Online detection method of ADSS optical cable defects based on machine vision

Abstract

The present application relates to the field of image processing technology, and specifically to an online detection method for defects in ADSS optical cables based on machine vision. The method comprises: obtaining the surface area of ​​the optical cable in the optical cable acquisition image at each position of the ADSS optical cable; extracting the electric trace area in the optical cable acquisition image; obtaining the electric trace difference characteristic value based on the area ratio of all electric trace areas to the optical cable surface area and the grayscale value difference; extracting the carbon particle area in each electric trace area, and obtaining the carbon trace distribution density based on the area and spatial distribution of all carbon particle areas; extracting the skeleton contour in each electric trace area; obtaining the electric trace intensity based on the distance and length between each corner point on the skeleton contour and all adjacent corner points; obtaining the electrocorrosion severity index by combining the electric trace difference characteristic value, carbon trace distribution density and electric trace intensity; performing anomaly detection analysis on all electrocorrosion severity indices to obtain defect detection results. The present application can improve the defect detection accuracy of optical cables.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to an online detection method for ADSS optical cable defects based on machine vision. Background Art

[0002] ADSS optical cable is a metal-free, tension-resistant, non-inductive, and highly insulated all-dielectric, self-supporting optical cable. It is typically used in transmission lines operating at voltages of 35kV and above, providing high-speed, stable, and reliable communication transmission capabilities. During operation, ADSS optical cable can be susceptible to electrical corrosion caused by factors such as dry arcing and corona discharge, particularly during rainy seasons and when humidity is high.

[0003] Optical cable defects caused by electrical corrosion seriously affect the stability of the power communication system. In the existing technology, machine learning and deep learning image segmentation methods are usually used to detect defects in ADSS optical cable images. However, the surrounding environment of the optical cable is complex. During the detection process of the optical cable image, it is easily affected by the surrounding trees and houses, resulting in misjudgment of optical cable defects, resulting in low optical cable defect detection accuracy. Summary of the Invention

[0004] In order to solve the above technical problems, the present application provides an online detection method for ADSS optical cable defects based on machine vision to solve the existing problems.

[0005] The machine vision-based online detection method for ADSS optical cable defects in this application adopts the following technical solutions:

[0006] An embodiment of the present application provides an online detection method for ADSS optical cable defects based on machine vision, the method comprising the following steps:

[0007] S1, obtaining optical cable acquisition images at various positions of the ADSS optical cable; extracting the optical cable surface area in the optical cable acquisition images;

[0008] S2, for each optical cable acquisition image, extract the electrical trace area according to the grayscale distribution in the optical cable acquisition image; obtain the electrical trace difference characteristic value according to the area ratio of all electrical trace areas to the optical cable surface area and the grayscale value difference;

[0009] S3, extracting the carbon particle area according to the grayscale change in each electrical trace area, and obtaining the carbon trace distribution density of the optical cable acquisition image according to the area and spatial distribution of all carbon particle areas in the optical cable acquisition image;

[0010] S4, extracting the skeleton contour in each trace area; obtaining the carbonization density index of each trace area based on the distance and length between each corner point on the skeleton contour and all adjacent corner points; and obtaining the trace intensity of the optical cable collection image by combining the carbonization density indexes of all trace areas in the optical cable collection image;

[0011] S5, comprehensively analyzing the electrical trace difference characteristic values, carbon trace distribution density, and electrical trace intensity of the optical cable collected images to obtain the electrical corrosion severity index of the optical cable collected images; performing anomaly detection and analysis on the electrical corrosion severity indexes of all optical cable collected images to obtain the optical cable defect detection results.

[0012] Furthermore, the method for obtaining the electric trace area includes:

[0013] After grayscale enhancement processing is performed on the optical cable image, threshold analysis is performed to obtain the segmentation threshold; the area smaller than the segmentation threshold is regarded as the electrical trace area.

