A power transmission line image enhancement method and system based on super-resolution reconstruction

Abstract

The application relates to the technical field of image processing, and particularly discloses a power transmission line image enhancement method and system based on super-resolution reconstruction, which comprises the following steps: firstly, carrying out gray scale statistics on a power transmission line original inspection image to obtain global brightness distribution characteristics; secondly, carrying out contrast layering on the original image according to the characteristics to obtain a brightness level image set; thirdly, constructing a scale space and a multi-scale gradient field to generate a significant geometric structure feature map; fourthly, splitting a low-frequency basic layer and a high-frequency detail layer through guided filtering, combining the feature map to complete structure perception reconstruction, and obtaining a hierarchical super-resolution image; fifthly, obtaining an original brightness level mapping relationship through brightness field backtracking, obtaining a composite super-resolution image through Bayesian inference; and finally, obtaining a final enhanced image through visual optimization; and the application can improve the quality and precision of the power transmission line image.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method and system for enhancing transmission line images based on super-resolution reconstruction. Background Technology

[0002] Existing image enhancement technologies for power transmission line inspections only employ global brightness adjustment and uniform filtering, failing to implement differentiated layering operations based on the image's own brightness distribution. This results in an unbalanced overall brightness distribution, excessive smoothing or masking of the detailed features of the core geometric structure of the transmission line, and an inability to fully preserve the texture and contour information of the line itself. Furthermore, existing technologies lack targeted processing in scale-space construction and gradient field extraction, making it difficult to accurately locate the linear structural features of the transmission line. This leads to blurred image edges, loss of detail, and structural distortion and artifacts during super-resolution reconstruction, ultimately failing to restore the true visual state of the inspection image.

[0003] Existing image enhancement schemes lack a brightness correlation mechanism between hierarchical super-resolution images and the original images. The lack of brightness hierarchy constraints during reconstruction leads to brightness shifts and pixel anomalies in the output image, failing to maintain the brightness distribution logic of the original image. Furthermore, current visual optimization techniques employ only a single global correction method, failing to simultaneously improve global contrast and local detail. Edge sharpening and brightness correction lack precise benchmarks, resulting in low clarity, integrity, and recognizability of the enhanced image. This makes it difficult to support refined inspection and accurate fault identification of transmission lines, reducing the reliability of line condition detection. Therefore, improving the quality and accuracy of transmission line images has become an urgent problem to be solved. Summary of the Invention

[0004] This invention provides a method and system for enhancing transmission line images based on super-resolution reconstruction, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a method for enhancing transmission line images based on super-resolution reconstruction, comprising: C1. Perform grayscale statistics on the original images of the original inspection image sequence of the transmission line to obtain the global brightness distribution characteristics of the original images; C2. Based on the global brightness distribution features, the original image is contrast-layered to obtain the brightness layer image set of the transmission line; C3. Scale space construction is performed on the brightness level images in the brightness level image set to obtain the multi-scale gradient field of the brightness level images, and the significant geometric structure feature map of the transmission line is constructed based on the multi-scale gradient field. C4. Perform guided filtering on the brightness level image to obtain the low-frequency base layer and high-frequency detail layer of the brightness level image, and perform structure-aware reconstruction on the low-frequency base layer and the high-frequency detail layer based on the significant geometric structure feature map to obtain the level super-resolution image of the brightness level image. C5. Perform luminance field backtracking on the hierarchical super-resolution image and the original image to obtain the original luminance hierarchy mapping relationship of the hierarchical super-resolution image, and perform Bayesian inference on the hierarchical super-resolution image based on the original luminance hierarchy mapping relationship to obtain the composite super-resolution image of the transmission line. C6. Perform visual optimization on the composite super-resolution image to obtain the final enhanced image of the transmission line.

[0006] In a preferred embodiment, the grayscale statistics of the original images in the original inspection image sequence of the transmission line to obtain the global brightness distribution features of the original images include: Obtain the original inspection image sequence of the transmission line; The pixels of the original image in the original inspection image sequence are converted to brightness to obtain the gray value of the pixel, and the frequency of occurrence of the gray value is accumulated to obtain the gray frequency distribution record of the original image. The frequencies of adjacent gray values ​​in the gray-level frequency distribution record are merged to obtain the gray-level interval and gray-level interval aggregate frequency of the transmission line. The distribution of the gray-level interval and gray-level interval aggregate frequency is constructed to obtain the gray-level interval aggregate distribution of the original image. Based on the gray-level interval aggregation frequency, peak location is performed on the gray-level interval aggregation distribution to obtain the maximum aggregation frequency interval of the gray-level interval aggregation distribution, and the gray-level lower limit and gray-level upper limit of the maximum aggregation frequency interval are used as the main brightness interval of the original image; The main brightness range and the grayscale range are aggregated and mapped collaboratively to obtain the global brightness distribution features of the original image.

[0007] In a preferred embodiment, the step of contrast-layering the original image based on the global brightness distribution features to obtain a brightness-level image set of the transmission line includes: Based on the main brightness interval of the global brightness distribution features, the pixel points of the original image are classified to obtain the main brightness level of the pixel points; The main brightness level pixels are extracted by masking to obtain the main brightness level image of the power transmission line; Within the non-primary brightness range of the global brightness distribution features, grayscale values ​​in the original image are clustered to obtain high-proportion secondary brightness level images, medium-proportion secondary brightness level images, and low-proportion secondary brightness level images of the transmission line. The main brightness level image, the high-proportion secondary brightness level image, the medium-proportion secondary brightness level image, and the low-proportion secondary brightness level image are jointly filtered to obtain the brightness level image set of the transmission line.

[0008] In a preferred embodiment, the step of constructing a scale space for the luminance level images in the luminance level image set to obtain a multi-scale gradient field of the luminance level images, and constructing a significant geometric feature map of the transmission line based on the multi-scale gradient field, includes: Gaussian smoothing kernel convolution is performed on the brightness level images in the brightness level image set to obtain the scale space image sequence of the brightness level images; Perform a difference transform on the scale space images in the scale space image sequence to obtain the horizontal gray-scale change rate and the vertical gray-scale change rate of the scale space images; The horizontal gray-level change rate and the vertical gray-level change rate are vector-synthesized to obtain the single-scale gradient field of the scale space image; Based on a preset scale order, the single-scale gradient fields are stacked sequentially to obtain the multi-scale gradient field of the brightness level image. Extremum detection is performed on the multi-scale gradient field to obtain local maxima points of the multi-scale gradient field, and point set fitting is performed on the local maxima points to obtain candidate structure line segments of the multi-scale gradient field. Spatial projection is performed on the candidate structural line segments to obtain a significant geometric structural feature map of the transmission line.

