Power transmission line image enhancement method

Abstract

The power transmission line image enhancement method replaces the traditional single gray level hard clipping histogram processing logic, avoids the damage of single gray level frequency adjustment to the continuous gray distribution of the core components such as the conductor and fittings of the power transmission line from the processing mechanism, and then performs frequency redistribution of the redundant frequency generated by the soft limiting within the corresponding continuous gray cluster, completes the optimization adjustment of the histogram distribution, and finally performs gray mapping processing based on the adjusted gray histogram, realizes the image contrast improvement and accurately highlights the defect area features, at the same time, guarantees the smooth and continuous gray distribution of the power transmission components, effectively avoids the generation of gray discontinuity and edge artifacts, improves the distinguishability of the defect area and the background, and provides high-quality image data basis for subsequent power transmission line state detection and defect identification.

Description

Technical Field

[0001] This application relates to the field of power system equipment condition monitoring, specifically to a method for enhancing images of transmission lines. Background Technology

[0002] Currently, power transmission lines are the core transmission carrier of smart grids, and their inspection visualization images have become the core data foundation for line defect identification and condition assessment. In complex outdoor environments, acquired images are easily affected by uneven lighting, noise interference, and background obfuscation, resulting in low contrast and blurred details in defect areas. There is an urgent need for high-precision image enhancement technology adapted to power transmission scenarios to ensure the accuracy of subsequent intelligent detection.

[0003] In related technologies, transmission line image enhancement is based on the CLAHE algorithm. Existing solutions mostly use swarm intelligence algorithms to optimize the key parameters of CLAHE in order to solve the problem of poor adaptability of manual parameter tuning.

[0004] However, it still has not broken through the limitations of the underlying logic of the traditional CLAHE "single gray level hard clipping + uniform redistribution". This processing method will destroy the continuous gray level distribution characteristics of core components such as conductors and fittings in the transmission line, which can easily cause local gray level abrupt changes, gray level discontinuities and edge artifacts, resulting in blurred boundaries between the defect area and the background, directly reducing the accuracy of subsequent defect identification. Summary of the Invention

[0005] This application provides a method for enhancing images of power transmission lines, which can solve the technical problems existing in related technologies, such as the destruction of continuous grayscale distribution of power transmission components, grayscale discontinuity and edge artifacts, and insufficient identification of defect areas due to the continued use of CLAHE hard clipping histogram processing logic.

[0006] This application provides a method for enhancing images of power transmission lines, the method comprising: Extract the effective grayscale range of the grayscale image to be enhanced, and perform grayscale clustering on the grayscale levels within the effective grayscale range to obtain multiple continuous grayscale clusters; Perform cluster-level soft clipping on the total grayscale frequency of each of the continuous grayscale clusters to obtain clipped grayscale frequency data. The cluster-level soft clipping is used to replace the single grayscale hard clipping operation. The redundant frequencies generated by the cluster-level soft limiting process are redistributed within the corresponding continuous gray-level clusters to obtain an adjusted gray-level histogram. Based on the adjusted grayscale histogram, grayscale mapping processing is performed to obtain an enhanced transmission line image. The gray-level clustering within the effective gray-level range yields multiple consecutive gray-level clusters, including: Based on gray-level similarity, unsupervised clustering is performed on the gray levels within the effective gray-level range, dividing continuous or similar gray-level values ​​into multiple continuous gray-level clusters, which correspond to different gray-level feature regions of the transmission line image.

[0007] In one embodiment, performing cluster-level soft clipping on the total grayscale frequency of each of the consecutive grayscale clusters to obtain clipped grayscale frequency data includes: Calculate the total gray frequency corresponding to each consecutive gray cluster, where the total gray frequency is the sum of the histogram frequencies of all gray levels within the cluster; Determine the cluster-level limiting threshold and compare the total gray frequency of each consecutive gray cluster with the cluster-level limiting threshold; For continuous gray-level clusters whose total gray-level frequency exceeds the cluster-level limiting threshold, the histogram frequency of all gray-levels within the cluster is synchronously scaled proportionally to complete the soft limiting process. For consecutive grayscale clusters whose total grayscale frequency does not exceed the cluster-level amplitude limit threshold, their original grayscale frequency is retained.

[0008] In one implementation, the redundant frequencies generated by the cluster-level soft clipping process are redistributed within the corresponding consecutive gray-level clusters to obtain an adjusted gray-level histogram, including: Calculate the redundancy frequency generated by consecutive grayscale clusters after cluster-level soft limiting processing; Obtain the texture detail features of the image region corresponding to each gray level within the continuous gray cluster, and determine the frequency allocation weight corresponding to each gray level based on the texture detail features; Based on the frequency allocation weight, the redundant frequencies are allocated to the corresponding gray levels within the continuous gray cluster, completing the frequency redistribution and obtaining the adjusted gray histogram.

[0009] In one implementation, the step of performing grayscale mapping processing based on the adjusted grayscale histogram to obtain an enhanced transmission line image includes: Based on the adjusted grayscale histogram, local grayscale stretching mapping within each continuous grayscale cluster is performed to obtain the preliminary mapped grayscale value corresponding to the grayscale within each cluster. Perform global gray-level normalization mapping on the initial mapped gray values ​​of all continuous gray-level clusters to map the gray values ​​to the standard gray-level range; Smoothing is performed on the gray-level transition regions of adjacent continuous gray-level clusters to obtain the final enhanced gray-level data. An enhanced transmission line image is generated and output based on the enhanced gray-level data.

[0010] In one embodiment, extracting the effective grayscale range of the grayscale image to be enhanced includes: Statistically analyze the gray-level histogram distribution of the gray-level image to be enhanced; Based on the grayscale histogram distribution, extreme grayscale values ​​at both ends of the histogram are removed, and the grayscale range including the core components and defect areas of the transmission line is retained to form the effective grayscale range.

[0011] In one embodiment, before extracting the effective grayscale range of the grayscale image to be enhanced, the method further includes: Acquire images of power transmission line inspections to be processed; The inspection image of the transmission line is preprocessed to obtain the grayscale image to be enhanced. The preprocessing includes edge-preserving denoising, which is used to filter out image acquisition noise and environmental interference noise while retaining the edge contour information of the core components of the transmission line.

[0012] In one embodiment, the preprocessing of the transmission line inspection image to obtain the grayscale image to be enhanced includes: When the transmission line inspection image is a color image, the grayscale data corresponding to its brightness channel is extracted as the data to be processed. When the transmission line inspection image is a grayscale image, it is directly used as the data to be processed; The edge-preserving denoising process is performed on the data to be processed to obtain the grayscale image to be enhanced.

