specifically includes: Deblurring algorithm: for image blur caused by camera shake or motion blur, a blind deblurring algorithm or a point spread function-based deblurring algorithm is used to perform clear processing; Super-resolution reconstruction uses deep learning models to perform super-resolution reconstruction on images to improve the resolution and detail of the images.

[0015] Preferably, the implementation of the comprehensive optimization strategy specifically includes: Modular design modularizes the above optimization technologies and allows them to be flexibly combined according to actual needs to achieve comprehensive optimization effects; Automation and intelligence: using machine learning algorithms to automatically adjust various optimization parameters, realize intelligent processing, and improve processing efficiency and effects; Real-time and resource optimization: While ensuring the optimization effect, the real-time performance and resource usage of the algorithm are considered to optimize the algorithm implementation and reduce the computational complexity.

[0016] In summary, compared with the prior art, the present invention provides an image enhancement method, which has the following beneficial effects: The present invention uses the image enhancement method to accurately identify and repair bad pixel blocks at specified positions, thereby significantly improving the integrity and visual quality of the image. The 5x5 neighborhood interpolation process is used to effectively restore the pixel values ​​of the bad block area. At the same time, by calculating the RGB channel mean conversion grayscale image, an accurate assessment of the image brightness level is achieved, and an optimized enhancement strategy is adopted for different brightness levels to balance brightness enhancement and detail retention. In addition, the method can also effectively detect and remove noise after enhancement to prevent interference with subsequent color adjustment. Finally, the pixel color deviation is carefully corrected to make the image color more natural and realistic, thereby improving the overall viewing and practicality of the image, thereby solving the problems raised in the background technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is the overall flow chart of the image enhancement of the invention. DETAILED DESCRIPTION

[0018] The present invention provides a technical solution, an image enhancement method, please refer to Figure 1 , the method comprising: Input an image, identify bad pixel blocks at specified positions in the image, and record the row and column indexes of the pixels corresponding to the bad blocks; Upgrade bad block identification and repair technology and multi-scale bad block detection and layered repair; Under the premise of not crossing the boundary, define a 5x5 neighborhood with the row and column index as the center, and find the non-bad block pixels in the neighborhood; Perform adaptive texture synthesis and restoration; Perform detail enhancement technology processing; Interpolate these non-bad block pixels and use an average value or other interpolation algorithm to calculate new pixel values; Replace the pixel value of the bad block with the value obtained by interpolation processing to complete the bad block repair; Calculate the mean of each RGB channel of the image, and use this mean to convert the image into a grayscale image to obtain the basic brightness level of the image; Set the brightness adjustment coefficient, adaptively enhance the brightness, and perform targeted enhancement processing according to different brightness levels. Specifically, there are three brightness enhancement strategies: low, medium, and high. By comparing the effects of different enhancement levels, comprehensively consider the balance between brightness improvement and image detail retention, select the best enhancement solution, and perform color correction and enhancement. In addition, it is necessary to detect the noise after image enhancement, use advanced filtering technology combined with color correction and enhancement and image clarity technology, and remove noise through filtering or threshold segmentation methods to prevent noise from interfering with subsequent color adjustments; Finally, the pixel color deviation is carefully corrected, the pixel RGB offset value is adjusted, and a comprehensive optimization strategy is implemented to make the color more natural and realistic; Output image; In the above complex image enhancement method, each step is carefully designed to ensure that bright and clear images can be captured under extreme lighting conditions, such as at night or in low-light environments. This system not only relies on traditional image processing technology, but also deeply integrates modern artificial intelligence, especially the latest achievements in the field of deep learning, thereby achieving unprecedented image quality improvement and deep optimization of bad block identification and repair. The bad block identification module is the first step in image preprocessing, and its efficiency and accuracy are directly related to the effects of subsequent steps. Traditional bad block identification often relies on simple threshold judgment or statistical methods, but in complex and changeable image scenes, these methods are prone to misjudgment, resulting in loss of image information. Or error repair, for this reason, the system introduced a bad block detection algorithm based on deep learning, which automatically identifies abnormal pixel blocks in the image by training a convolutional neural network (CNN). This algorithm can learn and understand advanced features such as image texture and color distribution, so as to more accurately locate bad blocks and reduce false positives and false negatives. In the bad block repair stage, in addition to the traditional local weighted average method, the system also integrates pixel prediction technology based on deep learning. This technology uses a trained neural network to predict and generate the most likely pixel value to replace the bad block based on the valid pixel information around the bad block. Compared with the traditional interpolation algorithm, this method can better maintain the local structure and texture characteristics of the image and reduce the repair time. In addition, the system also uses conditional generative adversarial networks (GANs) for image restoration. Through adversarial training between the generator and the discriminator, GAN can generate more realistic and natural image patches, further improving the restoration quality. Brightness enhancement and adaptive adjustment. The brightness enhancement module is one of the cores of the system. It analyzes the RGB components of the image and combines the preset brightness adjustment curve to intelligently enhance the overall brightness of the image. In this process, the system not only considers the uniform improvement of global brightness, but also adds a local adaptive adjustment mechanism. It can dynamically adjust the enhancement strategy according to the brightness distribution of different areas of the image to avoid overexposure or underexposure. In order to achieve this The system adopts a regional brightness prediction model based on deep learning, which can accurately evaluate the brightness requirements of each small area and make fine adjustments accordingly. The system uses intelligent noise detection and removal. In the process of image enhancement, the increase in brightness is often accompanied by an increase in noise, especially when the gain is high. Therefore, the noise detection and removal module is particularly important. The system uses advanced edge-preserving filters, such as bilateral filters or guided filters. These filters can effectively retain the edge details of the image while removing noise and avoid image blur. In addition, the system also combines denoising algorithms based on deep learning, such as Denoising Convolutional Neural Networks (DnCNN), which learns the mapping relationship between noise and clean images through training the network to achieve more efficient and accurate denoising effects, color adjustment and detail retention.The color adjustment module is dedicated to correcting the color cast in the image to make the image color more realistic and natural. This module analyzes the ratio of the three primary colors of red, green and blue for each pixel in the image, combined with color space conversion technology (such as conversion from RGB to HSV or Lab space), to make fine adjustments to the color. At the same time, in order to retain the detailed information in the image, the system adopts a multi-level image fusion strategy. This strategy uses wavelet transform or fast Fourier transform to decompose the image into different frequency bands, and enhances the details of high-frequency components, such as edge sharpening and texture enhancement; while the low-frequency components are smoothed to enhance the overall structure and contour. This processing method not only enhances the brightness level of the image, but also effectively retains the detailed features of the scene, avoiding distortion caused by excessive processing. The introduction of deep learning and iterative optimization has brought revolutionary changes to the system in the entire image processing process, whether it is bad block repair, brightness adjustment, Whether it is noise removal or color adjustment, the deep learning model has demonstrated strong learning ability and adaptability. In order to further improve the performance and accuracy of the system, the system adopts a two-stage deep learning training strategy: offline pre-training and online reasoning. In the offline pre-training stage, the model is trained with a large amount of labeled data to enable the model to have preliminary recognition and prediction capabilities; in the online reasoning stage, the model parameters are continuously optimized through real-time data feedback to achieve continuous learning and self-improvement of the model. This mechanism ensures that the system can always maintain a high level of processing power and accuracy when facing complex and changing image scenes. In summary, this image enhancement and processing system achieves high-quality image preview under extreme lighting conditions by integrating traditional image processing technology with deep learning algorithms. It not only improves the brightness and clarity of the image, but also retains rich detail information and true color expression, providing users with a more intuitive and realistic visual experience. ,