[0014] Furthermore, the method for obtaining the electrical trace difference characteristic value includes:

[0015] Acquire a standard image of an optical cable; extract the surface area of ​​the optical cable in the standard image of the optical cable;

[0016] The electrical trace difference characteristic value of the i-th optical cable acquisition image is recorded as EI i , Where, PA i represents the total area of ​​all electrical traces in the i-th optical cable acquisition image; TA i G represents the total area of ​​the optical cable surface in the i-th optical cable acquisition image; i,j represents the mean grayscale value of all pixels in the jth electrical trace area in the i-th optical cable acquisition image, G0 represents the mean grayscale value of all pixels in the optical cable surface area in the optical cable standard image, n represents the number of electrical trace areas, |G i,j -G0| means calculate G i,j -The absolute value of G0.

[0017] Furthermore, the method for obtaining the carbon mark distribution density of the optical cable acquisition image includes:

[0018] For each electric trace area, each circular area in the electric trace area is extracted and recorded as each carbon particle area;

[0019] The carbon trace distribution density of the i-th optical cable acquisition image is recorded as CP i , Where A i,m represents the area of ​​the mth carbon particle region in the i-th optical cable acquisition image, d i,(m,k)represents the distance between the center pixel points of the mth and kth carbon particle regions in the i-th optical cable acquisition image, where m≠k; s represents the number of carbon particle regions in the i-th optical cable acquisition image.

[0020] Furthermore, the method for obtaining the carbonization density index of each electric trace area includes:

[0021] Obtain corner points on the skeleton contour; record any corner point as a target point; obtain the corner points on each line centered on the target point in the skeleton contour that are closest to the target point, and record them as adjacent corner points of the target point; obtain the skeleton contour between the target point and its adjacent corner points as the carbonization channel between the target point and its adjacent corner points;

[0022] The carbonization density index of the tracking area is obtained according to the length and number of carbonization channels between all corner points in the tracking area and their adjacent corner points.

[0023] Furthermore, the carbonization density index of the tracking area is obtained according to the length and number of carbonization channels between all corner points and adjacent corner points in the tracking area, including: recording the carbonization density index of the a-th tracking area as TC a , Where, P a represents the area of ​​the ath electric trace region; l a represents the total length of all carbonized channels in the ath tracking area, E a Represents the carbonization channel characteristic number of all corner points in the a-th tracking area.

[0024] Furthermore, the characteristic number of carbonized channels at all corner points in the tracking area is: the average number of connected carbonized channels at all corner points in the tracking area.

[0025] Furthermore, the method for obtaining the electrical trace strength of the optical cable acquisition image includes:

[0026] The average carbonization density index of all electrical trace areas in the optical cable acquisition image is calculated as the electrical trace intensity of the optical cable acquisition image.

[0027] Furthermore, the method for obtaining the electrical corrosion severity index of the optical cable acquisition image includes:

[0028] The sum of the electric trace intensity and the carbon trace distribution density of each optical cable acquisition image is calculated; and the electric trace difference characteristic value of each optical cable acquisition image and the sum are fused to obtain the electric corrosion severity index of each optical cable acquisition image.

[0029] Furthermore, the detection results of the electrical corrosion defects of the optical cable include:

[0030] Anomaly detection is performed on the electrical corrosion severity index of all optical cable images to obtain an anomaly score for each optical cable image. Optical cable images with an anomaly score greater than a preset anomaly threshold are recorded as optical cable defect images. The acquisition location of the optical cable defect image is used as the location where the electrical corrosion defect occurs on the optical cable.