[0013] In one embodiment, the method further includes: Before performing image enhancement processing, the core control parameters of the image enhancement process are adaptively optimized using the Newton's Downhill Iterative Optimization Algorithm to obtain the optimal parameter combination that is suitable for the current transmission line inspection image. Based on the optimal parameter combination, the corresponding gray-level clustering, cluster-level soft limiting, frequency redistribution, and effective gray-level interval extraction steps are executed.

[0014] In one implementation, the adaptive optimization of the core control parameters of the image enhancement process using the Newton's Downhill Iterative Optimization Algorithm includes: The value range of the core control parameters to be optimized is initialized to generate an initial optimization population; the core control parameters include the number of gray-level clusters, the limiting coefficient of cluster-level soft limiting, the weight coefficient of frequency redistribution, and the extreme gray-level removal ratio of the effective gray-level range. Construct a fitness function that integrates multi-dimensional image enhancement effect evaluation metrics; With the goal of finding the optimal value of the fitness function, Newton's Downhill Iterative Optimization is performed to complete the iterative update of the population and the selection of optimal solutions. After the iteration terminates, the globally optimal parameter combination is output.

[0015] The beneficial effects of the technical solutions provided in this application include: By extracting the effective grayscale range of the grayscale image to be enhanced, the target grayscale range is accurately locked. Then, grayscale clustering is performed on the grayscale levels within the effective grayscale range to obtain multiple continuous grayscale clusters. A clustering processing unit based on grayscale feature clustering is established to adapt to the multi-peak grayscale distribution characteristics of transmission line inspection images. Subsequently, cluster-level soft limiting processing is performed on the continuous grayscale clusters as a whole unit, replacing the traditional histogram processing logic of hard cropping of single grayscale levels. From the processing mechanism, this avoids the destruction of the continuous grayscale distribution of core components such as transmission line conductors and fittings by adjusting the frequency of single grayscale levels. Then, by redistributing the redundant frequencies generated by soft limiting within the corresponding continuous grayscale clusters, the histogram distribution is optimized and adjusted. Finally, grayscale mapping processing is performed based on the adjusted grayscale histogram. While improving image contrast and accurately highlighting the features of defect areas, this process ensures the smooth and continuous grayscale distribution of transmission components, effectively avoids the generation of grayscale discontinuities and edge artifacts, and improves the distinction between defect areas and background. This provides a high-quality image data foundation for subsequent transmission line condition detection and defect identification. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating an embodiment of the transmission line image enhancement method of this application; Figure 2 This is a schematic flowchart of another embodiment of the transmission line image enhancement method of this application. Detailed Implementation

[0017] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0018] Currently, the construction of smart grids is continuously deepening and advancing. As the core carrier of power transmission in the power grid, the operating status of transmission lines directly determines the safe and stable power supply level of the grid. With the large-scale application of drone inspections and fixed visual monitoring equipment in the operation and maintenance of transmission lines, the visual acquisition of images of the lines has become the core data foundation for tower structure inspection, conductor defect identification, hardware condition assessment, and insulator fault diagnosis. However, transmission lines are mostly located in complex outdoor environments, and the acquired images are easily affected by severe uneven lighting (backlight, overexposure, underexposure, etc.), equipment acquisition noise (salt-and-pepper noise, Gaussian noise, etc.), and complex background interference (sky, vegetation, and grayscale confusion with transmission components, etc.). These images suffer from problems such as low contrast in key defect areas, blurred details, and significant noise interference, which directly restrict the detection accuracy and reliability of subsequent intelligent recognition algorithms. Therefore, it is urgent to develop image enhancement technologies adapted to the characteristics of transmission line scenarios, which can improve the contrast of defect areas while ensuring the structural integrity, grayscale continuity, and noise suppression capabilities of transmission components, providing high-quality image data support for the accurate operation and maintenance of intelligent inspection of transmission lines.

[0019] Among them, contrast-limited adaptive histogram equalization (CLAHE) is the mainstream technology for enhancing transmission line images. To address the shortcomings of traditional CLAHE, such as reliance on manual parameter tuning and poor scene adaptability, a technical solution has emerged that combines swarm intelligence algorithms such as particle swarm optimization and differential evolution with CLAHE. By using swarm intelligence algorithms to automatically optimize key parameters of CLAHE, such as block size and cropping threshold, the algorithm's adaptability to different inspection scenarios is improved, thereby enhancing the contrast of transmission line images.

[0020] However, the aforementioned existing technologies still have insurmountable technical defects. First, existing technologies have not broken through the limitations of the traditional CLAHE's underlying logic of "single gray-level hard cropping + uniform redistribution." Hard cropping processing will destroy the continuous gray-level distribution characteristics of core components such as conductors and fittings in transmission lines, easily causing local gray-level abrupt changes, resulting in problems such as gray-level discontinuity and edge artifacts, leading to blurred boundaries between defect areas and the background, directly reducing the accuracy of subsequent defect identification. Second, the swarm intelligence algorithm used in existing technologies has poor adaptability to the scene characteristics of multi-peak gray-level distribution and local low contrast in defect areas of transmission line images. It is prone to slow convergence speed and getting trapped in local optima during parameter optimization, and cannot achieve an effective balance between image enhancement accuracy and algorithm processing efficiency. Third, existing solutions lack customized design for transmission line inspection scenarios. Parameter optimization lacks clear targeting, only pursuing the improvement of global contrast without focusing on the detail enhancement needs of defect areas. It cannot effectively solve scene-specific problems such as overexposure, underexposure, and gray-level confusion between components and the background, and cannot meet the core application requirements of intelligent transmission line inspection for accurate highlighting of defect features.

[0021] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0022] In a first aspect, embodiments of this application provide a method for enhancing images of power transmission lines.

[0023] In one embodiment, reference is made to Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the transmission line image enhancement method of this application. Figure 1 As shown, the image enhancement method for transmission lines includes: S100: Extract the effective grayscale range of the grayscale image to be enhanced, and perform grayscale clustering on the grayscale levels within the effective grayscale range to obtain multiple continuous grayscale clusters; S200: Perform cluster-level soft clipping on the total grayscale frequency of each of the continuous grayscale clusters to obtain clipped grayscale frequency data. The cluster-level soft clipping is used to replace the single grayscale hard clipping operation. S300: The redundant frequencies generated by the cluster-level soft limiting process are redistributed within the corresponding continuous gray-level clusters to obtain an adjusted gray-level histogram. S400: Perform grayscale mapping processing based on the adjusted grayscale histogram to obtain an enhanced transmission line image.

