A method and system for surface oxidation detection of overhead strand

Abstract

The present application belongs to the technical field of image processing, and particularly relates to a surface oxidation detection method and system for overhead strand, which comprises the following steps: obtaining gradient field data of the overhead strand image; determining the local main direction entropy of the pixel point based on the angle difference between the gradient direction of all neighborhood pixel points of the pixel point and the local main direction, and determining the entropy threshold through clustering to screen anchor points; constructing a gradient profile line along the local main direction of the anchor point, determining the local rust width of the anchor point according to the gradient amplitude change of the pixel points on the gradient profile line; obtaining the adaptive structure element parameters based on the local rust width and the local main direction of the anchor point, and performing morphological top-hat transformation to calculate the response value of the anchor point, and marking the anchor points with a response value greater than the response threshold as oxidation defects. The present application significantly improves the accuracy of overhead strand surface oxidation rust recognition by adaptively matching the local geometric morphology of the rust.

Description

Technical Field

[0001] This invention relates to the field of image processing technology. More specifically, this invention relates to a method and system for detecting surface oxidation in overhead stranded cables. Background Technology

[0002] As a key load-bearing and conductive component of power transmission systems, the surface quality of overhead stranded wires before leaving the factory directly affects the long-term stability and safety of the power grid. During storage and quality inspection, the surface of the stranded wires is prone to oxidation and rust of varying shapes due to environmental factors. If these are not detected and treated in time, they will seriously affect product quality and even pose safety hazards. Therefore, in order to replace inefficient manual visual inspection, the development of automated image inspection methods has become an urgent need in the industry.

[0003] In the field of automated inspection, a core technical problem for the fine, linear oxidation rust on the surface of overhead stranded wires is the complexity and variability of rust morphology. Under different storage conditions and oxidation levels, the width, length, curvature and local orientation of rust all show significant inconsistency, which poses a severe challenge to the adaptability and accuracy of inspection algorithms.

[0004] Existing technologies often employ morphological top-hat transformation algorithms, which perform morphological operations on stranded wire images using a preset fixed structuring element (SE) to extract bright features smaller than the structuring element as defect areas. Due to the variability of actual rust morphology, if the actual width or local orientation of the rust deviates significantly from the preset structuring element parameters, matching will fail, resulting in missed defect detection. Conversely, when the texture of the stranded wire itself in the background is similar in morphology to the fixed structuring element, it is very easy to be misidentified as the target, leading to false detection and seriously affecting the reliability and accuracy of the detection results. Summary of the Invention

[0005] To address the technical problem of low detection accuracy caused by the inability of the aforementioned structural elements to adaptively match the varied shapes of rust, this invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a method for detecting surface oxidation of overhead stranded wires, comprising: acquiring gradient field data of an image of an overhead stranded wire, the gradient field data including the gradient magnitude and gradient direction of each pixel; acquiring the local principal direction of any pixel, the local principal direction being equal to the gradient direction corresponding to the peak value of the gradient directions of all neighboring pixels of that pixel; determining the local principal direction entropy of the pixel based on the angle difference between the gradient directions of all neighboring pixels of that pixel and the local principal direction; statistically analyzing the distribution of the local principal direction entropy of all pixels and determining an entropy threshold; and extracting pixels whose local principal direction entropy is less than the entropy threshold as... Anchor points are defined; a gradient profile line is constructed along the local principal direction of each anchor point, and the local rust width of each anchor point is determined based on the gradient magnitude change of the pixels on the gradient profile line; the local rust width is the distance between the left and right edge points of the anchor point, and the left and right edge points are the pixels corresponding to the initial local maximum value of the gradient magnitude when searching along the gradient profile line from the anchor point to both sides; for each anchor point, adaptive structural element parameters are obtained, including the vertical direction of the local principal direction, the local rust width, and the empirical length; and morphological top-hat transformation is performed using the parameters to calculate the response value of the anchor point; anchor points with response values ​​greater than the response threshold are marked as oxidation defects.