[0031] This application has at least the following beneficial effects:

[0032] This application analyzes the characteristics of the formation of electrocorrosion defects in optical cables. First, the grayscale changes on the surface of the optical cable in the collected image of the optical cable when electrocorrosion occurs are analyzed. The electric trace severity coefficient is constructed based on the area of ​​the electric trace area and the grayscale value difference. The color and brightness changes on the surface of the optical cable when electrocorrosion occurs are taken into account, and the severity of the electrocorrosion phenomenon on the surface of the optical cable is preliminarily evaluated. The characteristics of the carbon particle traces deposited on the surface of the optical cable during the electrocorrosion process are further analyzed. The carbon trace distribution density is constructed based on the area and spatial distribution of the carbon particle area to accurately judge the severity of the electrocorrosion. Furthermore, the shape characteristics of the carbonization channels formed during the electrocorrosion process are analyzed and the carbonization density index is calculated, which can accurately reflect the distribution density of the carbonization channels. The greater the electric trace intensity, the more severe the electrocorrosion phenomenon in the area. Finally, the electric trace difference characteristic value, the carbon trace distribution density and the electric trace intensity are combined to obtain the electric corrosion severity index, which accurately judges the severity of the electric corrosion phenomenon on the surface of each section of the optical cable. The defect detection result of the optical cable is obtained based on the electric corrosion severity index, which fully considers the characteristics of the electrocorrosion defect, can eliminate the interference of environmental factors, and improve the defect detection accuracy of the optical cable. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0034] Figure 1 A flowchart of the steps of the online detection method for ADSS optical cable defects based on machine vision provided in this application;

[0035] Figure 2 A schematic diagram of selecting adjacent corner points provided for one embodiment of the present application. DETAILED DESCRIPTION

[0036] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. In the absence of conflict, the embodiments of this application and the technical features in the embodiments can be combined with each other. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0037] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0038] The specific scheme of the online detection method for ADSS optical cable defects based on machine vision provided by this application is described in detail below with reference to the accompanying drawings.

[0039] An embodiment of the present application provides an online detection method for ADSS optical cable defects based on machine vision. Specifically, the following online detection method for ADSS optical cable defects based on machine vision is provided. Figure 1 , the method comprises the following steps:

[0040] S1, obtaining optical cable acquisition images at various positions of the ADSS optical cable; extracting the optical cable surface area in the optical cable acquisition images.

[0041] A specific implementation scenario of the embodiment of the present application is the detection scenario of electrical corrosion defects of ADSS optical cables. For ADSS optical cables that work in outdoor environments for a long time, as time accumulates, soluble salt substances and dust in the air will accumulate on the surface of the optical cable. After continuous erosion by moisture in the air, the soluble salt substances attached to the outer surface of the optical cable will form a resistance layer. Driven by the induced voltage, a discharge arc will be generated. The arc discharges repeatedly, and the heat generated by the resistance layer causes the optical cable protective sheath to form dendritic carbonized electrical traces, thereby destroying the polymer of the optical cable protective sheath and forming electrical corrosion. The method of defect detection on the image of the ADSS optical cable is usually used for electrical corrosion defect detection.

[0042] Because ADSS optical cables often extend for thousands of meters in practical applications, it's impossible to capture images of the cable's entire surface all at once. Therefore, in this embodiment, a high-resolution camera is used to continuously capture segment-by-segment images of outdoor optical cables under good lighting conditions. The image acquisition process inevitably involves some interference noise, which degrades image quality and impacts the accuracy of ADSS cable defect detection. To reduce this noise, median filtering is used to denoise the acquired standard and captured cable images. Median filtering is a well-known technique, and the specific implementation is omitted for clarity.

[0043] Since background information, such as the ground, trees, etc., is inevitably included when acquiring the surface image of the optical cable, it is necessary to extract the optical cable from the entire image in order to prevent background information from interfering with the accuracy of optical cable defect detection. This embodiment uses a neural network to extract the optical cable area from the acquired optical cable acquisition image. The training details of the neural network are as follows: 2,000 optical cable surface acquisition images are acquired to mark the optical cable surface area, and normalized, that is, the image matrix is ​​converted to a floating point number between [0,1], so that the model converges better. The neural network uses a convolutional neural network for image segmentation training, using the ResNet18 network structure and the Softmax function as the activation function, and finally obtains the trained neural network model.

[0044] The currently captured optical cable image and the standard optical cable image are input into the trained neural network model to obtain the optical cable surface area in the captured and standard optical cable images. In this embodiment, to eliminate interference from environmental factors and accurately detect electrical corrosion defects, the characteristics of electrical corrosion in the captured optical cable image are fully analyzed.