[0024] In this embodiment, the target grayscale range is accurately locked by extracting the effective grayscale range of the grayscale image to be enhanced. Then, grayscale clustering is performed on the grayscale levels within the effective grayscale range to obtain multiple continuous grayscale clusters. A clustering processing unit based on grayscale feature clustering is established to adapt to the multi-peak grayscale distribution characteristics of transmission line inspection images. Subsequently, cluster-level soft limiting processing is performed on the continuous grayscale clusters as a whole unit, replacing the traditional histogram processing logic of hard clipping of single grayscale levels. From the processing mechanism, this avoids the impact of single grayscale level frequency adjustment on core components such as transmission line conductors and fittings. The disruption of the continuous grayscale distribution of components is addressed by redistributing redundant frequencies generated by soft limiting within the corresponding continuous grayscale clusters to optimize the histogram distribution. Finally, grayscale mapping processing is performed based on the adjusted grayscale histogram. This process enhances image contrast, accurately highlights defective area features, ensures the smooth and continuous grayscale distribution of transmission components, effectively avoids grayscale discontinuities and edge artifacts, improves the distinction between defective areas and the background, and provides a high-quality image data foundation for subsequent transmission line condition detection and defect identification.

[0025] Furthermore, in one embodiment, step S200 includes the following steps: S201: Calculate the total gray frequency corresponding to each consecutive gray cluster, wherein the total gray frequency is the sum of the histogram frequencies of all gray levels within the cluster; S202: Determine the cluster-level limiting threshold and compare the total gray frequency of each continuous gray cluster with the cluster-level limiting threshold; S203: For continuous gray-level clusters whose total gray-level frequency exceeds the cluster-level limiting threshold, the histogram frequency of all gray-levels in the cluster is synchronously scaled proportionally to complete the soft limiting process; S204: For continuous grayscale clusters whose total grayscale frequency does not exceed the cluster-level amplitude limiting threshold, retain their original grayscale frequency.

[0026] In this embodiment, the total gray-level frequency is counted for each continuous gray-level cluster as a whole. Based on the total gray-level frequency, a cluster-level limiting threshold is determined and a comparison and verification are performed. For continuous gray-level clusters whose total gray-level frequency exceeds the cluster-level limiting threshold, soft limiting is performed by synchronously scaling the histogram frequency of all gray-levels within the cluster. For continuous gray-level clusters that do not exceed the threshold, the original gray-level frequency is directly retained. This replaces the traditional single-gray-level hard cropping frequency processing method, avoids the gray-level distribution break caused by single-gray-level frequency removal, ensures the relative distribution relationship and continuous transition characteristics between gray levels within the cluster, and simultaneously achieves the normalization adjustment of the over-concentrated area of ​​the histogram. This balances the dual requirements of image contrast enhancement and gray-level distribution continuity, and provides a standardized histogram data foundation for subsequent frequency redistribution processing.

[0027] Furthermore, in one embodiment, step S300 includes the following steps: S301: Calculate the redundant frequency of consecutive gray-scale clusters after cluster-level soft limiting processing; S302: Obtain the texture detail features of the image region corresponding to each gray level within the continuous gray cluster, and determine the frequency allocation weight corresponding to each gray level based on the texture detail features; S303: Based on the frequency allocation weight, the redundant frequencies are allocated to the corresponding gray levels within the continuous gray cluster to complete the frequency redistribution and obtain the adjusted gray histogram.

[0028] In this embodiment, the redundant frequencies generated by the continuous gray-level clusters after cluster-level soft limiting are first accurately calculated. Then, the texture detail features of the image regions corresponding to each gray level within the continuous gray-level cluster are obtained. Based on this, the frequency allocation weights corresponding to each gray level are determined. Finally, based on the frequency allocation weights, the redundant frequencies are allocated to the corresponding gray levels within the continuous gray-level cluster, completing the frequency redistribution and obtaining an adjusted gray-level histogram. This achieves differentiated allocation of redundant frequencies, allowing regions with richer texture details to receive more frequency allocations for their corresponding gray levels. While targeting and improving the detail contrast of the transmission line defect area, it avoids ineffective enhancement and noise amplification of the smooth background area, and ensures the continuous transition characteristics of the gray-level distribution within the cluster. This provides optimized histogram data adapted to the inspection defect identification requirements for subsequent gray-level mapping processing.

[0029] Furthermore, in one embodiment, step S400 includes the following steps: S401: Based on the adjusted grayscale histogram, perform local grayscale stretching mapping within each continuous grayscale cluster to obtain the preliminary mapped grayscale value corresponding to the grayscale within each cluster. S402: Perform global gray-level normalization mapping on the initial mapped gray values ​​of all continuous gray-level clusters to map the gray values ​​to the standard gray-level range; S403: Perform smoothing processing on the gray-level transition area of ​​adjacent continuous gray-level clusters to obtain the final enhanced gray-level data, and generate and output the enhanced transmission line image based on the enhanced gray-level data.

[0030] In this embodiment, based on the adjusted grayscale histogram, local grayscale stretching mapping within each continuous grayscale cluster is first performed to specifically enhance the internal contrast of each grayscale feature region, obtaining the preliminary mapped grayscale value corresponding to the grayscale within each cluster. Then, global grayscale normalization mapping is performed on the preliminary mapped grayscale values ​​of all continuous grayscale clusters to uniformly map the grayscale values ​​to the standard grayscale range, ensuring the consistency of the global grayscale distribution of the entire image. Finally, smoothing processing is performed on the grayscale transition areas of adjacent continuous grayscale clusters to eliminate possible grayscale jumps between clusters, ensuring smooth grayscale transitions of continuous structural components such as transmission line conductors and fittings. Ultimately, enhanced grayscale data is obtained and an enhanced transmission line image is generated and output. While accurately highlighting the detailed features of each region, it completely avoids problems such as grayscale discontinuity and block effects, outputting a high-quality enhanced image that meets the needs of intelligent inspection and defect identification of transmission lines.

[0031] Furthermore, in one embodiment, S100 includes the following steps: S101: Statistically analyze the gray-level histogram distribution of the gray-level image to be enhanced; S102: Based on the grayscale histogram distribution, the extreme grayscale values ​​at both ends of the histogram are removed, and the grayscale range including the core components and defect areas of the transmission line is retained to form the effective grayscale range.