[0007] This invention acquires gradient field data from images of overhead stranded wires, uses local principal direction entropy to measure the orderliness of gradient directions, effectively filters out anchor points with linear features, initially eliminates interference from messy textured backgrounds, and abandons the globally fixed structural elements used in traditional morphological detection. Instead, it constructs gradient profile lines along the local principal directions of the anchor points, and accurately determines the local rust width of each anchor point by analyzing gradient amplitude changes. Based on this, it constructs adaptive structural elements containing specific directions, widths, and empirical lengths and performs a top-hat transformation. This local adaptive matching mechanism ensures that morphological operations can maximize the enhancement of target signals that highly match the actual rust morphology, while suppressing background textures. It solves the problem of missed detections and false detections caused by the inability of structural elements to adapt to varied rust morphologies in existing technologies, thus improving the accuracy of oxide rust detection.

[0008] Preferably, the method for obtaining the neighboring pixels includes: defining a region centered on each pixel. A neighborhood window is defined as any set of pixels within a given neighborhood window that is considered a neighboring pixel. It is a pre-defined positive odd number.

[0009] Preferably, the method for obtaining the local principal direction entropy includes: calculating the angle difference between the gradient direction of each neighboring pixel and the local principal direction of the pixel, and calculating the probability of each type of angle difference; calculating the local principal direction entropy of the pixel. , In the formula, For the first Local principal direction entropy of each pixel; For the first Within all neighboring pixels of the i-th pixel, the i-th The probability of the difference in angle between the two classes; For the first The index value and total number of the category of the angle difference among all neighboring pixels of a pixel; It is a logarithmic function.

[0010] This invention constructs a local principal direction entropy index and uses the probability distribution of the angle difference between the gradient direction of neighboring pixels and the local principal direction to evaluate the degree of texture disorder in local areas of an image. Since oxidized rust usually presents linear features with consistent direction, while the texture distribution of twisted wire background is disordered, by calculating the entropy value, potential defect areas can be quickly located from complex metal backgrounds, providing high-quality anchor points for subsequent morphological processing.

[0011] Preferably, the step of statistically analyzing the distribution of local principal direction entropy of all pixels and determining the entropy threshold includes: clustering the local principal direction entropy of all pixels using the DBSCAN clustering algorithm to obtain the clustering results; calculating the average value of the local principal direction entropy of pixels in each cluster, and using the minimum local principal direction entropy in the cluster corresponding to the maximum value of the average value as the entropy threshold.

[0012] This invention utilizes the DBSCAN clustering algorithm to analyze the distribution of local principal direction entropy and dynamically determine the optimal entropy threshold. Compared to a fixed threshold, this method can adaptively identify data clusters representing the background and defects based on the specific texture distribution characteristics of each image. It effectively avoids the problem of inaccurate anchor point selection caused by twisted wire texture characteristics and significantly reduces the risk of false detection and false negative detection.

[0013] Preferably, constructing a gradient profile line along the local principal direction of each anchor point includes: defining a line segment along the local principal direction centered on each anchor point as the gradient profile line of that anchor point.

[0014] Preferably, the method for obtaining the pixel corresponding to the initial local maximum value of the gradient magnitude includes: when searching along the anchor point to both sides of the gradient profile line, extracting the pixel corresponding to the gradient magnitude that is initially greater than the two adjacent pixels.

[0015] This invention defines the edge by searching for the initial local maximum value of the gradient amplitude along the gradient profile line. This method utilizes the gradient peak features generated by the gray-scale abrupt change at the edge of the rust to accurately locate the left and right boundaries of the rust at the physical level. This not only eliminates the interference of the internal texture fluctuation of the rust on the width measurement, but also ensures that the width parameter of the subsequently generated adaptive structural element is close to the actual defect size, which is the key to achieving high-precision morphological matching.

[0016] Preferably, the empirical length is equal to the width of the local rust at the anchor point. times, and Take 3.

[0017] Preferably, the calculation of the response value of the anchor point includes: performing an opening operation on the overhead stranded line image using the adaptive structural element parameters of each anchor point, obtaining the operation result image of the anchor point, and extracting the gray value of the anchor point in the operation result image as the operation gray value of the anchor point; the response value of the anchor point is equal to the normalized result of the absolute difference between the gray value of the anchor point and the operation gray value.

[0018] This invention utilizes the difference between the original image's grayscale value and the background estimate after adaptive structuring element opening operation to effectively amplify the real defect signal with high morphological matching, while suppressing background textures with morphological mismatch, ensuring keen capture of subtle defects.