[0009] In a preferred embodiment, the step of performing guided filtering on the luminance level image to obtain a low-frequency base layer and a high-frequency detail layer of the luminance level image, and then performing structure-aware reconstruction on the low-frequency base layer and the high-frequency detail layer based on the salient geometric structure feature map to obtain a hierarchical super-resolution image of the luminance level image, includes: Using the brightness level image as a guide image, the brightness level image is truncated in the frequency domain to obtain the low-frequency base layer of the brightness level image; The residual between the brightness level image and the low-frequency base layer is extracted to obtain the high-frequency detail layer of the brightness level image, and bilinear interpolation is performed on the low-frequency base layer to obtain the magnified low-frequency layer of the low-frequency base layer. Angle marking is performed on the salient geometric structure feature map to obtain the geometric direction label of the salient geometric structure feature map; Based on the geometric direction label, anisotropic diffusion is performed on the high-frequency detail layer to obtain the directional enhancement detail layer of the brightness level image; Based on the amplified low-frequency layer, the directional enhancement detail layer is amplified proportionally to obtain the amplified high-frequency layer of the directional enhancement detail layer. The amplified low-frequency layer and the amplified high-frequency layer are cascaded and reconstructed to obtain the hierarchical super-resolution image of the brightness level image.

[0010] In a preferred embodiment, the step of performing luminance field backtracking on the hierarchical super-resolution image and the original image to obtain the original luminance hierarchy mapping relationship of the hierarchical super-resolution image, and performing Bayesian inference on the hierarchical super-resolution image based on the original luminance hierarchy mapping relationship to obtain the composite super-resolution image of the transmission line, includes: The pixels in the hierarchical super-resolution image are reverse-mapped back to the pixel coordinate space of the original image to obtain the original image coordinate position of the hierarchical super-resolution image. Based on the original image coordinates, pixel attribution is determined for the brightness level image set to obtain the brightness level label of the original image coordinates. The brightness level label is then associated and bound with the pixels in the super-resolution image to obtain the original brightness level mapping relationship of the super-resolution image. Based on the original brightness level mapping relationship, the pixel density in the super-resolution image is estimated to obtain the prior distribution features of the pixel. Anomaly detection is performed on the pixels in the hierarchical super-resolution image and the prior distribution features to obtain the outlier pixels in the hierarchical super-resolution image; Based on the prior distribution characteristics, prior constraint optimization is performed on the outlier pixels to obtain the posterior estimate of the outlier pixels. The posterior estimates are used to drive reconstruction to obtain a composite super-resolution image of the transmission line.

[0011] In a preferred embodiment, the posterior estimate is calculated using the following formula: ; In the formula, The index of the pixel in the super-resolution image is given. For the hierarchical super-resolution image, the first The posterior estimate of each outlier pixel. For the hierarchical super-resolution image, the first The original observations of the outlier pixels, The first in the original brightness level mapping relationship Brightness level labels for each pixel For the brightness level label The corresponding prior distribution mean, For the brightness level label The corresponding prior distribution variance, This is the preset observation noise variance.

[0012] In a preferred embodiment, the step of visually optimizing the composite super-resolution image to obtain the final enhanced image of the transmission line includes: Global histogram statistics are performed on the composite super-resolution image to obtain the pixel value distribution range of the composite super-resolution image, and the pixel values ​​of the composite super-resolution image are linearly mapped according to the pixel value distribution range to obtain the global contrast-enhanced image of the transmission line. The global contrast enhancement image is divided into blocks to obtain local regions of the global contrast enhancement image, and local cumulative distribution fitting is performed on the local regions to obtain the enhancement mapping relationship of each local region. Based on the enhanced mapping relationship, the value range of the pixels in the local region is transformed to obtain the local contrast adjusted image of the transmission line; High-pass filtering is applied to the local contrast-adjusted image to obtain the edge-sharpened image of the local contrast-adjusted image; Gamma correction is applied to the edge-sharpened image to obtain the final enhanced image of the transmission line.

[0013] In a preferred embodiment, performing gamma correction on the edge-sharpened image to obtain the final enhanced image of the transmission line includes: The center trend of the pixels in the edge-sharpened image is measured to obtain the overall brightness benchmark of the edge-sharpened image; Gain derivation is performed on the preset target luminance median and the overall luminance reference to obtain the correction intensity coefficient of the transmission line; Based on the overall brightness reference and the correction intensity coefficient, the pixels of the edge-sharpened image are grayscale corrected to obtain the corrected pixel values ​​of the edge-sharpened image; The corrected pixel values ​​are then used for image reconstruction to obtain the final enhanced image of the transmission line.

[0014] To address the aforementioned problems, the present invention also provides a transmission line image enhancement system based on super-resolution reconstruction, the system comprising: The brightness statistics module is used to perform grayscale statistics on the original images of the original inspection image sequence of the transmission line to obtain the global brightness distribution characteristics of the original images. The layering processing module is used to perform contrast layering on the original image based on the global brightness distribution features to obtain a brightness layer image set of the transmission line. The scale space module is used to construct the scale space of the brightness level images in the brightness level image set, obtain the multi-scale gradient field of the brightness level images, and construct the significant geometric structure feature map of the transmission line based on the multi-scale gradient field. The structure reconstruction module is used to perform guided filtering on the brightness level image to obtain the low-frequency base layer and high-frequency detail layer of the brightness level image, and to perform structure-aware reconstruction on the low-frequency base layer and the high-frequency detail layer based on the significant geometric structure feature map to obtain the level super-resolution image of the brightness level image. The backtracking inference module is used to perform brightness field backtracking on the hierarchical super-resolution image and the original image to obtain the original brightness hierarchy mapping relationship of the hierarchical super-resolution image, and to perform Bayesian inference on the hierarchical super-resolution image based on the original brightness hierarchy mapping relationship to obtain the composite super-resolution image of the transmission line. The visual optimization module is used to perform visual optimization on the composite super-resolution image to obtain the final enhanced image of the transmission line.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention can accurately extract the global brightness distribution features of the original inspection images of transmission lines, complete the image contrast layering processing, construct multi-scale gradient fields and significant geometric structure feature maps, realize the structural perception reconstruction of the low-frequency basic layer and the high-frequency detail layer, completely preserve the line texture, contour and linear geometric structure, improve the detail integrity and structural restoration accuracy of the layered super-resolution image, and ensure the stability of image super-resolution reconstruction.