[0019] See also Figure 1 , the bad block identification and repair technology upgrade specifically includes: Introducing machine learning algorithms and using deep learning models (such as convolutional neural networks, Transformer, or generative adversarial networks (GANs)) to automatically identify bad pixel blocks in images, improving recognition accuracy and efficiency; Improves the accuracy and efficiency of bad block identification, reduces false positives and false negatives, and provides a more reliable foundation for subsequent repair steps; Adaptive neighborhood selection: dynamically adjust the neighborhood size according to the distribution of pixels around the module to include more effective information for interpolation processing; Through adaptive neighborhood selection, the interpolation process is optimized, the accuracy and naturalness of the restored image are improved, and the restoration effect is more in line with the user's visual expectations; Advanced interpolation algorithm, using bilinear interpolation and bicubic interpolation or more advanced interpolation algorithms to improve the accuracy and naturalness of the restored pixels, so as to obtain a more natural restoration effect; The use of advanced interpolation algorithms significantly improves the clarity and detail expression of the restored image, making the restoration effect more natural and delicate, meeting users' needs for high-quality images.

[0020] See also Figure 1 , Adaptive texture synthesis and restoration specifically include: Combining local and global texture features of the image, an adaptive texture synthesis algorithm is used to generate a repair result that is highly consistent with the surrounding area based on the texture and brightness information of the pixels around the bad block, achieving a more natural repair effect. By combining the local and global texture features of the image, the generated restoration result is highly consistent with the surrounding area, significantly improving the naturalness and visual effect of the restoration.

[0021] See also Figure 1 ,Multi-scale bad block detection and hierarchical repair specifically include: The bad blocks of the image are detected at different scales. According to the size and distribution characteristics of the bad blocks, a layered repair strategy is adopted to repair large-scale bad blocks first, and then gradually refine them to small-scale bad blocks to achieve a more refined repair effect. Bad block detection at different scales and layered repair ensures that the repair process is gradually refined from large to small, achieving a more refined repair effect and reducing discontinuities and artifacts after repair.