[0019] Preferably, the method for obtaining the gradient field data includes: using the Sobel operator to calculate the gradient magnitude and gradient direction of each pixel in the overhead stranded wire image.

[0020] Secondly, the present invention provides a surface oxidation detection system for overhead stranded wires, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned surface oxidation detection method for overhead stranded wires is implemented.

[0021] By adopting the above technical solution, a computer program for surface oxidation detection of overhead stranded wires is generated and stored in a memory for loading and execution by a processor. A terminal device is then manufactured based on the memory and processor for convenient use.

[0022] The beneficial effects of this invention are as follows:

[0023] (1) This invention introduces local principal direction entropy as a pre-screening feature and uses the direction consistency of the gradient field to distinguish rust from the twisted wire background. Before performing complex morphological operations, most of the background interference is eliminated, which greatly improves the pertinence of the detection algorithm.

[0024] (2) This invention measures the width of rust by constructing gradient profile lines perpendicular to the edge direction. Compared with traditional image segmentation or fixed scale estimation, this method can more realistically reflect the physical size changes of rust at different locations, providing reliable data support for morphological processing.

[0025] (3) By allowing the structural elements to deform in real time with the local features of each anchor point, the present invention solves the technical problem that traditional fixed structural elements cannot adapt to irregular linear defect morphology, and significantly improves the accuracy of oxide rust identification. Attached Figure Description

[0026] Figure 1 This is a flowchart illustrating a surface oxidation detection method for overhead stranded wires according to the present invention. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0029] This invention discloses a method for detecting surface oxidation in overhead stranded wires, referring to... Figure 1 This includes steps S1-S5:

[0030] S1. Obtain gradient field data of the overhead stranded line image.

[0031] It should be noted that in the quality inspection process before the overhead stranded wire leaves the factory, surface oxidation rust is a key inspection indicator. Rust usually appears as thin strip-shaped structures that are brighter or darker than the background of the stranded wire. These structures are manifested as significant local changes in grayscale values ​​in the image. Therefore, it is necessary to obtain a surface image of the stranded wire and calculate its gradient field. The gradient amplitude can reflect the severity of grayscale changes, while the gradient direction provides the orientation of the edge, and this direction is perpendicular to the edge or texture direction in the image. These two parameters are the basic data source for subsequent analysis of rust morphology, direction and width.

[0032] Specifically, at the inspection station before the overhead stranded cable leaves the factory, an industrial camera and a uniform light source are set up. As the overhead stranded cable passes by, its surface image is captured. The captured image is then converted to grayscale, and the Sobel operator is used to calculate the gradient magnitude and gradient direction of each pixel in the image. It should be noted that the Sobel operator for calculating gradient magnitude and gradient direction is a well-known technique and will not be elaborated upon here.

[0033] At this point, the gradient field data of the overhead stranded line image has been obtained.

[0034] S2. Obtain the local principal direction of any pixel; determine the local principal direction entropy of the pixel based on the angle difference between the gradient directions of all neighboring pixels and the local principal direction; statistically analyze the distribution of the local principal direction entropy of all pixels and determine the entropy threshold; extract pixels whose local principal direction entropy is less than the entropy threshold as anchor points.

[0035] It should be noted that the background of the image on the surface of the overhead stranded cable contains a uniform metallic texture, while when there is oxidation rust on the actual surface, it is typically characterized by lines or stripes. This indicates that the gradient direction of the area with oxidation should be relatively consistent, while the gradient direction distribution in the uniform background area will be very chaotic. Therefore, local principal direction entropy is introduced to measure the orderliness of the gradient direction in this local area. If the direction of the area is highly consistent, its entropy value will be very low; conversely, the entropy value will be very high. By using a lower entropy threshold, the pixels most likely to contain linear rust structures are selected as anchor points to avoid the subsequent width estimation being interfered with by a large number of normal texture areas.

[0036] Specifically, the local principal direction of each pixel is obtained by defining a direction centered on each pixel. The neighborhood window is defined as all pixels within the neighborhood window as the neighboring pixels of the current pixel. The histogram of the gradient directions of all neighboring pixels is calculated, and the gradient direction corresponding to the peak value is extracted as the local principal direction.