[0045] S2: For each optical cable acquisition image, extract the electrical trace area according to the grayscale distribution in the optical cable acquisition image; obtain the electrical trace difference characteristic value according to the area ratio of all electrical trace areas to the optical cable surface area and the grayscale value difference.

[0046] Due to the different manifestations of the arc's effect on the cable's outer sheath during electrocorrosion, electrocorrosion typically occurs in three common forms: corrosion, tracking, and breakdown. This application primarily detects defects in the cable's outer sheath due to tracking. Tracking refers to the formation of dendritic carbonized channels in the cable's outer sheath under the action of the arc. As the carbonization deepens, the sheath cracks under the tension of the cable.

[0047] To extend the life of the cable, the outer sheath of ADSS cables is typically dark in color. However, when electrical corrosion occurs, the cable's color changes significantly. The heating caused by the current on the cable's surface causes oxidation or partial degradation of the outer sheath material, resulting in a darker, matte-colored outer sheath in the area of ​​the electrical traces.

[0048] Since the electrical trace area is usually darker and matte, it will appear darker in the grayscale image. In order to identify the electrical trace area in the optical cable acquisition image, each optical cable acquisition image is grayscale enhanced and then subjected to threshold analysis to obtain the segmentation threshold; the area smaller than the segmentation threshold is regarded as the electrical trace area.

[0049] It should be noted that grayscale enhancement of an image can improve image contrast and facilitate the identification of objects in the image. In this embodiment, a histogram equalization algorithm is used to enhance the grayscale of the optical cable image, further amplifying the brightness difference between the electrical trace area and the surrounding normal black area. Other image enhancement algorithms may be used in other embodiments of this application, and this application does not limit them.

[0050] Furthermore, this embodiment uses the Otsu method to perform threshold analysis on the optical cable acquisition image after grayscale enhancement processing to obtain a segmentation threshold; since the color of the electric trace area is obviously different from other areas and the color is darker, the area smaller than the segmentation threshold is regarded as the electric trace area.

[0051] Based on the above analysis, the grayscale value of the electrical trace area is lower than that of the normal optical cable area, and the degree of damage to the cable surface caused by electrical traces is related to the area and grayscale value of the electrical trace area. First, under the same acquisition equipment conditions as the optical cable acquisition image, an image of a section of ADSS optical cable with no surface defects is obtained as the optical cable standard image; wherein, the optical cable acquisition image and the optical cable standard image are both grayscale images; using the same acquisition method as the optical cable surface area in the optical cable acquisition image, the optical cable surface area in the optical cable standard image is obtained. Furthermore, based on the area ratio of all electrical trace areas to the optical cable surface area, and the grayscale value differences between different electrical trace areas, the electrical trace difference characteristic value is obtained.

[0052] Specifically, in this embodiment, the electrical trace difference characteristic value of the i-th optical cable acquisition image is recorded as EI i , Where PA i represents the total area of ​​all electrical traces in the i-th optical cable acquisition image; TA i Represents the total area of ​​the optical cable surface in the i-th optical cable acquisition image; the above areas are all represented by the number of pixels; G i,j represents the mean grayscale value of all pixels in the jth electrical trace area in the i-th optical cable acquisition image, G0 represents the mean grayscale value of all pixels in the optical cable surface area in the optical cable standard image, n represents the number of electrical trace areas, |G i,j -G0| means calculate G i,j -The absolute value of G0.

[0053] When electrical corrosion occurs on the surface of the optical cable, the area of ​​the electrical trace area accounts for the proportion of the total area of ​​the entire optical cable area. The larger the value, the gray value of the cable surface will change when the cable surface is corroded. The greater the difference between the gray value and the average gray value of the cable area in the standard cable image, the larger the calculated electrical trace difference characteristic value EI iThe larger the value, the more serious the electrical corrosion on the cable surface.