[0032] In this embodiment, by statistically analyzing the grayscale histogram distribution of the grayscale image to be enhanced, the grayscale distribution characteristics of the entire image range are fully obtained. Then, based on the grayscale histogram distribution, extreme grayscale values ​​at both ends of the histogram are removed, accurately preserving the grayscale range including the core components and defect areas of the transmission line and forming an effective grayscale interval. This eliminates the interference of invalid grayscale values ​​such as overexposure and pure black that have no inspection value on the subsequent grayscale clustering and histogram adjustment stages, improves the accuracy of subsequent clustering and the targeting of enhancement processing, and provides accurate and effective grayscale processing objects for the subsequent core steps of image enhancement.

[0033] Furthermore, in one embodiment, S100 includes the following steps: S103: Based on gray-level similarity, perform unsupervised clustering on the gray levels within the effective gray-level range, dividing continuous or similar gray-level values ​​into multiple continuous gray-level clusters, each of which corresponds to a different gray-level feature region of the transmission line image.

[0034] In this embodiment, gray-level similarity is used as the clustering basis to perform unsupervised clustering on gray levels within the effective gray-level range. Continuous or similar gray values ​​are divided into multiple continuous gray-level clusters, so that each continuous gray-level cluster corresponds to a different gray-level feature region of the transmission line image. This adapts to the multi-peak gray-level distribution characteristics of transmission line inspection images, achieving accurate division of different gray-level feature regions. This provides a clear division basis for subsequent histogram soft limiting and frequency redistribution processing based on continuous gray-level clusters, while ensuring the continuous correlation characteristics of gray levels within the same cluster, providing a prerequisite for protecting the continuity of gray-level distribution in subsequent processing stages.

[0035] Furthermore, in one embodiment, before extracting the effective grayscale range of the grayscale image to be enhanced, the following steps are also included: S010: Acquire the inspection images of the power transmission lines to be processed; S020: Perform preprocessing on the transmission line inspection image to obtain the grayscale image to be enhanced. The preprocessing includes edge-preserving denoising processing, which is used to retain the edge contour information of the core components of the transmission line while filtering out image acquisition noise and environmental interference noise.

[0036] In this embodiment, the inspection images of the transmission line to be processed are first acquired, and then preprocessing operations including edge-preserving noise reduction are performed on the acquired transmission line inspection images to obtain grayscale images to be enhanced. While effectively filtering out various noises introduced by equipment acquisition and outdoor environmental interference, the edge contour information of core components such as transmission line conductors, hardware, and insulators is strictly preserved. This avoids problems such as edge blurring and structural distortion in subsequent image enhancement processing, and provides a high-quality and highly reliable image data foundation for subsequent core enhancement steps such as effective grayscale interval extraction and grayscale clustering. This adapts to the image data processing needs of various transmission line inspection acquisition terminals.

[0037] Furthermore, in one embodiment, S020 includes the following steps: S021: When the transmission line inspection image is a color image, extract the grayscale data corresponding to its brightness channel as the data to be processed. S022: When the transmission line inspection image is a grayscale image, it is directly used as the data to be processed; S023: Perform the edge-preserving denoising process on the data to be processed to obtain the grayscale image to be enhanced.

[0038] In this embodiment, a differentiated grayscale data extraction strategy is adopted for the color and grayscale image formats output by different acquisition terminals during transmission line inspection. The grayscale data corresponding to the brightness channel of the color image is extracted as the data to be processed to avoid color distortion problems in subsequent enhancement processing. The grayscale image is directly used as the data to be processed to ensure the adaptability of inspection images of different formats. Then, edge-preserving denoising processing is uniformly performed on the extracted data to be processed. While effectively filtering out various noises introduced by image acquisition and environmental interference, the edge contour information of the core components of the transmission line is preserved. Finally, a standardized grayscale image to be enhanced is obtained, which provides high-quality and highly adaptable basic data for subsequent core enhancement steps such as effective grayscale interval extraction and grayscale clustering.

[0039] Furthermore, in one embodiment, a parameter adaptive optimization step is included before performing image enhancement processing: S001: By using the Newton's Downhill Iterative Optimization Algorithm, the core control parameters of the image enhancement process are adaptively optimized to obtain the optimal parameter combination that is suitable for the current transmission line inspection image. S002: Based on the optimal parameter combination, perform the corresponding gray-level clustering, cluster-level soft limiting, frequency redistribution, and effective gray-level interval extraction steps.

[0040] In this embodiment, before performing image enhancement processing, the core control parameters of the image enhancement process are adaptively optimized using the Newton's Downhill Iterative Optimization Algorithm to obtain the optimal parameter combination that fits the current transmission line inspection image. Then, based on the optimal parameter combination, the corresponding gray-level clustering, cluster-level soft limiting, frequency redistribution, and effective gray-level interval extraction steps are executed. This achieves automated and scenario-based adaptation of the core control parameters for image enhancement, eliminating the need for manual parameter adjustment for images under different inspection scenarios and lighting conditions. At the same time, relying on the characteristics of the Newton's Downhill Iterative Optimization Algorithm, the convergence speed and global solution accuracy of parameter optimization are improved, avoiding the defect of traditional optimization algorithms that are prone to getting trapped in local optima. This ensures the adaptability of each image enhancement processing step to the current inspection image, improves the accuracy of image enhancement processing, and enhances its versatility for complex and ever-changing outdoor transmission line inspection scenarios.

[0041] Furthermore, in one embodiment, step S001 includes the following steps: S0011: Initialize the value range of the core control parameters to be optimized and generate an initial optimization population; the core control parameters include the number of gray-level clusters, the limiting coefficient of cluster-level soft limiting, the weighting coefficient of frequency redistribution, and the extreme gray-level removal ratio of the effective gray-level range. S0012: Construct a fitness function that integrates multi-dimensional image enhancement effect evaluation metrics; S0013: With the optimal value of the fitness function as the objective, perform Newton's Downhill Iterative Optimization to complete the iterative update of the population and the selection of optimal solutions. After the iteration terminates, output the globally optimal parameter combination.