[0016] 2. This invention establishes the original brightness level mapping relationship through brightness field backtracking, completes pixel feature optimization based on brightness level constraints, eliminates image brightness offset and pixel anomalies, maintains the original brightness distribution logic of the image, and then improves the image contrast and edge sharpening effect simultaneously through global and local collaborative visual optimization, thereby improving the clarity and recognizability of the final enhanced image, providing high-quality image support for the refined inspection of power transmission lines, and enhancing the overall effect and application reliability of image enhancement. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a method for enhancing transmission line images based on super-resolution reconstruction, provided in an embodiment of the present invention. Figure 2 A functional block diagram of a power transmission line image enhancement system based on super-resolution reconstruction is provided in an embodiment of the present invention. The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] This application provides a method for enhancing transmission line images based on super-resolution reconstruction. The execution entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method for enhancing transmission line images based on super-resolution reconstruction can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cluster of cloud servers. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0020] Reference Figure 1 The diagram shown is a flowchart illustrating a method for enhancing transmission line images based on super-resolution reconstruction according to an embodiment of the present invention. In this embodiment, the method for enhancing transmission line images based on super-resolution reconstruction includes: C1. Perform grayscale statistics on the original images of the original inspection image sequence of the transmission line to obtain the global brightness distribution characteristics of the original images; In this embodiment of the invention, the grayscale statistics of the original images in the original inspection image sequence of the transmission line to obtain the global brightness distribution features of the original images include: Obtain the original inspection image sequence of the transmission line; The pixels of the original image in the original inspection image sequence are converted to brightness to obtain the gray value of the pixel, and the frequency of occurrence of the gray value is accumulated to obtain the gray frequency distribution record of the original image. The frequencies of adjacent gray values ​​in the gray-level frequency distribution record are merged to obtain the gray-level interval and gray-level interval aggregate frequency of the transmission line. The distribution of the gray-level interval and gray-level interval aggregate frequency is constructed to obtain the gray-level interval aggregate distribution of the original image. Based on the gray-level interval aggregation frequency, peak location is performed on the gray-level interval aggregation distribution to obtain the maximum aggregation frequency interval of the gray-level interval aggregation distribution, and the gray-level lower limit and gray-level upper limit of the maximum aggregation frequency interval are used as the main brightness interval of the original image; The main brightness range and the grayscale range are aggregated and mapped collaboratively to obtain the global brightness distribution features of the original image.

[0021] The original inspection image sequence of the transmission line is directly retrieved from the transmission line inspection and acquisition equipment. This sequence contains multiple unprocessed original images used for line detection.

[0022] Read all pixels in each original image in the original inspection image sequence one by one, convert the color brightness information carried by each pixel into an independent gray value, after traversing all pixels of an original image, accumulate the number of occurrences of each gray value in the image one by one, and record each gray value and the corresponding accumulated frequency to form a gray frequency distribution record of the original image.

[0023] Following the continuous arrangement of gray values ​​from smallest to largest, the frequencies corresponding to adjacent gray values ​​in the gray-level frequency distribution record are added and merged to form continuous and uninterrupted gray-level intervals. The sum of these values ​​is used as the aggregated frequency of the corresponding gray-level interval. All gray-level intervals and their corresponding aggregated frequencies are combined and arranged in order of gray values ​​to construct the aggregated distribution of the gray-level intervals of the original image.

[0024] The gray-level interval aggregation frequency corresponding to each gray-level interval in the gray-level interval aggregation distribution is read sequentially. The gray-level interval with the largest gray-level interval aggregation frequency value is determined as the maximum aggregation frequency interval of the gray-level interval aggregation distribution. The minimum gray-level value corresponding to the maximum aggregation frequency interval is extracted as the lower gray-level value, and the maximum gray-level value corresponding to the maximum aggregation frequency interval is extracted as the upper gray-level value. The lower gray-level value and the upper gray-level value are combined to form the main brightness interval of the original image.

[0025] The grayscale range corresponding to the main brightness range is matched bidirectionally in terms of position and value with all grayscale ranges and their corresponding grayscale range aggregation frequencies in the grayscale range aggregation distribution. The core brightness information of the main brightness range is integrated into the overall structure of the grayscale range aggregation distribution to form a global brightness distribution feature that can completely characterize the overall brightness distribution of the original image.

[0026] The beneficial effect is that by processing the pixel brightness information of the original inspection images of transmission lines layer by layer, the global brightness distribution features of the images can be accurately extracted, the main brightness range of the images can be clearly determined, and the overall brightness distribution pattern of the images can be completely preserved, providing accurate and stable brightness feature support for the subsequent analysis and processing of transmission line inspection images.

[0027] C2. Based on the global brightness distribution features, the original image is contrast-layered to obtain the brightness layer image set of the transmission line; In this embodiment of the invention, the step of performing contrast layering on the original image based on the global brightness distribution features to obtain the brightness layer image set of the transmission line includes: Based on the main brightness interval of the global brightness distribution features, the pixel points of the original image are classified to obtain the main brightness level of the pixel points; The main brightness level pixels are extracted by masking to obtain the main brightness level image of the power transmission line; Within the non-primary brightness range of the global brightness distribution features, grayscale values ​​in the original image are clustered to obtain high-proportion secondary brightness level images, medium-proportion secondary brightness level images, and low-proportion secondary brightness level images of the transmission line. The main brightness level image, the high-proportion secondary brightness level image, the medium-proportion secondary brightness level image, and the low-proportion secondary brightness level image are jointly filtered to obtain the brightness level image set of the transmission line.

[0028] The grayscale value of each pixel in the original image is compared one by one with the lower and upper grayscale limits of the main brightness range corresponding to the global brightness distribution features. Pixels with grayscale values ​​greater than or equal to the lower limit and less than or equal to the upper limit are classified as main brightness level pixels, while pixels with grayscale values ​​less than the lower limit or greater than the upper limit are classified as non-main brightness level pixels. This process completes the classification of all pixels and obtains the main brightness level of each pixel. Based on the pixel location information corresponding to the main brightness level, the original grayscale values ​​and spatial positions of the main brightness level pixels are completely preserved in the original image. The grayscale values ​​of non-main brightness level pixels are uniformly replaced with fixed background grayscale values, enabling independent extraction of the main brightness level region and obtaining the main brightness level image of the transmission line. After excluding the main brightness interval in the global brightness distribution features, the remaining gray range is taken as the non-main brightness interval. All pixels in the original image whose gray values ​​are in the non-main brightness interval are traversed. These pixels are divided into three consecutive gray groups in descending order of gray value. The three gray groups correspond to the high proportion sub-brightness level, the medium proportion sub-brightness level, and the low proportion sub-brightness level, respectively. The gray value information and position information of the pixels in each group are retained and the pixels in other groups are removed to obtain the high proportion sub-brightness level image, the medium proportion sub-brightness level image, and the low proportion sub-brightness level image of the transmission line. Isolated outlier pixels were removed from the main brightness level image, high-proportion secondary brightness level image, medium-proportion secondary brightness level image, and low-proportion secondary brightness level image. Each pixel was checked for adjacent pixels at the same level; outlier pixels without adjacent pixels were removed. All valid pixels with adjacent pixels were retained. After filtering the four images, they were integrated into a unified image set, resulting in a brightness level image set for the transmission line. The beneficial effects are: accurate brightness stratification of the original image based on global brightness distribution features; clear distinction between the main brightness level and secondary brightness levels with different proportions; preservation of complete features of each level image through mask extraction and targeted filtering; and formation of a structurally sound and feature-clear brightness level image set. This provides stable and accurate layered image support for subsequent analysis and processing of transmission line inspection images.