[0037] in, The local analysis range of a pixel is defined to capture gradient changes within that local area. The value cannot be too small to avoid failing to capture gradient changes. Furthermore, the analysis process involves acquiring neighboring pixels within a neighborhood window centered on the pixel. The value is an odd number; this invention will The value is set to 7, and implementers can adjust it as needed. It should be added that for pixels in the edge region where a complete neighborhood window cannot be obtained, the gradient information of the pixel itself is used for completion processing.

[0038] For any pixel, calculate the angle difference between the gradient direction of each neighboring pixel and the local principal direction of the pixel. Treat each different angle difference as an independent category and count the probability of each type of angle difference.

[0039] The local principal direction entropy is determined based on the angle difference between the gradient directions of all neighboring pixels of any given pixel and the local principal direction; the local principal direction entropy satisfies the expression:

[0040]

[0041] In the formula, For the first Local principal direction entropy of each pixel; For the first Within all neighboring pixels of the i-th pixel, the i-th The probability of such angle differences occurring; For the first The index value and total number of the category of the angle difference among all neighboring pixels of a pixel; It is a logarithmic function.

[0042] in, Reflecting the The orderliness of the gradient directions of the neighboring pixels of the i-th pixel; the larger this value, the better. The more chaotic the gradient direction distribution of the neighboring pixels of a given pixel, the greater the probability that the pixel is a textured region; conversely, the more chaotic the gradient direction distribution, the greater the probability that the pixel is a textured region. The more consistent the gradient direction distribution of a pixel's neighboring pixels, the greater the probability that the pixel is rust.

[0043] Considering that the local principal direction entropy calculated for oxidized rust areas is significantly smaller than that calculated for uniform metallic texture areas, the local principal direction entropy of all pixels is obtained. The DBSCAN clustering algorithm is then used to cluster the local principal direction entropies of all pixels to obtain the clustering results. The cluster containing the maximum average local principal direction entropy of all pixels in the cluster is calculated and extracted. The minimum local principal direction entropy in the cluster is used as the entropy threshold. Pixels with local principal direction entropy less than the entropy threshold are then extracted as anchor points. It should be noted that the DBSCAN clustering algorithm is a well-known technique and will not be elaborated upon here.

[0044] At this point, the local principal direction entropy of all pixels and all anchor points have been obtained.

[0045] S3. Construct gradient profile lines along the local main direction of each anchor point, and determine the local rust width of each anchor point based on the gradient amplitude changes of the pixels on the gradient profile lines.

[0046] It should be noted that traditional top-hat transformation uses structuring elements with fixed widths. If the actual width of the rust does not match the width of the structuring element, the detection effect will be inaccurate. Therefore, it is necessary to estimate the possible local rust width for each anchor point. Considering that the local principal direction of the anchor point is perpendicular to the edge of the rust, if a gradient profile line is drawn along the local principal direction of the anchor point, that is, the direction perpendicular to the cutting of the rust, two significant local maxima will be formed at the two edges of the rust, that is, at the points where the gray level jumps from the background to the rust and from the rust back to the background. In the background area inside the rust and outside the background, the gradient amplitude is relatively low because the gray level is relatively flat. Therefore, by finding these two peak points, the left and right edges of the rust can be located, and its width can be calculated.

[0047] Specifically, the gradient profile line of each anchor point is constructed by defining a line segment along its local principal direction, centered on each anchor point, as the gradient profile line of that anchor point.

[0048] The left and right edge points of each anchor point are obtained by sampling pixel by pixel from the anchor point to both sides along the gradient profile line of each anchor point, recording the gradient magnitude of each pixel to form a one-dimensional gradient magnitude sequence. When searching along the gradient profile line to both sides of the anchor point, the pixel whose gradient magnitude is first greater than the gradient magnitude of its two adjacent pixels is extracted and marked as the left and right edge points of the anchor point.

[0049] Calculate the Euclidean distance between the left and right edge points of each anchor point, which is used as the local rust width for each anchor point.

[0050] At this point, the local rust width of each anchor point has been obtained.

[0051] S4. For each anchor point, obtain the adaptive structural element parameters, which include the vertical direction of the local main direction, the width of the local rust, and the empirical length.