[0042] In this embodiment, the value range of the core control parameters to be optimized is initialized and an initial optimization population is generated, clarifying the composition of the core control parameters and providing an initial search space and search object for parameter optimization. Then, a fitness function that integrates multi-dimensional image enhancement effect evaluation indicators is constructed to clarify the quantitative optimization target of parameter optimization, ensuring that the optimization direction aligns with the actual needs of transmission line image enhancement. Subsequently, with the optimal value of the fitness function as the objective, Newton's downhill iterative optimization is performed to complete the iterative update of the population and the selection of optimal solutions. After the iteration terminates, the globally optimal parameter combination is output. Relying on the mechanism of Newton's downhill iterative optimization, the convergence speed and solution accuracy of parameter optimization are improved, avoiding getting trapped in local optima. This provides core control parameters that are accurately adapted to the current inspection image for subsequent image enhancement steps, ensuring the effect of image enhancement processing and its versatility for complex and ever-changing outdoor transmission line inspection scenarios.

[0043] Secondly, embodiments of this application propose an end-to-end solution that integrates grayscale cluster soft constraints with Newton's downhill iteration.

[0044] Briefly: First, edge-preserving denoising of the transmission line image is achieved through a combination of median filtering and bilateral filtering, and extreme gray levels are adaptively removed to extract the effective gray level range. Then, the Newton's Downhill Iterative Algorithm (NDO) is used, with the specific fitness function of the transmission line as the target, to adaptively optimize four core parameters: gray level cluster number K, soft limiting coefficient α, variance weight coefficient γ, and extreme gray level removal ratio T, solving the problems of slow convergence and easy getting trapped in local optima in traditional algorithms. The core enhancement stage abandons the traditional CLAHE hard cropping logic, and uses fast K-means to cluster the effective gray levels into continuous gray level clusters. Soft limiting is applied to the total frequency of the cluster level instead of hard cropping of a single gray level, and redundant frequencies are non-uniformly distributed according to the gray level variance within the cluster. Combined with intra-cluster and global two-layer CDF mapping, gray level smoothing is achieved. Finally, after guided filtering and gray level normalization post-processing, an enhanced image with no discontinuities, clear defects, and effective noise suppression is output. The entire solution is tailored to the characteristics of transmission line images with multi-peak grayscale and complex outdoor lighting, taking into account both real-time engineering requirements and enhanced accuracy. It can be directly adapted to the intelligent inspection needs of field equipment such as drones and edge computing boxes.

[0045] Step 1: Data Acquisition and Preprocessing The data for this solution comes from power transmission line visualization monitoring equipment, including image data retrieved by drones, high-definition inspection cameras, and line inspection robots. The data format supports grayscale and color images. Color images are subsequently enhanced only on the brightness channel to avoid color distortion. The solution comprehensively covers different outdoor scenarios, including complex lighting conditions such as backlight, strong light, and rainy days, diverse background environments such as sky, trees, and mountains, and images of power transmission lines with different defect types such as broken conductor strands, corroded hardware, and cracked insulators, ensuring the scenario adaptability and versatility of the solution.

[0046] The acquired raw images need to undergo a preprocessing process. First, to address the salt-and-pepper noise introduced by the equipment acquisition and the Gaussian noise caused by environmental interference, a combined strategy of "3×3 median filtering + 5×5 bilateral filtering" is adopted. The 3×3 small window median filtering can quickly remove salt-and-pepper noise without blurring the edges of fine structures such as conductors and fittings. The 5×5 bilateral filtering, while smoothing Gaussian noise, strictly preserves the edge contours of core components through gray-level similarity weights, adapting to the structural characteristics of transmission lines. Subsequently, to avoid meaningless extreme gray-level interference from overexposed skies (gray-level 240-255) and pure black backgrounds (gray-level 0-5) in subsequent gray-level clustering and parameter optimization, the cumulative proportion of the gray-level histogram of the denoised image is statistically analyzed to adaptively remove the first T% and the last T% of extreme gray levels, where T is the parameter to be optimized later. This accurately extracts the effective gray-level range containing towers, conductors, and defect areas, providing a high-quality data foundation for subsequent core enhancement stages.

[0047] Step 2: NDO Iterative Parameter Optimization This step uses NDO to adaptively find the optimal parameter combination [K, α, γ, T] that fits the transmission line image within a preset parameter space, solving the problems of slow convergence and easy getting trapped in local optima in traditional swarm intelligence algorithms, and ensuring that subsequent enhancement steps accurately meet the needs of the scenario.

[0048] 1. Parameter initialization First, it is necessary to clarify the four core parameters to be optimized and their physical meaning. Based on the characteristics of the scene, such as the multi-peak grayscale of the transmission line image and the features of the defect area, reasonable value boundaries should be set to avoid meaningless optimization and improve efficiency.

[0049] (1) Number of grayscale clusters K The number of clusters in the fast K-means algorithm determines the fineness of the gray-level cluster division; the value range is as follows. K ∈[4, 16], take the integer, 4~16 can fully distinguish the core areas such as dark area, medium gray conductor area, bright area hardware; (2) Cluster-level soft limiting coefficient α : Determines the threshold value for the total cluster frequency, balancing contrast enhancement and noise suppression; value range α∈[1.0, 5.0], a floating-point number, consistent with the logic of traditional CLAHE clipping coefficients; (3) Intra-cluster variance weighting coefficient c : Controls the allocation bias of redundant frequencies within a cluster, with a value range c ∈[0.0,1.0], floating-point number. c =0 indicates uniform distribution. c =1 is allocated entirely according to variance; (4) Extreme grayscale removal ratio T : Controls the degree of removal of extreme gray levels during the preprocessing stage, with a value range of T ∈[5,20], take integers, unit is %, to avoid removing too much effective grayscale or too little.