[0029] C3. Scale space construction is performed on the brightness level images in the brightness level image set to obtain the multi-scale gradient field of the brightness level images, and the significant geometric structure feature map of the transmission line is constructed based on the multi-scale gradient field. In this embodiment of the invention, the step of constructing a scale space for the luminance level images in the luminance level image set to obtain a multi-scale gradient field of the luminance level images, and constructing a significant geometric structure feature map of the transmission line based on the multi-scale gradient field, includes: Gaussian smoothing kernel convolution is performed on the brightness level images in the brightness level image set to obtain the scale space image sequence of the brightness level images; Perform a difference transform on the scale space images in the scale space image sequence to obtain the horizontal gray-scale change rate and the vertical gray-scale change rate of the scale space images; The horizontal gray-level change rate and the vertical gray-level change rate are vector-synthesized to obtain the single-scale gradient field of the scale space image; Based on a preset scale order, the single-scale gradient fields are stacked sequentially to obtain the multi-scale gradient field of the brightness level image. Extremum detection is performed on the multi-scale gradient field to obtain local maxima points of the multi-scale gradient field, and point set fitting is performed on the local maxima points to obtain candidate structure line segments of the multi-scale gradient field. Spatial projection is performed on the candidate structural line segments to obtain a significant geometric structural feature map of the transmission line.

[0030] Each brightness level image in the brightness level image set is selected. A Gaussian smoothing kernel with a fixed side length of nine pixels is used as the basic processing unit. Starting from the upper left corner of the brightness level image, the Gaussian smoothing kernel is moved row by row and column by column to completely cover the entire pixel area of ​​the image. The gray values ​​of all pixels within the coverage area of ​​the Gaussian smoothing kernel are read. These gray values ​​are weighted and fused according to the preset weight rule that the center position has the highest proportion and the edge position has the lowest proportion. The unique value obtained by fusion replaces the original gray value of the pixel at the center position of the Gaussian smoothing kernel. After the full pixel smoothing processing of a single brightness level image is completed, all processing results are arranged in order according to the processing scale from small to large to obtain the scale space image sequence of brightness level images.

[0031] Read all pixel information of each scale-space image within the scale-space image sequence. Starting from the first pixel in the first row and first column of the image, compare the gray values ​​of two adjacent pixels in the same row horizontally. Subtract the gray value of the previous pixel from the gray value of the next pixel, and use the calculated value as the horizontal gray-level change rate of that adjacent position. Starting from the first pixel in the first column and first row of the image, compare the gray values ​​of two adjacent pixels in the same column vertically. Subtract the gray value of the previous pixel from the gray value of the next pixel, and use the calculated value as the vertical gray-level change rate of that adjacent position. Completely record the horizontal and vertical gray-level change rates corresponding to each pixel in the scale-space image.

[0032] The horizontal and vertical grayscale change rates corresponding to each pixel position in the scale space image are synchronously integrated, and the horizontal spatial direction and magnitude of the horizontal grayscale change rate and the vertical spatial direction and magnitude of the vertical grayscale change rate are fully preserved. The change rates in the two directions are integrated into a single gradient description information that can simultaneously represent the position and the magnitude of change. After the gradient description information covers all pixels of the scale space image, a single-scale gradient field exclusive to the scale space image is formed.

[0033] Following a fixed arrangement of scale values ​​from smallest to largest, all single-scale gradient fields obtained at different processing scales are stacked in a one-to-one correspondence of pixel coordinates, strictly maintaining the spatial positional relationship of pixels within each single-scale gradient field without shifting or misaligning. After completing the stacking and combination of all single-scale gradient fields, the multi-scale gradient field corresponding to the brightness level image is obtained.

[0034] The gradient description information of each pixel in the multi-scale gradient field is traversed. Taking the current pixel to be detected as the center, eight neighboring pixels are selected as a fixed comparison range. The gradient description information of the current pixel is compared with the gradient description information of all neighboring pixels in the comparison range one by one. When the gradient description information value of the current pixel is greater than the gradient description information values ​​of all neighboring pixels in the comparison range, the pixel is marked as a local maximum point. All marked local maximum points in the multi-scale gradient field are collected. The adjacent points are connected in sequence according to the spatial coordinate position of the local maximum points to form a continuous and complete line structure, thus obtaining the candidate structure line segment corresponding to the multi-scale gradient field.

[0035] Candidate structural line segments are mapped to the spatial coordinate system of the original inspection image of the transmission line according to the pixel coordinate correspondence. The conventional direction of the transmission line conductors and fittings is used as the matching benchmark. Line segments that are consistent with the direction of the transmission line are retained, and scattered line segments with a length less than a fixed pixel threshold and no actual geometric meaning are removed. All effective line segments are integrated into a unified feature image to obtain the significant geometric structure feature map corresponding to the transmission line.

[0036] The beneficial effects are that by constructing a scale-space image sequence through multi-scale Gaussian smoothing, and by accurately extracting gradient change information by combining horizontal and vertical differential transformations, local maxima are located after synthesizing single-scale and multi-scale gradient fields, and significant geometric structure feature maps are formed by point set fitting and coordinate projection. This fully preserves the core geometric morphological information of the transmission line, such as the conductor direction and component outline, and provides accurate and stable feature support for geometric structure recognition and fault detection of transmission line inspection images.

[0037] C4. Perform guided filtering on the brightness level image to obtain the low-frequency base layer and high-frequency detail layer of the brightness level image, and perform structure-aware reconstruction on the low-frequency base layer and the high-frequency detail layer based on the significant geometric structure feature map to obtain the level super-resolution image of the brightness level image. In this embodiment of the invention, the step of performing guided filtering on the brightness level image to obtain a low-frequency base layer and a high-frequency detail layer of the brightness level image, and then performing structure-aware reconstruction on the low-frequency base layer and the high-frequency detail layer based on the salient geometric structure feature map to obtain a hierarchical super-resolution image of the brightness level image, includes: Using the brightness level image as a guide image, the brightness level image is truncated in the frequency domain to obtain the low-frequency base layer of the brightness level image; The residual between the brightness level image and the low-frequency base layer is extracted to obtain the high-frequency detail layer of the brightness level image, and bilinear interpolation is performed on the low-frequency base layer to obtain the magnified low-frequency layer of the low-frequency base layer. Angle marking is performed on the salient geometric structure feature map to obtain the geometric direction label of the salient geometric structure feature map; Based on the geometric direction label, anisotropic diffusion is performed on the high-frequency detail layer to obtain the directional enhancement detail layer of the brightness level image; Based on the amplified low-frequency layer, the directional enhancement detail layer is amplified proportionally to obtain the amplified high-frequency layer of the directional enhancement detail layer. The amplified low-frequency layer and the amplified high-frequency layer are cascaded and reconstructed to obtain the hierarchical super-resolution image of the brightness level image.