[0052] It should be noted that due to the varied forms of actual oxidation rust, if a globally uniform structural element is used for identification, the actual width of the rust will deviate significantly from the preset structural element parameters, leading to matching failure and missed detection. Conversely, when the texture of the stranded wire itself is similar in form to a fixed structural element, it is very easy to be misidentified as a target, resulting in false detection. Therefore, the structural element parameters are adaptively determined by combining the actual direction of each rust anchor point and the width of the rust to ensure accurate extraction in subsequent transformations.

[0053] Considering that oxidation rust on the surface of overhead stranded wires usually presents as a slender strip structure with a certain direction, and that the width and length of the rust are generally significantly different, with the length being much greater than the width; therefore, in order to construct a template that can accurately match this linear feature, it is necessary to reasonably set the length parameters of the structural elements.

[0054] Specifically, for each anchor point, adaptive structural element parameters are obtained. The method for obtaining these parameters is as follows: the vertical direction of the local principal direction of each anchor point is used as the direction of the structural element for that anchor point; the width of the local rust corresponding to each anchor point is used as the width of the structural element for that anchor point; and the empirical length of the structural element is set to the width of the local rust at that anchor point. times.

[0055] in, The value of is chosen to ensure that the generated adaptive structuring element can fully cover the longitudinal features along the direction of the rust, thereby preserving the rust signal to the greatest extent in subsequent morphological operations, while suppressing spots or background textures that do not conform to this elongated shape. Therefore, The value cannot be too small, according to the embodiments of the present invention. The value is 3, and can be adjusted according to the actual situation.

[0056] At this point, the adaptive structuring element parameters for each anchor point have been obtained.

[0057] S5. Perform a morphological top-hat transformation using the adaptive structuring element parameters of the anchor points and calculate the response value of the anchor points; mark anchor points with response values ​​greater than the response threshold as oxidation defects.

[0058] It should be noted that, considering the need for accurate rust identification, the globally fixed structuring element parameters of the traditional top-hat transform are replaced with locally adaptive structuring element parameters. In the region of the selected anchor points, i.e., the region of possible rust points, the adaptive structuring element is used to perform the morphological top-hat transform. That is, the result of the image opening operation is subtracted from the original image. Only real rust with a high match between the local morphology and the adaptive structuring element parameters will be significantly enhanced, i.e., it will show a high response value in the top-hat transform result. The background texture with a mismatched morphology will have its response effectively suppressed due to template mismatch, thereby greatly reducing the false detection rate.

[0059] Specifically, for each anchor point, an opening operation is performed on the overhead stranded line image using its adaptive structuring element parameters to obtain the operation result image for that anchor point; and the grayscale value of that anchor point in the operation result image is extracted as the operation grayscale value for that anchor point. It should be noted that the opening operation is a well-known technique and will not be elaborated upon here, and the operation result image is the same size as the overhead stranded line image.

[0060] Calculate the response value of each anchor point based on the deviation between its grayscale value and the calculated grayscale value; the response value satisfies the expression:

[0061]

[0062] In the formula, For the first The response value of each anchor point; For the first The grayscale value of each anchor point; For the first The calculated grayscale value of each anchor point To determine the absolute value sign; This is the standard normalization function.

[0063] Among them, through the first When the adaptive structuring element at each anchor point slides across the image and performs an opening operation, the thin, bright rust spots are filtered out, leaving only a smooth, twisted-line textured background image; that is... Reflecting the The calculated grayscale value of the anchor point, i.e., the estimated grayscale value of the background, is further used in the original image. Subtracting the calculated gray value from the gray value of the first anchor point yields the first anchor point. The response value of each anchor point; if The larger the value, the higher the original gray value of the anchor point is compared to the gray value of the background after the opening operation and smoothing. This means that there is a bright color feature that is more prominent in this position compared to the twisted texture background, which means that the anchor point is more likely to be an oxide rust mark. Conversely, the smaller the value, the smaller the difference between the original gray value of the anchor point and the gray value of the background after the opening operation and smoothing. This means that the anchor point is less likely to be an oxide rust mark.

[0064] Obtain the response value of each anchor point, set the response threshold, and then... Anchor points exceeding the response threshold are marked as oxidation defects, facilitating timely re-inspection by testing personnel.