[0022] See also Figure 1 , detail enhancement technology specifically includes: Edge enhancement, using the Sobel operator and Canny edge detection algorithm to enhance the edge details in the image and improve the clarity of the image; Effectively enhances the edge details in the image, improving the clarity and recognition of the image; Frequency domain enhancement: convert the image to the frequency domain through Fourier transform, enhance the high-frequency components, and then inversely transform it back to the spatial domain to achieve detail enhancement; The texture and detail information of the image are enhanced, making the image more delicate and rich.

[0023] See also Figure 1 , brightness adaptive enhancement specifically includes: Based on histogram equalization, according to the histogram distribution of the image, the brightness distribution is dynamically adjusted to enhance the contrast of the image; Improved the brightness distribution of the image, making the image brighter and with moderate contrast, thus improving the visual effect of the image; Application of Retinex theory: using Retinex theory to simulate the human visual system's perception of brightness and color to achieve more natural brightness enhancement; It enhances the brightness of the image while maintaining the naturalness and authenticity of the color, improving the overall beauty of the image.

[0024] See also Figure 1 , color correction and enhancement specifically include: Color space conversion: convert the image from RGB color space to color space such as HSV or Lab, and adjust the brightness, saturation and hue respectively to achieve more refined color correction; Adjust brightness, saturation, and hue individually for more refined color correction; Color balance adjustment: automatically adjusts the proportion of the three primary colors of red, green and blue according to the overall color distribution of the image to make the colors more balanced and natural; Improves the color performance of images, making them more balanced, natural and vivid, enhancing the visual appeal of images.

[0025] See also Figure 1 , advanced filtering techniques include: Non-local mean filtering uses the information of similar pixel blocks in the image to filter, effectively removing noise while retaining image details; Improves image clarity and smoothness, reduces the impact of noise on image quality; Guided filtering, combining edge-preserving filtering and guided imaging, achieves smooth denoising while preserving edge details; While removing noise, the edge information of the image is maintained, making the image clearer and more delicate.

[0026] See also Figure 1 , image sharpening technology specifically includes: Deblurring algorithm: for image blur caused by camera shake or motion blur, a blind deblurring algorithm or a point spread function-based deblurring algorithm is used to perform clear processing; It restores image blur caused by camera shaking or motion blur, improving image clarity and recognition; Super-resolution reconstruction: Use deep learning models to perform super-resolution reconstruction on images to improve image resolution and detail. The resolution and detail richness of the image are significantly improved, making the image clearer and more delicate.

[0027] See also Figure 1 , the implementation of comprehensive optimization strategy includes: Modular design modularizes the above optimization technologies and allows them to be flexibly combined according to actual needs to achieve comprehensive optimization effects; Improves the flexibility and scalability of the system, making it easier to optimize and adjust according to different application scenarios. The modular design allows the system to flexibly select and combine optimization technologies according to different application scenarios and requirements. The system is no longer a fixed whole, but can be dynamically adjusted and optimized according to specific tasks. The modular design makes the optimization and adjustment of the system simpler and more intuitive. By adjusting the combination and parameter settings between different modules, the overall performance of the system can be easily improved. Automation and intelligence: using machine learning algorithms to automatically adjust various optimization parameters, realize intelligent processing, and improve processing efficiency and effects; Improves processing efficiency and effect, reduces the workload of manual intervention and parameter adjustment, and the application of automation and intelligent technology enables the system to automatically adjust various optimization parameters to achieve the most optimized processing effect, allowing users to focus more on image data analysis and application without spending a lot of time and energy on system adjustment and optimization. Intelligent processing also improves system reliability and stability because the system can automatically adapt to different image data and processing requirements; Real-time and resource optimization: While ensuring the optimization effect, the real-time performance and resource usage of the algorithm are considered to optimize the algorithm implementation and reduce the computational complexity; It ensures the real-time performance and resource utilization of the system when processing large-scale image data, and improves the overall performance and stability of the system. Real-time performance is one of the important indicators of modern image processing systems. By optimizing algorithm implementation and considering resource occupancy, the system can achieve the ability to quickly process large-scale image data while ensuring optimization effects. This is crucial for application scenarios that require real-time processing of image data (such as video surveillance). The system can process and analyze image data in real time to provide users with timely and accurate information. Resource optimization enables the system to utilize computing resources and storage resources more efficiently. By reducing computational complexity and optimizing algorithm implementation, the system can reduce resource occupancy and consumption while ensuring processing effects, thereby improving the overall performance of the system and reducing the system's operating and maintenance costs. This resource optimization effect is particularly significant for image enhancement and optimization systems that require long-term operation and large-scale data processing.