[0038] Using the brightness level image as the guide image, a threshold is determined based on the smoothness of gray-level changes within the image. Pixel gray-level information with gray-level change rates within the threshold range is retained, while pixel gray-level information with gray-level change rates exceeding the threshold range is removed. This completes the frequency domain truncation process, resulting in the low-frequency base layer of the brightness level image.

[0039] The gray value of each pixel in the brightness level image is subtracted from the gray value of the pixel at the same coordinate position in the low-frequency base layer. The gray value difference of all coordinate positions is extracted and retained as independent pixel information to obtain the high-frequency detail layer of the brightness level image. New pixels are added between adjacent pixels in the low-frequency base layer according to the coordinate linear relationship using bilinear interpolation, so that the number of horizontal and vertical pixels in the low-frequency base layer increases by a fixed ratio, resulting in the magnified low-frequency layer after the low-frequency base layer.

[0040] The extension direction of candidate structural line segments in the significant geometric structure feature map is identified one by one. The angle between each candidate structural line segment and the horizontal baseline is measured. The angle value is used as the identification information of the corresponding line segment. After all line segments are identified, the geometric direction label of the significant geometric structure feature map is obtained.

[0041] According to the included angle and direction marked by the geometric direction label, the pixels in the high-frequency detail layer are arranged with grayscale continuity along the corresponding direction, so that the pixel grayscale change is consistent with the geometric direction, and the anisotropic diffusion process is completed to obtain the direction enhancement detail layer of the brightness level image.

[0042] Referring to the pixel size and pixel distribution ratio of the magnified low-frequency layer, the pixels of the directional enhancement detail layer are supplemented in equal numbers and arranged at equal intervals, while keeping the grayscale and directional features of the directional enhancement detail layer unchanged. This process is then performed to complete the same-proportion magnification, resulting in the magnified high-frequency layer of the directional enhancement detail layer.

[0043] The pixels at the same coordinate position on the magnified low-frequency layer and the magnified high-frequency layer are superimposed in grayscale, maintaining a one-to-one correspondence between the spatial coordinates of all pixels. After the superposition is completed, a complete image is formed, resulting in a super-resolution image of the brightness level.

[0044] The beneficial effects are that by guiding filtering to separate the high and low frequency information of the brightness level image, and combining it with the significant geometric structure feature map to achieve directional enhancement and layered magnification of the detail layer, the structure-aware image reconstruction is accurately completed, resulting in a layered super-resolution image with both high resolution and clear geometric structure, providing high-quality image support for the refined analysis of transmission line inspection images.

[0045] C5. Perform luminance field backtracking on the hierarchical super-resolution image and the original image to obtain the original luminance hierarchy mapping relationship of the hierarchical super-resolution image, and perform Bayesian inference on the hierarchical super-resolution image based on the original luminance hierarchy mapping relationship to obtain the composite super-resolution image of the transmission line. In this embodiment of the invention, the step of performing luminance field backtracking on the hierarchical super-resolution image and the original image to obtain the original luminance hierarchy mapping relationship of the hierarchical super-resolution image, and performing Bayesian inference on the hierarchical super-resolution image based on the original luminance hierarchy mapping relationship to obtain the composite super-resolution image of the transmission line, includes: The pixels in the hierarchical super-resolution image are reverse-mapped back to the pixel coordinate space of the original image to obtain the original image coordinate position of the hierarchical super-resolution image. Based on the original image coordinates, pixel attribution is determined for the brightness level image set to obtain the brightness level label of the original image coordinates. The brightness level label is then associated and bound with the pixels in the super-resolution image to obtain the original brightness level mapping relationship of the super-resolution image. Based on the original brightness level mapping relationship, the pixel density in the super-resolution image is estimated to obtain the prior distribution features of the pixel. Anomaly detection is performed on the pixels in the hierarchical super-resolution image and the prior distribution features to obtain the outlier pixels in the hierarchical super-resolution image; Based on the prior distribution characteristics, prior constraint optimization is performed on the outlier pixels to obtain the posterior estimate of the outlier pixels. The posterior estimates are used to drive reconstruction to obtain a composite super-resolution image of the transmission line.

[0046] The formula for calculating the posterior estimate is as follows: ; In the formula, The index of the pixel in the super-resolution image is given. For the hierarchical super-resolution image, the first The posterior estimate of each outlier pixel. For the hierarchical super-resolution image, the first The original observations of the outlier pixels, The first in the original brightness level mapping relationship Brightness level labels for each pixel For the brightness level label The corresponding prior distribution mean, For the brightness level label The corresponding prior distribution variance, This is the preset observation noise variance.

[0047] Based on the proportional relationship between the resolution size of the hierarchical super-resolution image and the resolution size of the original inspection image, each pixel in the hierarchical super-resolution image is traversed, and each pixel is reversed to the standard pixel coordinate system of the original image according to the coordinate scaling rules. The unique coordinate point corresponding to each pixel in the hierarchical super-resolution image in the original image is accurately located, and the original image coordinate position of the hierarchical super-resolution image is obtained.

[0048] Using the original image coordinates of the hierarchical super-resolution image as the basis for precise retrieval, the brightness level category to which the coordinates belong is matched one by one in the pre-generated brightness level image set. A unique brightness level label is marked for the original image coordinates. The brightness level label is fixedly associated with the corresponding pixel in the hierarchical super-resolution image, forming a stable correspondence between the pixels of the hierarchical super-resolution image and the brightness level of the original image, thus obtaining the original brightness level mapping relationship of the hierarchical super-resolution image.

[0049] Based on the original brightness level mapping relationship, all pixels in the hierarchical super-resolution image are grouped and classified. The numerical distribution range and concentration position of all pixel gray values ​​within the same brightness level group are statistically analyzed. The normal distribution range and central state of the pixel gray values ​​in the group are clarified, and the prior distribution characteristics of pixels in the hierarchical super-resolution image are obtained.

[0050] The grayscale value of each pixel in the hierarchical super-resolution image is compared with the prior distribution features of the corresponding group one by one. Pixels whose grayscale values ​​are outside the normal distribution range of the corresponding group are directly identified as anomalous pixels. After filtering out all anomalous pixels, the outlier pixels of the hierarchical super-resolution image are obtained.