[0054] S3, extracting carbon particle regions according to the grayscale changes in each electrical trace region, and obtaining the carbon trace distribution density of the optical cable acquisition image according to the area and spatial distribution of all carbon particle regions in the optical cable acquisition image.

[0055] During electrical corrosion of optical cables, arc discharges caused by the electric field generate high temperatures, leading to physical and chemical changes in the cable's outer protective layer, including but not limited to melting, vaporization, and carbonization. These processes release carbon particles, which deposit on the surface or inside the cable, forming carbon traces.

[0056] Therefore, for each electrical trace region, first extract each circular region within the electrical trace region and record it as each carbon particle region. In this embodiment, the Hough circle detection algorithm is used to extract the carbon particle region. In other embodiments of this application, a circular template matching algorithm can be used to match within the electrical trace region to obtain the carbon particle region, which is not limited in this application. Furthermore, based on the distribution and area of ​​the carbon particles, the carbon trace distribution density on the cable surface is obtained.

[0057] Carbon particles exist in the form of spheres on the surface of the outer protective sheath of the optical cable. The more carbon particle traces there are and the denser the distribution, the more severe the corrosion of the optical cable surface is.

[0058] The carbon trace distribution density of the i-th optical cable acquisition image is recorded as CP i , the calculation formula is:

[0059]

[0060] Where A i,m represents the area of ​​the mth carbon particle region in the i-th optical cable acquisition image, d i,(m,k) represents the distance between the center pixel points of the mth and kth carbon particle regions in the i-th optical cable acquisition image, where m≠k. In this embodiment, the Euclidean distance is used to calculate the distance; s represents the number of carbon particle regions in the i-th optical cable acquisition image.

[0061] When carbon particle traces appear on the surface of the outer protective sheath of the optical cable, the larger the area of ​​the carbon particle region and the smaller the distance between the carbon particle regions, the larger the area of ​​the carbon particles and the denser the distribution, which in turn makes the calculated carbon mark distribution density larger, indicating that the outer protective sheath of the optical cable is more severely eroded by carbon particles.

[0062] S4, extracting the skeleton contour in each trace area; obtaining the carbonization density index of each trace area based on the distance and length between each corner point on the skeleton contour and all adjacent corner points; and obtaining the trace intensity of the optical cable acquisition image by combining the carbonization density indexes of all trace areas in the optical cable acquisition image.

[0063] Within the discharge channel, the concentrated effect of the arc discharge leads to more pronounced carbon deposition, forming a carbonized layer. With repeated and sustained arc discharge, carbon particles accumulate on the inner wall of the discharge channel, forming carbonized channels. Because the arc's position and direction on the cable's outer sheath are not fixed, the carbonized channels typically exhibit a dendritic pattern. These dendritic channels interweave and expand, ultimately forming a complete carbonized channel. Therefore, the denser the carbonized channels on the cable's outer sheath, the more severe the electrocorrosion in that area.

[0064] To extract the carbonized channels, this embodiment extracts the skeleton outline of each electrical trace region and obtains corner points on the skeleton outline. Specifically, this embodiment uses the Zhang-Suen algorithm to extract the skeleton outline of the carbonized channels in the image from the extracted electrical trace regions, and simultaneously uses the Harris algorithm to extract and detect the corner points of the carbonized channels from the obtained skeleton outline. The Zhang-Suen and Harris algorithms are well-known technologies, and their specific implementation processes will not be described in detail. Implementers can select the skeleton outline extraction algorithm and corner point detection method based on actual circumstances, and this application does not impose any restrictions.

[0065] Dendritic carbonization channels have complex morphologies and branching structures, and each intersection may connect multiple carbonization channels. Therefore, any corner point is recorded as the target point. The corner points closest to the target point on each line centered on the target point in the skeleton contour are recorded as the target point's adjacent corner points. The skeleton contour between the target point and its adjacent corner points is obtained as the carbonization channel between the target point and its adjacent corner points.