[0050] Then, initialize the NDO parameters, including the population size. N Problem Dimension d (This plan belongs to) d =4-dimensional), upper and lower bounds of variables [ l j , u j Maximum number of iterations T max Minimum value e =10 6 and generate individuals x i =[ K i , α i , c i , T i The individual's location is determined using these four variables as coordinates. The first position of each individual is calculated. j Dimensional parameter position x i，j Generate random numbers in the range [0, 1]. R The calculation formula is as follows: ;

[0051] In the formula, R The values ​​are random numbers in the range [0, 1]. The initial population matrix is ​​calculated using the results from the above formula. X :

[0052] 2. Fitness Function Construction Individual parameters x i Using the core independent variable, construct a fitness function that maximizes the objective.F ( x i ), F ( x i The larger the value, the stronger the fit of the parameter set. The core evaluation index for fusion transmission line image enhancement (all normalized to [0, 1]) is as follows:

[0053] In the formula, oh 1 represents the local contrast weight of the key region; oh 2 represents the edge shape preservation weight; oh 3 represents the grayscale continuity weight; oh 4 represents the noise suppression weight. The weight allocation should follow the principle of prioritizing the prominence of transmission line defects and preserving edge conformity as the basis. C key ( x i ) is a parameter x i The local contrast of key areas such as conductors, fittings, and defects is enhanced using the following formula:

[0054] In the formula, N ROI The number of pixels in the key area; ( i , j ) is a parameter x i Enhanced image grayscale values; The average gray level of the 3×3 neighborhood; =10 6 To prevent the denominator from being 0. E pres ( x i ) is a parameter x i The edge shape preservation of the enhanced image is calculated using the following formula:

[0055] In the formula, G pre This represents the gradient magnitude after preprocessing. For parameters x i Enhanced gradient magnitude It is an L2 norm. S gray ( x i ) is a parameter xi The grayscale continuity of the enhanced image is calculated using the following formula:

[0056] In the formula, M The effective grayscale range length. To enhance the grayscale level k The frequency of the histogram. N supp ( x i ) is a parameter x i The noise suppression degree of the enhanced image is calculated using the following formula:

[0057] s noise,pre The variance of the noise region after preprocessing. To enhance the variance of the noise region.

[0058] 3. NDO iteration Let the current iteration number t=1, and execute the loop. After initialization, calculate the fitness value of each individual. F ( x i Record the optimal position of an individual. x i , pbest Individual optimal fitness F i , pbest And select the globally optimal position. x best Global optimal fitness F best .

[0059] (1) Update of the downhill mechanism For each individual x i Calculate the downhill step length factor of Newton. l And initially update the location. To ensure that the iteration progresses in the direction of improving fitness, the formula is as follows:

[0060]

[0061] right Each dimension of the variable Implement boundary control:

[0062] (2) Population location update Generate random numbers in the range [0, 1]. r 3. Select update mode.

[0063] ① Exploration stage ( r 3≤0.5) For each individual x i Generate random weights W 1. W 2. W 3∈[0, 1], randomly select individuals from the population. x a , x b ( a , b (≠0), update the position using the following formula:

[0064] Boundary control will be implemented after the update.

[0065] ② Development stage ( r 3>0.5) Generate random numbers in the range [0, 1]. r 4. Two sub-modes:

[0066] In the formula, W 4. W 5. W 6. W 7∈[0, 1] represents random weights; x c ( c (≠0) represents randomly selected individuals in the population. Boundary control is still required after the update.

[0067] (3) Positive greedy choice For each individual x i Calculate the fitness of the new solution F ( The optimal solution is retained according to the following formula:

[0068] After each iteration, update the current individual optimum and the global optimum. For each individual... x i If its current fitness F ( x i () greater than the individual's own historical best fitness F i,pbest Then update the individual's optimal position. xi,pbest = x i Individual optimal fitness F i,pbest = F ( x i ); Traverse all individuals, if the current individual has the best fitness F i,pbest Greater than the best fitness in the entire population's history F best Then update the global optimal position. x best = x i,pbest Global optimal fitness F best = F i,pbest .

[0069] (4) Iteration count update t If it increments by 1, t > T If the loop terminates, then return to step 3 and iterate using NDO.

[0070] (5) Output of results Output the globally optimal individual x best =[ K best , α best , c best , T best ].

[0071] Step 3: Gray-scale cluster soft constraint CLAHE core enhancement This step is based on the optimal parameter combination obtained from the NDO iteration in step two. K best , α best , c best , T best The preprocessed transmission line image is enhanced by performing grayscale cluster soft constraint CLAHE enhancement, which solves the problems of grayscale discontinuity and edge distortion caused by traditional CLAHE hard cropping, and targets the contrast of defect areas while preserving the edge contours of core components such as conductors and fittings.

[0072] 1. Gray-scale unsupervised clustering Based on the "effective grayscale range" extracted in step one, fast K-means unsupervised clustering is performed on the image grayscale levels to divide continuous or similar grayscale values ​​into... K best Each grayscale cluster includes dark area clusters, medium gray conductor clusters, bright area hardware clusters, and defect feature clusters, adapting to the multi-peak grayscale distribution characteristics of transmission line images.

[0073] (1) Determine cluster samples: Extract all gray values ​​of the preprocessed image. g ∈[ g min , g max ], g min , g max These are the upper and lower bounds of the effective grayscale after removing extreme grayscale values ​​in step one; (2) Initialize cluster centers: randomly select K best One gray value is used as the initial cluster center. c 1, c 2, …, c Kbest},satisfy c 1< c 2<…< c Kbest ; (3) Iterative clustering convergence: For each gray value g Assigned to the nearest cluster center, i.e. g ∈ C m , C m For the first m A grayscale cluster, satisfying:

[0074] Then, the mean gray value of each cluster is recalculated as the new cluster center, using the following formula:

[0075] In the formula, N m For the first m The number of gray values ​​contained in each gray cluster. The K-means iteration terminates when the changes in the centers of all clusters satisfy the following relationship, ultimately yielding... K best A grayscale cluster { C 1, C 2, …, C Kbest}

[0076] 2. Cluster-level soft limiting calculation Traditional CLAHE uses "hard clipping" to remove frequencies exceeding a threshold, which easily leads to grayscale banding. This solution is based on... α best Cluster-level soft clipping is implemented to abandon hard cropping, balancing contrast enhancement and grayscale continuity.

[0077] (1) Calculate the first m The total frequency of each gray cluster H m :

[0078] In the formula, h ( g () represents the grayscale value g The histogram frequency, i.e., the gray values ​​in the image. g The number of pixels.

[0079] (2) Calculate the cluster-level soft limiting threshold T m :

[0080] In the formula, This represents the total frequency of all grayscale clusters.

[0081] (3) Perform soft-limiting scaling: If H m > T m Then, the frequency of all gray values ​​in that cluster is reduced proportionally to replace the hard removal, as shown in the following formula:

[0082] In the formula, h ′( g ) represents the frequency of grayscale values ​​after soft clipping; if H m ≤ T m ,but h ′( g )= h ( g ), retaining the original frequency.

[0083] 3. Non-uniform redistribution within clusters The redundant frequency after soft clipping ( H m T m The grayscale variance within a cluster is non-uniformly distributed, with high variance and more distribution in defective areas and low variance and less distribution in smooth areas, thus accurately improving the contrast of defect details.