[0051] Each pixel in the hierarchical super-resolution image is uniquely numbered according to its horizontal row-by-row and vertical column-by-column arrangement order to obtain an index for pixel location. ,index The value is directly determined by the pixel's position in the image. The formula for calculating the posterior estimate of outlier pixels in a hierarchical super-resolution image is based on the index. Based on this, the optimization of outlier pixels is completed by combining brightness levels and grayscale values.

[0052] The hierarchical super-resolution image generated by the structure-aware reconstruction process is the first The initial grayscale value carried by each outlier pixel is the original observation value. The first result obtained from the brightness field backtracking process Each pixel corresponds to a brightness level label. Brightness level labels obtained statistically based on the original brightness level mapping relationship The gray-level center value of the corresponding pixel group is the mean of the prior distribution. Statistical brightness level labels The degree of numerical dispersion, which represents the deviation of the grayscale values ​​of a corresponding pixel group from the prior distribution mean, is the variance of the prior distribution. The observed noise variance is determined by combining historical noise data from transmission line inspection images with pre-defined fixed noise fluctuation characteristics based on actual imaging experience. .

[0053] The original observed values ​​are calculated using the posterior estimate formula for outlier pixels in hierarchical super-resolution images. With brightness level labels By combining the corresponding prior distribution features, the original grayscale values ​​of outlier pixels are adjusted to fall within the normal distribution range of the corresponding group, generating standard grayscale values ​​that fully meet the distribution requirements of the brightness level, thus obtaining the posterior estimate of the outlier pixels. This formula provides the core numerical basis for the Bayesian inference process, ensuring the accuracy of the generated posterior estimates.

[0054] Using posterior estimates The original grayscale values ​​of the corresponding outlier pixels in the hierarchical super-resolution image are replaced one by one, while the original grayscale and position information of the non-outlier pixels in the hierarchical super-resolution image are completely preserved. The pixels of the entire image are recombined and shaped according to the coordinate relationship of the original image to obtain the composite super-resolution image of the transmission line.

[0055] The beneficial effects are that the original brightness level mapping relationship between the hierarchical super-resolution image and the original image is accurately constructed through a complete brightness field backtracking process. Based on the prior distribution features of pixels and the calculation formula of special posterior estimation values, the outlier pixels are accurately identified and constrained and optimized. The pixel anomaly problem in the hierarchical super-resolution image is effectively corrected, and the stability and consistency of the overall brightness distribution of the image are strictly maintained. Finally, a composite super-resolution image of the transmission line with complete details, balanced brightness, and accurate structure is generated.

[0056] C6. Perform visual optimization on the composite super-resolution image to obtain the final enhanced image of the transmission line.

[0057] In this embodiment of the invention, the step of visually optimizing the composite super-resolution image to obtain the final enhanced image of the transmission line includes: Global histogram statistics are performed on the composite super-resolution image to obtain the pixel value distribution range of the composite super-resolution image, and the pixel values ​​of the composite super-resolution image are linearly mapped according to the pixel value distribution range to obtain the global contrast-enhanced image of the transmission line. The global contrast enhancement image is divided into blocks to obtain local regions of the global contrast enhancement image, and local cumulative distribution fitting is performed on the local regions to obtain the enhancement mapping relationship of each local region. Based on the enhanced mapping relationship, the value range of the pixels in the local region is transformed to obtain the local contrast adjusted image of the transmission line; High-pass filtering is applied to the local contrast-adjusted image to obtain the edge-sharpened image of the local contrast-adjusted image; Gamma correction is applied to the edge-sharpened image to obtain the final enhanced image of the transmission line.

[0058] The step of performing gamma correction on the edge-sharpened image to obtain the final enhanced image of the transmission line includes: The center trend of the pixels in the edge-sharpened image is measured to obtain the overall brightness benchmark of the edge-sharpened image; Gain derivation is performed on the preset target luminance median and the overall luminance reference to obtain the correction intensity coefficient of the transmission line; Based on the overall brightness reference and the correction intensity coefficient, the pixels of the edge-sharpened image are grayscale corrected to obtain the corrected pixel values ​​of the edge-sharpened image; The corrected pixel values ​​are then used for image reconstruction to obtain the final enhanced image of the transmission line.

[0059] The process involves traversing all pixels within the composite super-resolution image, recording the pixel value of each pixel, and counting the frequency of each pixel value. By traversing and comparing, the minimum and maximum pixel values ​​in the composite super-resolution image are determined. The original pixel values ​​of the composite super-resolution image are then proportionally mapped to the standard pixel value range of the transmission line inspection image to perform linear stretching, making the overall pixel distribution of the image more uniform and obtaining a global contrast-enhanced image of the transmission line.

[0060] The global contrast-enhanced image is divided into local regions of uniform size and non-overlapping size, which can completely cover the entire image, according to a fixed side length. The cumulative distribution of internal pixel values ​​from low to high is calculated separately for each local region. A continuous distribution curve is fitted according to the trend of the cumulative distribution. The target enhanced pixel value corresponding to each original pixel value is determined, forming a unique correspondence rule, and the enhancement mapping relationship of each local region is obtained.

[0061] According to the enhancement mapping relationship corresponding to each local region, the target pixel value corresponding to the original pixel value of each pixel in the local region is found one by one. The original pixel value is replaced with the target pixel value to complete the value range conversion. After all local regions are processed, they are combined into a complete image to obtain the local contrast adjustment image of the transmission line.

[0062] Centered on each pixel of the locally contrast-adjusted image, a fixed range of neighboring pixels are selected for comparison. Pixel information with pixel value changes exceeding a fixed change threshold is extracted. Edge and detail features such as wires and hardware in the image are preserved, while smooth pixel information in the background area is weakened, resulting in an edge-sharpened image of the locally contrast-adjusted image.

[0063] Arrange the pixel values ​​of all pixels within the edge-sharpened image in ascending order, and select the pixel value in the middle position after the arrangement as the standard value for measuring the overall brightness. This value can stably reflect the overall brightness level of the image, thus obtaining the overall brightness benchmark of the edge-sharpened image.

[0064] The median target brightness value of the transmission line inspection image is preset and the overall brightness benchmark is calculated. Based on the fixed value range of the calculated difference, the corresponding brightness adjustment intensity level is determined. This level is used as a fixed correction parameter to obtain the correction intensity coefficient of the transmission line.

[0065] Using the overall brightness benchmark as the adjustment center, the pixel value of each pixel in the edge-sharpened image is corrected in a targeted manner by combining the correction intensity coefficient. If the overall brightness is too low, the pixel value is increased simultaneously; if the overall brightness is too high, the pixel value is decreased simultaneously, so that all pixel values ​​meet the preset brightness standard, and the corrected pixel values ​​of the edge-sharpened image are obtained.