[0066] In this embodiment, the pixels around each corner point are traversed, and starting from the corner point, the pixels along the skeleton outline are tracked until the next corner point is encountered, which is the adjacent corner point, which may be the intersection point or the end point of the skeleton outline; Figure 2 As shown, the adjacent corner points of corner point A are A1, A2 and A3.

[0067] Furthermore, based on the above analysis, the dendritic characteristics of carbonization channels indicate that the more divergent and densely branched the branches, the more severe the electrocorrosion in that area. The carbonization density index of the tracking area was obtained based on the length and number of carbonization channels between all corner points and their adjacent corner points in the tracking area.

[0068] As an embodiment of the present application, the number of pixel points on the skeleton outline passed during the tracking process is recorded as the length of the carbonization channel. For each corner point, all carbonization channels connected to it are recorded and identified, and finally the number of carbonization channels connected to each corner point and the length of each carbonization channel are counted. Based on the above analysis, the carbonization density index is constructed: the carbonization density index of the a-th trace area is recorded as TCa , Where, P a represents the area of ​​the ath electric trace region; l a represents the total length of all carbonized channels in the ath tracking area, E a Represents the carbonization channel characteristic number of all corner points in the a-th tracking area.

[0069] The characteristic number of carbonization channels of all corner points in the trace area reflects the number of carbonization channels connected to each corner point; the more connected carbonization channels are, the denser the distribution of carbonization channels is; in this embodiment, the characteristic number of carbonization channels of all corner points in the trace area is: the mean of the number of connected carbonization channels of all corner points in the trace area; in other embodiments of the present application, the mode of the number of connected carbonization channels of all corner points in the trace area can be selected as the characteristic number of carbonization channels of all corner points in the trace area.

[0070] When electrical corrosion occurs on the surface of the optical cable, the longer the length of the carbonized channel per unit area in the electrical trace area, the denser the carbonized channel. At this time, the more carbonized channels connected to each corner point, the more branches of the carbonized channels. In turn, the larger the calculated electrical trace density index, the denser the distribution of the carbonized channels, and the more serious the electrical corrosion phenomenon on the optical cable surface.

[0071] Therefore, to comprehensively evaluate the distribution of electrical traces in the cable image, the average carbonization density index of all electrical trace areas in the cable image is calculated as the electrical trace intensity of the cable image. The average carbonization density index of all electrical trace areas in the cable image represents the intensity of electrical traces in the cable image. The greater the electrical trace intensity, the more likely it is to cause severe corrosion on the cable surface.

[0072] S5, comprehensively analyzing the electrical trace difference characteristic values, carbon trace distribution density, and electrical trace intensity of the optical cable collected images to obtain the electrical corrosion severity index of the optical cable collected images; performing anomaly detection and analysis on the electrical corrosion severity indexes of all optical cable collected images to obtain the optical cable defect detection results.

[0073] The above analysis shows that electrical corrosion on the cable surface primarily occurs through two stages: repeated arc discharges that form electrical traces, and high temperatures that cause physical and chemical changes in the cable's outer protective layer, resulting in carbon particle traces. Therefore, the electrical trace intensity and carbon trace distribution density calculated above can be used to characterize the electrical corrosion phenomenon on the cable surface. Combined with the corrosion characteristics initially reflected by the calculated electrical trace difference characteristic values, the severity of the electrical corrosion phenomenon on the cable surface can be comprehensively assessed.

[0074] Specifically, the sum of the electric trace intensity and the carbon trace distribution density of each optical cable acquisition image is calculated; and the electric trace difference characteristic value of each optical cable acquisition image and the sum are fused to obtain the electric corrosion severity index of each optical cable acquisition image.

[0075] It should be noted that fusion means that there is a same-direction change relationship between each fused data and the fused result, that is, the larger the fused data, the larger the fused result. The specific form of fusion is determined by actual application, and this application does not impose any special restrictions.