[0065] The specific value of the response threshold can be set according to the actual application scenario and requirements. In this invention, the preset response threshold is set to 0.6.

[0066] This invention also discloses a surface oxidation detection system for overhead stranded wires, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a surface oxidation detection method for overhead stranded wires according to the present invention.

[0067] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

Claims

1. A method for detecting surface oxidation in overhead stranded wires, characterized in that, include: Acquire gradient field data of an overhead stranded line image, wherein the gradient field data includes the gradient magnitude and gradient direction of each pixel; Obtain the local principal direction of any pixel, where the local principal direction is equal to the gradient direction corresponding to the peak value of the gradient directions of all neighboring pixels of that pixel; The local principal direction entropy of a pixel is determined based on the angle difference between the gradient directions of all its neighboring pixels and the local principal direction, including: Calculate the angle difference between the gradient direction of each neighboring pixel and the local principal direction of the given pixel, and calculate the probability of each type of angle difference; calculate the local principal direction entropy of the given pixel. , In the formula, For the first Local principal direction entropy of each pixel; For the first Within all neighboring pixels of the i-th pixel, the i-th The probability of the difference in angle between the two classes; For the first The index value and total number of the category of the angle difference among all neighboring pixels of a pixel; It is a logarithmic function; The distribution of local principal direction entropy for all pixels is statistically analyzed, and the entropy threshold is determined, including: The local principal direction entropy of all pixels is clustered using the DBSCAN clustering algorithm to obtain the clustering results; the average local principal direction entropy of pixels in each cluster is calculated, and the minimum local principal direction entropy in the cluster corresponding to the maximum value of the average value is used as the entropy threshold. Pixels whose local principal direction entropy is less than the entropy threshold are extracted as anchor points; A gradient profile line is constructed along the local main direction of each anchor point. The local rust width of each anchor point is determined based on the gradient magnitude change of the pixels on the gradient profile line. The local rust width is the distance between the left and right edge points of the anchor point. The left and right edge points are the pixels corresponding to the first local maximum value of the gradient magnitude when searching along the gradient profile line from the anchor point to both sides. For each anchor point, adaptive structural element parameters are obtained, including the vertical direction of the local principal direction, the width of the local rust, and the empirical length; and morphological top-hat transformation is performed using the parameters to calculate the response value of the anchor point; anchor points with response values ​​greater than the response threshold are marked as oxidation defects.

2. The method for detecting surface oxidation of overhead stranded wires according to claim 1, characterized in that, The method for obtaining the neighboring pixels includes: Defined centered on each pixel A neighborhood window is defined as any set of pixels within a given neighborhood window that is considered a neighboring pixel. It is a pre-defined positive odd number.

3. The method for detecting surface oxidation of overhead stranded wires according to claim 1, characterized in that, The construction of gradient profile lines along the local principal direction of each anchor point includes: For each anchor point, define a line segment along its local principal direction as the gradient profile line of that anchor point.

4. The method for detecting surface oxidation of overhead stranded wires according to claim 1, characterized in that, The method for obtaining the pixel corresponding to the initial local maximum value of the gradient magnitude includes: When searching along the anchor point to both sides of the gradient profile line, extract the pixel point whose gradient magnitude is initially greater than the gradient magnitude of its two adjacent pixels.

5. The method for detecting surface oxidation of overhead stranded wires according to claim 1, characterized in that, The empirical length is equal to the width of the local rust at the anchor point. times, and Take 3.

6. The method for detecting surface oxidation of overhead stranded wires according to claim 1, characterized in that, The response value of the calculated anchor point includes: Using the adaptive structuring element parameters of each anchor point, an opening operation is performed on the overhead stranded line image to obtain the operation result image of that anchor point. The gray value of that anchor point in the operation result image is then extracted as the operation gray value of that anchor point. The response value of that anchor point is equal to the normalized result of the absolute difference between the gray value of that anchor point and the operation gray value.

7. The method for detecting surface oxidation of overhead stranded wires according to claim 1, characterized in that, The method for obtaining the gradient field data includes: using the Sobel operator to calculate the gradient magnitude and gradient direction of each pixel in the overhead stranded wire image.

8. A surface oxidation detection system for overhead stranded wires, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement a surface oxidation detection method for overhead stranded wires according to any one of claims 1-7.

Images