[0084] (1) Calculate the first m Each gray value within a gray cluster g Local variance :

[0085] In the formula, Ω g The grayscale value in the image g The 3×3 neighborhood of the pixel, N g,3×3 The number of neighboring pixels. The average gray level of the neighborhood.

[0086] Calculate variance weights: for Normalize to [0, 1] to obtain the weights. w g The formula is as follows:

[0087] (3) Non-uniform distribution of redundant frequencies: The redundant frequencies Δ H m = H m T m The weighted distribution of grayscale values ​​within a cluster is calculated using the following formula:

[0088] In the formula, h ′′( g () represents the frequency after the final redistribution; c best The optimal variance weighting coefficients are... c When γ=0, the distribution is uniform; when γ=1, the distribution is completely based on variance, thus meeting the needs for enhancing defect details.

[0089] Step 4: Intra-cluster integration with global two-layer CDF mapping This step will adjust the histogram frequency count after step three. h ′′( g The enhancement grayscale value is mapped to the final enhanced grayscale value through a two-layer CDF (cumulative distribution function) of "local stretching within the cluster + global normalization". The enhanced grayscale value is normalized to [0, 255] to avoid the block effect of traditional CLAHE.

[0090] 1. Local stretching of CDF within the cluster For each grayscale cluster C m CDF is calculated separately to achieve local stretching of gray values ​​within a cluster, in order to adapt to the enhancement requirements of different gray-level clusters:

[0091] In the formula, For the first m Intra-cluster CDF of the cluster; For the first m grayscale clusters C m Any value within that range that is less than or equal to the current target grayscale value g The grayscale level.

[0092] Map the gray values ​​within a cluster to the upper and lower bounds of the gray values ​​of that cluster. , ], , The first m The minimum and maximum gray values ​​of a cluster, stretching low-frequency gray values ​​and compressing high-frequency gray values ​​within the cluster, improve intra-cluster contrast. The formula is as follows:

[0093] 2. Global CDF Normalization The grayscale values ​​after stretching all clusters Integration is achieved by mapping global CDF to the standard grayscale range of [0, 255] to ensure global grayscale consistency. The formula is as follows:

[0094]

[0095] In the formula, For global CDF; To obtain the final enhanced grayscale value, the grayscale value of all pixels is replaced with... . 3. Edge smoothing correction Slight smoothing is applied to the grayscale transitions between clusters to avoid abrupt grayscale changes. The strategy involves applying a 3×3 mean filter to the grayscale overlap areas of adjacent clusters to ensure a smooth grayscale transition for continuous structures such as wires and fittings.

[0096] Step 5: Enhance Image Output This step, based on the mapping results from step four, outputs an enhanced image adapted to the power transmission line inspection scenario: (1) Grayscale image: Direct output final The corresponding grayscale matrix is ​​in PNG / JPG format, which conforms to the storage specifications of the inspection system. (2) Color image: The enhanced luminance channel (Y channel) is fused with the U / V channel of the original color image to restore the RGB color image and avoid color distortion.

[0097] To verify the improvement effect of the proposed scheme "NDO iterative parameter optimization + gray-scale cluster soft constraint CLAHE" on the image enhancement quality of transmission lines, a real-world dataset of transmission lines (containing 200 images of 5 types of complex lighting conditions such as backlight, strong light, and cloudy / rainy days, and 6 types of typical defects such as conductor strand breakage, hardware corrosion, and insulator cracks) was selected. The traditional CLAHE algorithm and the CLAHE algorithm optimized by genetic algorithm (GA-CLAHE) were compared, and the superiority of the technology was verified through core quantitative indicators.

[0098] 1. Evaluation Indicators and Calculation Methods (1) Information entropy ( ): This measures the richness of image information. A higher value indicates that defects, textures, and wire details are more fully preserved after enhancement. It is a core quantitative indicator for image quality enhancement, and the calculation method is as follows:

[0099] In the formula, p ( g (This refers to the grayscale values ​​of the enhanced image.) g The probability of its occurrence reflects the image's ability to carry details.

[0100] (2) Smoothness of grayscale transition between clusters ( SIT This quantifies the unique innovative effect of this solution, measuring the smoothness of grayscale transitions in continuous structures such as transmission line conductors and fittings. A higher value indicates fewer grayscale jumps, solving the structural distortion problem caused by traditional CLAHE hard clipping. The calculation method is as follows:

[0101] 2. The test results are shown in Table 1 below.

[0102] Table 1 3. Technical Effect Analysis As can be seen from the test results in the table above, the proposed solution significantly improves the CLAHE algorithm, achieving comprehensive superiority in both core image enhancement quality indicators and proprietary innovative indicators. The information entropy (Entropy) reaches 7.96, an improvement of 1.71 (27.36%) compared to the traditional CLAHE algorithm and 0.84 (11.80%) compared to the GA-CLAHE algorithm. By precisely optimizing the core parameters K, α, γ, and T using the NDO algorithm and combining it with a non-uniform frequency distribution strategy within gray-level clusters, key information such as transmission line defect textures and conductor details is maximized. This solves the detail loss problem caused by hard cropping in the traditional CLAHE algorithm, significantly improving the image information carrying capacity. The inter-cluster gray-level transition smoothness (SIT) reaches 98.53%, an improvement of 35.85 percentage points compared to the traditional CLAHE algorithm and 15.36 percentage points compared to the GA-CLAHE algorithm. The innovative gray-level cluster soft constraint mechanism and dual-layer CDF mapping strategy completely solve the gray-level jump problem caused by hard cropping in the traditional CLAHE algorithm, ensuring smooth gray-level transitions for continuous structures such as transmission line conductors and fittings, and avoiding structural integrity damage problems such as "conductor segmentation and fitting distortion" in the enhanced image.

[0103] In summary, this scheme, through the synergistic improvement of NDO adaptive parameter optimization and gray-level cluster soft constraints, not only significantly improves the richness of image information but also ensures the smoothness of gray-level transitions in continuous structures, comprehensively optimizing the image enhancement quality of transmission lines. It is significantly superior to the traditional CLAHE and GA-CLAHE algorithms and fully meets the main requirements of intelligent inspection of transmission lines for high-quality images.

[0104] It should be noted that the sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0105] The terms "comprising" and "having," and any variations thereof, in the specification, claims, and accompanying drawings of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus. The terms "first," "second," and "third," etc., are used to distinguish different objects, etc., and do not indicate a sequence, nor do they limit "first," "second," and "third" to different types.