[0066] Following the original horizontal and vertical pixel arrangement of the edge-sharpened image, all corrected pixel values ​​are sequentially placed in their original coordinate positions, maintaining the geometric structure and spatial feature distribution of the image unchanged, thus completing the reconstruction process of the complete image and obtaining the final enhanced image of the transmission line.

[0067] The beneficial effects are that the overall contrast of the composite super-resolution image is comprehensively improved through global histogram statistics and linear mapping, the local contrast is finely adjusted by relying on block cumulative distribution fitting, the image edge and detail features are enhanced by high-pass filtering, and the overall brightness is optimized by gamma correction. This allows the global and local visual effects of the transmission line image to be improved simultaneously, the details to be complete and clear, and the brightness distribution to be uniform and reasonable, providing high-quality image support for fault identification and status analysis of transmission line inspection.

[0068] like Figure 2 The diagram shown is a functional block diagram of a power transmission line image enhancement system based on super-resolution reconstruction, provided in an embodiment of the present invention.

[0069] The image enhancement system 100 for power transmission lines based on super-resolution reconstruction described in this invention can be installed in an electronic device. Depending on the functions implemented, the image enhancement system 100 may include a brightness statistics module 101, a layered processing module 102, a scale space module 103, a structure reconstruction module 104, a backtracking inference module 105, and a visual optimization module 106. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, and are stored in the memory of the electronic device.

[0070] In this embodiment, the functions of each module / unit are as follows: The brightness statistics module 101 is used to perform grayscale statistics on the original images of the original inspection image sequence of the transmission line to obtain the global brightness distribution characteristics of the original images. The layering processing module 102 is used to perform contrast layering on the original image based on the global brightness distribution features to obtain the brightness layer image set of the transmission line. The scale space module 103 is used to construct the scale space of the brightness level images in the brightness level image set to obtain the multi-scale gradient field of the brightness level images, and to construct the significant geometric structure feature map of the transmission line based on the multi-scale gradient field. The structure reconstruction module 104 is used to perform guided filtering on the brightness level image to obtain the low-frequency base layer and high-frequency detail layer of the brightness level image, and perform structure-aware reconstruction on the low-frequency base layer and the high-frequency detail layer based on the significant geometric structure feature map to obtain the level super-resolution image of the brightness level image. The backtracking inference module 105 is used to perform brightness field backtracking on the hierarchical super-resolution image and the original image to obtain the original brightness hierarchy mapping relationship of the hierarchical super-resolution image, and to perform Bayesian inference on the hierarchical super-resolution image based on the original brightness hierarchy mapping relationship to obtain the composite super-resolution image of the transmission line. The visual optimization module 106 is used to perform visual optimization on the composite super-resolution image to obtain the final enhanced image of the transmission line.

[0071] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0072] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0073] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0074] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0075] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0076] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for enhancing transmission line images based on super-resolution reconstruction, characterized in that, The method includes: C1. Perform grayscale statistics on the original images of the original inspection image sequence of the transmission line to obtain the global brightness distribution characteristics of the original images; C2. Based on the global brightness distribution features, the original image is contrast-layered to obtain the brightness layer image set of the transmission line; C3. Scale space construction is performed on the brightness level images in the brightness level image set to obtain the multi-scale gradient field of the brightness level images, and the significant geometric structure feature map of the transmission line is constructed based on the multi-scale gradient field. C4. Perform guided filtering on the brightness level image to obtain the low-frequency base layer and high-frequency detail layer of the brightness level image, and perform structure-aware reconstruction on the low-frequency base layer and the high-frequency detail layer based on the significant geometric structure feature map to obtain the level super-resolution image of the brightness level image. C5. Perform luminance field backtracking on the hierarchical super-resolution image and the original image to obtain the original luminance hierarchy mapping relationship of the hierarchical super-resolution image, and perform Bayesian inference on the hierarchical super-resolution image based on the original luminance hierarchy mapping relationship to obtain the composite super-resolution image of the transmission line. C6. Perform visual optimization on the composite super-resolution image to obtain the final enhanced image of the transmission line.

2. The method for enhancing transmission line images based on super-resolution reconstruction as described in claim 1, characterized in that, The grayscale statistics of the original images in the original inspection image sequence of the transmission line are performed to obtain the global brightness distribution features of the original images, including: Obtain the original inspection image sequence of the transmission line; The pixels of the original image in the original inspection image sequence are converted to brightness to obtain the gray value of the pixel, and the frequency of occurrence of the gray value is accumulated to obtain the gray frequency distribution record of the original image. The frequencies of adjacent gray values ​​in the gray-level frequency distribution record are merged to obtain the gray-level interval and gray-level interval aggregate frequency of the transmission line. The distribution of the gray-level interval and gray-level interval aggregate frequency is constructed to obtain the gray-level interval aggregate distribution of the original image. Based on the gray-level interval aggregation frequency, peak location is performed on the gray-level interval aggregation distribution to obtain the maximum aggregation frequency interval of the gray-level interval aggregation distribution, and the gray-level lower limit and gray-level upper limit of the maximum aggregation frequency interval are used as the main brightness interval of the original image; The main brightness range and the grayscale range are aggregated and mapped collaboratively to obtain the global brightness distribution features of the original image.

3. The method for enhancing transmission line images based on super-resolution reconstruction as described in claim 1, characterized in that, The step of performing contrast layering on the original image based on the global brightness distribution features to obtain a brightness layer image set for the transmission line includes: Based on the main brightness interval of the global brightness distribution features, the pixel points of the original image are classified to obtain the main brightness level of the pixel points; The main brightness level pixels are extracted by masking to obtain the main brightness level image of the power transmission line; Within the non-primary brightness range of the global brightness distribution features, grayscale values ​​in the original image are clustered to obtain high-proportion secondary brightness level images, medium-proportion secondary brightness level images, and low-proportion secondary brightness level images of the transmission line. The main brightness level image, the high-proportion secondary brightness level image, the medium-proportion secondary brightness level image, and the low-proportion secondary brightness level image are jointly filtered to obtain the brightness level image set of the transmission line.

4. The method for enhancing transmission line images based on super-resolution reconstruction as described in claim 1, characterized in that, The process of constructing a scale space for the luminance level images in the luminance level image set to obtain a multi-scale gradient field of the luminance level images, and constructing a significant geometric structure feature map of the transmission line based on the multi-scale gradient field, includes: Gaussian smoothing kernel convolution is performed on the brightness level images in the brightness level image set to obtain the scale space image sequence of the brightness level images; Perform a difference transform on the scale space images in the scale space image sequence to obtain the horizontal and vertical gray-scale change rates of the scale space images; The horizontal gray-level change rate and the vertical gray-level change rate are vector-synthesized to obtain the single-scale gradient field of the scale space image; Based on a preset scale order, the single-scale gradient fields are stacked sequentially to obtain the multi-scale gradient field of the brightness level image. Extremum detection is performed on the multi-scale gradient field to obtain local maxima points of the multi-scale gradient field, and point set fitting is performed on the local maxima points to obtain candidate structure line segments of the multi-scale gradient field. Spatial projection is performed on the candidate structural line segments to obtain a significant geometric structural feature map of the transmission line.