[0076] In this embodiment, the electrical corrosion severity index of the i-th optical cable acquisition image is recorded as SI i , SI i =EI i ×(TC i +CP i );where CP i represents the carbon trace distribution density of the i-th optical cable acquisition image; EI i represents the electrical trace difference characteristic value of the i-th optical cable acquisition image; TC i Represents the electrical trace intensity of the i-th optical cable acquisition image.

[0077] As another embodiment of the present application, the corrosion severity index SI of the i-th optical cable acquisition image is i =EI i ×exp(TC i +CP i ); where exp() is an exponential function with the natural constant e as the base, which is used to amplify the importance of the electrical trace intensity and carbon trace distribution density for calculating the electrical corrosion severity index.

[0078] When carbon particle traces appear on the surface of the outer protective sheath of the optical cable, the greater the electric trace intensity, the more likely it is to cause serious corrosion to the optical cable surface; at this time, the larger the area of ​​the carbon particles, the denser the distribution, and the greater the carbon trace distribution density, the more serious the carbon particle erosion of the outer protective sheath of the optical cable is, and the larger the electric trace difference characteristic value, the more significant the electric corrosion feature, the larger the electric corrosion severity index, and the more serious the electric corrosion phenomenon in the optical cable acquisition image.

[0079] The above analysis yields the electrocorrosion severity index for each optical cable image. When electrocorrosion occurs in an optical cable, it is often concentrated in a specific section or area, rather than being present throughout the entire cable. Therefore, the electrocorrosion severity index of an electrocorroded cable differs significantly from that of a healthy cable. Therefore, anomaly detection is performed on the electrocorrosion severity index of all optical cable images to obtain an anomaly score for each image. Images with an anomaly score greater than the preset anomaly threshold γ = 0.8 are classified as defective cable images. These defective images correspond to images of severely corroded cables.

[0080] In this embodiment, the HBOS algorithm is used to perform anomaly detection on the electrical corrosion severity index of all optical cable collected images; as other embodiments of the present application, the LOF anomaly detection algorithm can be selected, and this application does not impose any restrictions.

[0081] Finally, the location where the optical cable defect image is collected is the location where the electrical corrosion defect occurs on the ADSS optical cable. The larger the anomaly score, the more severe the electrical corrosion phenomenon.

[0082] Through the above description of the implementation method in combination with the accompanying drawings, technical personnel in the relevant field can understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0083] The above content is only a specific implementation method of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, and they should all be covered by the protection scope of the present application.

Claims

1. The online detection method of ADSS optical cable defects based on machine vision is characterized by: The method comprises the following steps: S1, obtaining optical cable acquisition images at various positions of the ADSS optical cable; extracting the optical cable surface area in the optical cable acquisition images; S2, for each optical cable acquisition image, extract the electrical trace area according to the grayscale distribution in the optical cable acquisition image; obtain the electrical trace difference characteristic value according to the area ratio of all electrical trace areas to the optical cable surface area and the grayscale value difference; S3, extracting the carbon particle area according to the grayscale change in each electrical trace area, and obtaining the carbon trace distribution density of the optical cable acquisition image according to the area and spatial distribution of all carbon particle areas in the optical cable acquisition image; S4, extracting the skeleton contour in each trace area; obtaining the carbonization density index of each trace area based on the distance and length between each corner point on the skeleton contour and all adjacent corner points; and obtaining the trace intensity of the optical cable collection image by combining the carbonization density indexes of all trace areas in the optical cable collection image; S5, comprehensively analyzing the electrical trace difference characteristic values, carbon trace distribution density, and electrical trace intensity of the optical cable collected images to obtain the electrical corrosion severity index of the optical cable collected images; performing anomaly detection and analysis on the electrical corrosion severity indexes of all optical cable collected images to obtain the optical cable defect detection results.

2. The ADSS optical cable defect online detection method based on machine vision according to claim 1, characterized in that: The method for obtaining the electric trace area includes: After grayscale enhancement processing is performed on the optical cable image, threshold analysis is performed to obtain the segmentation threshold; the area smaller than the segmentation threshold is regarded as the electrical trace area.