[0106] In the description of the embodiments of this application, terms such as "exemplary," "for example," or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a concrete manner.

[0107] In the description of the embodiments of this application, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The "and / or" in the text is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of this application, "multiple" means two or more.

[0108] In some processes described in the embodiments of this application, multiple operations or steps are included in a specific order. However, it should be understood that these operations or steps may not be executed in the order they appear in the embodiments of this application, or they may be executed in parallel. The sequence number of the operation is only used to distinguish different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed sequentially or in parallel, and these operations or steps may be combined.

[0109] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device to execute the methods described in the various embodiments of this application.

[0110] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method for enhancing images of power transmission lines, characterized in that, The transmission line image enhancement method includes: Extract the effective grayscale range of the grayscale image to be enhanced, and perform grayscale clustering on the grayscale levels within the effective grayscale range to obtain multiple continuous grayscale clusters; Perform cluster-level soft clipping on the total grayscale frequency of each of the continuous grayscale clusters to obtain clipped grayscale frequency data. The cluster-level soft clipping is used to replace the single grayscale hard clipping operation. The redundant frequencies generated by the cluster-level soft limiting process are redistributed within the corresponding continuous gray-level clusters to obtain an adjusted gray-level histogram. Based on the adjusted grayscale histogram, grayscale mapping processing is performed to obtain an enhanced transmission line image. The gray-level clustering within the effective gray-level range yields multiple consecutive gray-level clusters, including: Based on gray-level similarity, unsupervised clustering is performed on the gray levels within the effective gray-level range, dividing continuous or similar gray-level values ​​into multiple continuous gray-level clusters, which correspond to different gray-level feature regions of the transmission line image.

2. The transmission line image enhancement method as described in claim 1, characterized in that, The process of performing cluster-level soft clipping on the total gray-level frequency of each of the continuous gray-level clusters to obtain clipped gray-level frequency data includes: Calculate the total gray frequency corresponding to each consecutive gray cluster, where the total gray frequency is the sum of the histogram frequencies of all gray levels within the cluster; Determine the cluster-level limiting threshold and compare the total gray frequency of each consecutive gray cluster with the cluster-level limiting threshold; For continuous gray-level clusters whose total gray-level frequency exceeds the cluster-level limiting threshold, the histogram frequency of all gray-levels within the cluster is synchronously scaled proportionally to complete the soft limiting process. For consecutive grayscale clusters whose total grayscale frequency does not exceed the cluster-level amplitude limit threshold, their original grayscale frequency is retained.

3. The transmission line image enhancement method as described in claim 1, characterized in that, The redundant frequencies generated by the cluster-level soft clipping process are redistributed within the corresponding consecutive gray-level clusters to obtain an adjusted gray-level histogram, including: Calculate the redundancy frequency generated by consecutive grayscale clusters after cluster-level soft limiting processing; Obtain the texture detail features of the image region corresponding to each gray level within the continuous gray cluster, and determine the frequency allocation weight corresponding to each gray level based on the texture detail features; Based on the frequency allocation weight, the redundant frequencies are allocated to the corresponding gray levels within the continuous gray cluster, completing the frequency redistribution and obtaining the adjusted gray histogram.

4. The transmission line image enhancement method as described in claim 1, characterized in that, The step of performing grayscale mapping processing based on the adjusted grayscale histogram to obtain the enhanced transmission line image includes: Based on the adjusted grayscale histogram, local grayscale stretching mapping within each continuous grayscale cluster is performed to obtain the preliminary mapped grayscale value corresponding to the grayscale within each cluster. Perform global gray-level normalization mapping on the initial mapped gray values ​​of all continuous gray-level clusters to map the gray values ​​to the standard gray-level range; Smoothing is performed on the gray-level transition regions of adjacent continuous gray-level clusters to obtain the final enhanced gray-level data. An enhanced transmission line image is generated and output based on the enhanced gray-level data.

5. The transmission line image enhancement method as described in claim 1, characterized in that, The extraction of the effective grayscale range of the grayscale image to be enhanced includes: Statistically analyze the gray-level histogram distribution of the gray-level image to be enhanced; Based on the grayscale histogram distribution, extreme grayscale values ​​at both ends of the histogram are removed, and the grayscale range including the core components and defect areas of the transmission line is retained to form the effective grayscale range.

6. The transmission line image enhancement method as described in claim 1, characterized in that, Before extracting the effective grayscale range of the grayscale image to be enhanced, the process further includes: Acquire images of power transmission line inspections to be processed; The inspection image of the transmission line is preprocessed to obtain the grayscale image to be enhanced. The preprocessing includes edge-preserving denoising, which is used to filter out image acquisition noise and environmental interference noise while retaining the edge contour information of the core components of the transmission line.

7. The transmission line image enhancement method as described in claim 6, characterized in that, The preprocessing of the transmission line inspection image to obtain the grayscale image to be enhanced includes: When the transmission line inspection image is a color image, the grayscale data corresponding to its brightness channel is extracted as the data to be processed. When the transmission line inspection image is a grayscale image, it is directly used as the data to be processed; The edge-preserving denoising process is performed on the data to be processed to obtain the grayscale image to be enhanced.

8. The transmission line image enhancement method according to any one of claims 1 to 7, characterized in that, The method further includes: Before performing image enhancement processing, the core control parameters of the image enhancement process are adaptively optimized using the Newton's Downhill Iterative Optimization Algorithm to obtain the optimal parameter combination that is suitable for the current transmission line inspection image. Based on the optimal parameter combination, the corresponding gray-level clustering, cluster-level soft limiting, frequency redistribution, and effective gray-level interval extraction steps are executed.

9. The transmission line image enhancement method as described in claim 8, characterized in that, The process of adaptively optimizing the core control parameters of the image enhancement process using the Newton's Downhill Iterative Optimization Algorithm includes: The value range of the core control parameters to be optimized is initialized to generate an initial optimization population; the core control parameters include the number of gray-level clusters, the limiting coefficient of cluster-level soft limiting, the weight coefficient of frequency redistribution, and the extreme gray-level removal ratio of the effective gray-level range. Construct a fitness function that integrates multi-dimensional image enhancement effect evaluation metrics; With the goal of finding the optimal value of the fitness function, Newton's Downhill Iterative Optimization is performed to complete the iterative update of the population and the selection of optimal solutions. After the iteration terminates, the globally optimal parameter combination is output.

Images