[0028] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0029] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An image enhancement method, characterized in that: The method comprises: Input an image, identify bad pixel blocks at specified positions in the image, and record the row and column indexes of the pixels corresponding to the bad blocks; Upgrade bad block identification and repair technology and multi-scale bad block detection and layered repair; Under the premise of not crossing the boundary, define a 5x5 neighborhood with the row and column index as the center, and find the non-bad block pixels in the neighborhood; Perform adaptive texture synthesis and restoration; Perform detail enhancement technology processing; Interpolate these non-bad block pixels and use an average value or other interpolation algorithm to calculate new pixel values; Replace the pixel value of the bad block with the value obtained by interpolation processing to complete the bad block repair; Calculate the mean of each RGB channel of the image, and use this mean to convert the image into a grayscale image to obtain the basic brightness level of the image; Set the brightness adjustment coefficient, adaptively enhance the brightness, and perform targeted enhancement processing according to different brightness levels. Specifically, there are three brightness enhancement strategies: low, medium, and high. By comparing the effects of different enhancement levels, comprehensively consider the balance between brightness improvement and image detail retention, select the best enhancement solution, and perform color correction and enhancement. In addition, it is necessary to detect the noise after image enhancement, use advanced filtering technology combined with color correction and enhancement and image clarity technology, and remove noise through filtering or threshold segmentation methods to prevent noise from interfering with subsequent color adjustments; Finally, the pixel color deviation is carefully corrected, the pixel RGB offset value is adjusted, and a comprehensive optimization strategy is implemented to make the color more natural and realistic; Output image.

2. The image enhancement method according to claim 1, characterized in that: The bad block identification and repair technology upgrade specifically includes: Introducing machine learning algorithms and using deep learning models to automatically identify pixel modules in images to improve recognition accuracy and efficiency; Adaptive neighborhood selection: dynamically adjust the neighborhood size according to the distribution of pixels around the module to include more effective information for interpolation processing; Advanced interpolation algorithms use bilinear interpolation and bicubic interpolation or more advanced interpolation algorithms to improve the accuracy and naturalness of restored pixels.

3. The image enhancement method according to claim 2, characterized in that: The adaptive texture synthesis and restoration specifically includes: Combining the local and global texture features of the image, an adaptive texture synthesis algorithm is used to generate a repair result that is highly consistent with the surrounding area based on the texture and brightness information of the pixels around the bad block, achieving a more natural repair effect.

4. The image enhancement method according to claim 3, characterized in that: The multi-scale bad block detection and hierarchical repair specifically include: The bad blocks of the image are detected at different scales. According to the size and distribution characteristics of the bad blocks, a layered repair strategy is adopted to repair large-scale bad blocks first and then gradually refine them to small-scale bad blocks to achieve a more refined repair effect.

5. The image enhancement method according to claim 4, characterized in that: The detail enhancement technology specifically includes: Edge enhancement, using the Sobel operator and Canny edge detection algorithm to enhance the edge details in the image and improve the clarity of the image; Frequency domain enhancement converts the image to the frequency domain through Fourier transform, enhances the high-frequency components, and then inversely transforms back to the spatial domain to achieve detail enhancement.

6. The image enhancement method according to claim 5, characterized in that: The brightness adaptive enhancement specifically includes: Based on histogram equalization, according to the histogram distribution of the image, the brightness distribution is dynamically adjusted to enhance the contrast of the image; Application of Retinex theory, using Retinex theory to simulate the human visual system's perception of brightness and color to achieve more natural brightness enhancement.

7. The image enhancement method according to claim 6, characterized in that: The color correction and enhancement specifically include: Color space conversion: convert images from RGB color space to HSV or Lab color space, and adjust brightness, saturation and hue respectively to achieve more refined color correction; Color balance adjustment automatically adjusts the proportion of the three primary colors of red, green and blue according to the overall color distribution of the image to make the colors more balanced and natural.

8. An image enhancement method according to claim 7, characterized in that: The advanced filtering technology specifically includes: Non-local mean filtering uses the information of similar pixel blocks in the image to filter, effectively removing noise while retaining image details; Guided filtering, combining edge-preserving filtering and guided images, achieves smooth denoising while preserving edge details.

9. The image enhancement method according to claim 8, characterized in that: The image clearing technology specifically includes: Deblurring algorithm: for image blur caused by camera shake or motion blur, a blind deblurring algorithm or a point spread function-based deblurring algorithm is used to perform clear processing; Super-resolution reconstruction uses deep learning models to perform super-resolution reconstruction of images.

10. An image enhancement method according to claim 9, characterized in that: The implementation of the comprehensive optimization strategy specifically includes: Modular design, modularizing various optimization technologies; Automation and intelligence: using machine learning algorithms to adjust parameters and achieve intelligent processing; Real-time and resource optimization: While ensuring the optimization effect, the real-time performance and resource occupancy of the algorithm are considered to optimize the algorithm implementation and reduce the computational complexity.

Images