5. The method for enhancing transmission line images based on super-resolution reconstruction as described in claim 1, characterized in that, The process involves guiding filtering the brightness level image to obtain a low-frequency base layer and a high-frequency detail layer, and then performing structure-aware reconstruction on the low-frequency base layer and the high-frequency detail layer based on the salient geometric structure feature map to obtain a hierarchical super-resolution image of the brightness level image, including: Using the brightness level image as a guide image, the brightness level image is truncated in the frequency domain to obtain the low-frequency base layer of the brightness level image; The residual between the brightness level image and the low-frequency base layer is extracted to obtain the high-frequency detail layer of the brightness level image, and bilinear interpolation is performed on the low-frequency base layer to obtain the magnified low-frequency layer of the low-frequency base layer. Angle marking is performed on the salient geometric structure feature map to obtain the geometric direction label of the salient geometric structure feature map; Based on the geometric direction label, anisotropic diffusion is performed on the high-frequency detail layer to obtain the directional enhancement detail layer of the brightness level image; Based on the amplified low-frequency layer, the directional enhancement detail layer is amplified proportionally to obtain the amplified high-frequency layer of the directional enhancement detail layer. The amplified low-frequency layer and the amplified high-frequency layer are cascaded and reconstructed to obtain the hierarchical super-resolution image of the brightness level image.

6. The method for enhancing transmission line images based on super-resolution reconstruction as described in claim 3, characterized in that, The process of performing luminance field backtracking on the hierarchical super-resolution image and the original image to obtain the original luminance hierarchy mapping relationship of the hierarchical super-resolution image, and performing Bayesian inference on the hierarchical super-resolution image based on the original luminance hierarchy mapping relationship to obtain the composite super-resolution image of the transmission line, includes: The pixels in the hierarchical super-resolution image are reverse-mapped back to the pixel coordinate space of the original image to obtain the original image coordinate position of the hierarchical super-resolution image. Based on the original image coordinates, pixel attribution is determined for the brightness level image set to obtain the brightness level label of the original image coordinates. The brightness level label is then associated and bound with the pixels in the super-resolution image to obtain the original brightness level mapping relationship of the super-resolution image. Based on the original brightness level mapping relationship, the pixel density in the super-resolution image is estimated to obtain the prior distribution features of the pixel. Anomaly detection is performed on the pixels in the hierarchical super-resolution image and the prior distribution features to obtain the outlier pixels in the hierarchical super-resolution image; Based on the prior distribution characteristics, prior constraint optimization is performed on the outlier pixels to obtain the posterior estimate of the outlier pixels. The posterior estimates are used to drive reconstruction to obtain a composite super-resolution image of the transmission line.

7. The method for enhancing transmission line images based on super-resolution reconstruction as described in claim 6, characterized in that, The formula for calculating the posterior estimate is as follows: ； In the formula, The index of the pixel in the super-resolution image is given. For the hierarchical super-resolution image, the first The posterior estimate of each outlier pixel. For the hierarchical super-resolution image, the first The original observations of the outlier pixels, The first in the original brightness level mapping relationship Brightness level labels for each pixel For the brightness level label The corresponding prior distribution mean, For the brightness level label The corresponding prior distribution variance, This is the preset observation noise variance.

8. The method for enhancing transmission line images based on super-resolution reconstruction as described in claim 1, characterized in that, The step of visually optimizing the composite super-resolution image to obtain the final enhanced image of the transmission line includes: Global histogram statistics are performed on the composite super-resolution image to obtain the pixel value distribution range of the composite super-resolution image, and the pixel values ​​of the composite super-resolution image are linearly mapped according to the pixel value distribution range to obtain the global contrast-enhanced image of the transmission line. The global contrast enhancement image is divided into blocks to obtain local regions of the global contrast enhancement image, and local cumulative distribution fitting is performed on the local regions to obtain the enhancement mapping relationship of each local region. Based on the enhanced mapping relationship, the value range of the pixels in the local region is transformed to obtain the local contrast adjusted image of the transmission line; High-pass filtering is applied to the local contrast-adjusted image to obtain the edge-sharpened image of the local contrast-adjusted image; Gamma correction is applied to the edge-sharpened image to obtain the final enhanced image of the transmission line.

9. The method for enhancing transmission line images based on super-resolution reconstruction as described in claim 8, characterized in that, The step of performing gamma correction on the edge-sharpened image to obtain the final enhanced image of the transmission line includes: The center trend of the pixels in the edge-sharpened image is measured to obtain the overall brightness benchmark of the edge-sharpened image; Gain derivation is performed on the preset target luminance median and the overall luminance reference to obtain the correction intensity coefficient of the transmission line; Based on the overall brightness reference and the correction intensity coefficient, the pixels of the edge-sharpened image are grayscale corrected to obtain the corrected pixel values ​​of the edge-sharpened image; The corrected pixel values ​​are then used for image reconstruction to obtain the final enhanced image of the transmission line.

10. A transmission line image enhancement system based on super-resolution reconstruction, characterized in that, The system for implementing the transmission line image enhancement method based on super-resolution reconstruction as described in claim 1 includes: The brightness statistics module is used to perform grayscale statistics on the original images of the original inspection image sequence of the transmission line to obtain the global brightness distribution characteristics of the original images. The layering processing module is used to perform contrast layering on the original image based on the global brightness distribution features to obtain a brightness layer image set of the transmission line. The scale space module is used to construct the scale space of the brightness level images in the brightness level image set, obtain the multi-scale gradient field of the brightness level images, and construct the significant geometric structure feature map of the transmission line based on the multi-scale gradient field. The structure reconstruction module is used to perform guided filtering on the brightness level image to obtain the low-frequency base layer and high-frequency detail layer of the brightness level image, and to perform structure-aware reconstruction on the low-frequency base layer and the high-frequency detail layer based on the significant geometric structure feature map to obtain the level super-resolution image of the brightness level image. The backtracking inference module is used to perform brightness field backtracking on the hierarchical super-resolution image and the original image to obtain the original brightness hierarchy mapping relationship of the hierarchical super-resolution image, and to perform Bayesian inference on the hierarchical super-resolution image based on the original brightness hierarchy mapping relationship to obtain the composite super-resolution image of the transmission line. The visual optimization module is used to perform visual optimization on the composite super-resolution image to obtain the final enhanced image of the transmission line.

Images