3. The ADSS optical cable defect online detection method based on machine vision according to claim 1, characterized in that: The method for obtaining the electrical trace difference characteristic value includes: Acquire a standard image of an optical cable; extract the surface area of ​​the optical cable in the standard image of the optical cable; The electrical trace difference characteristic value of the i-th optical cable acquisition image is recorded as EI i , Where, PA i represents the total area of ​​all electrical traces in the i-th optical cable acquisition image; TA i G represents the total area of ​​the optical cable surface in the i-th optical cable acquisition image; i,j represents the mean grayscale value of all pixels in the jth electrical trace area in the i-th optical cable acquisition image, G0 represents the mean grayscale value of all pixels in the optical cable surface area in the optical cable standard image, n represents the number of electrical trace areas, |G i,j -G0| means calculate G i,j -The absolute value of G0.

4. The method for online detection of ADSS optical cable defects based on machine vision according to claim 1, wherein: The method for obtaining the carbon mark distribution density of the optical cable acquisition image includes: For each electric trace area, each circular area in the electric trace area is extracted and recorded as each carbon particle area; The carbon trace distribution density of the i-th optical cable acquisition image is recorded as CP i , Where A i,m represents the area of ​​the mth carbon particle region in the i-th optical cable acquisition image, d i,(m,k) represents the distance between the center pixel points of the mth and kth carbon particle regions in the i-th optical cable acquisition image, where m≠k; s represents the number of carbon particle regions in the i-th optical cable acquisition image.

5. The ADSS optical cable defect online detection method based on machine vision according to claim 1, characterized in that: The method for obtaining the carbonization density index of each electric trace area includes: Obtain corner points on the skeleton contour; record any corner point as a target point; obtain the corner points on each line centered on the target point in the skeleton contour that are closest to the target point, and record them as adjacent corner points of the target point; obtain the skeleton contour between the target point and its adjacent corner points as the carbonization channel between the target point and its adjacent corner points; The carbonization density index of the tracking area is obtained according to the length and number of carbonization channels between all corner points in the tracking area and their adjacent corner points.

6. The ADSS optical cable defect online detection method based on machine vision according to claim 5, characterized in that: The carbonization density index of the tracking area is obtained according to the length and number of carbonization channels between all corner points and adjacent corner points in the tracking area, including: recording the carbonization density index of the a-th tracking area as TC a , Where, P a represents the area of ​​the ath electric trace region; l a represents the total length of all carbonized channels in the ath tracking area, E a Represents the carbonization channel characteristic number of all corner points in the a-th tracking area.

7. The ADSS optical cable defect online detection method based on machine vision according to claim 6, characterized in that: The characteristic number of carbonized channels at all corner points in the trace area is: the average number of connected carbonized channels at all corner points in the trace area.

8. The method for online detection of ADSS optical cable defects based on machine vision according to claim 1, wherein: The method for obtaining the electrical trace strength of the optical cable acquisition image comprises: The average carbonization density index of all electrical trace areas in the optical cable acquisition image is calculated as the electrical trace intensity of the optical cable acquisition image.

9. The method for online detection of ADSS optical cable defects based on machine vision according to claim 1, wherein: The method for obtaining the electrical corrosion severity index of the optical cable acquisition image comprises: The sum of the electric trace intensity and the carbon trace distribution density of each optical cable acquisition image is calculated; and the electric trace difference characteristic value of each optical cable acquisition image and the sum are fused to obtain the electric corrosion severity index of each optical cable acquisition image.

10. The method for online detection of ADSS optical cable defects based on machine vision according to claim 1, wherein: The detection results of the electrical corrosion defects of the optical cable include: Anomaly detection is performed on the electrical corrosion severity index of all optical cable images to obtain an anomaly score for each optical cable image. Optical cable images with an anomaly score greater than a preset anomaly threshold are recorded as optical cable defect images. The acquisition location of the optical cable defect image is used as the location where the electrical corrosion defect occurs on the optical cable.

Images