A method for detecting surface defects of pantograph sliding plate
By detecting and filtering straight line segments based on the edge characteristics of the pantograph, a selection box is constructed to filter out the pantograph area and detect defects. This solves the problems of low efficiency and high implementation difficulty of existing detection methods, and realizes high-precision and fast pantograph pantograph surface defect detection.
Patent Information
- Application Number
- CN202311871637.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-12-29
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-12-29
AI Technical Summary
Existing methods for detecting wear defects in pantograph sliders are inefficient and pose safety hazards, and existing technologies are difficult to implement or have poor feasibility.
Based on the characteristic that the upper and lower edges of the skateboard are straight lines, straight line segment contour detection is used to filter out straight line contours that are similar to the slope and intercept of the upper and lower edges of the skateboard. A selection box is constructed to select the skateboard area, and the actual defect area is filtered out by area and contrast filtering conditions.
It improves the accuracy and speed of detection, effectively suppresses background interference, accurately locates the pantograph area, and is suitable for surface defect detection of various types of pantograph pantographs, meeting the accuracy and time requirements of visual inspection in rail transit.
Smart Images

Figure CN117808782B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of pantograph inspection, and more specifically to a method for detecting surface defects on the pantograph sliding plate. Background Technology
[0002] The pantograph is an important piece of equipment for train power supply, used to allow electric locomotives to obtain electrical energy from the overhead contact line. The pantograph is installed on the roof of the locomotive or EMU and can be divided into single-arm and double-arm types. Both consist of a sliding plate, upper frame, lower arm (lower frame for double-arm pantographs), base frame, lifting spring, transmission cylinder, and support insulators.
[0003] Due to material characteristics and operating principles, the pantograph's sliding plate surface is prone to wear defects such as cracks, chipping, and abrasion. If defective pantograph sliding plates are not replaced promptly, they can not only affect the normal operation of trains but also damage equipment, causing incalculable losses. Therefore, defect detection of the pantograph sliding plate surface is of great significance for the efficient operation and maintenance of railway systems.
[0004] Current methods for detecting pantograph slider wear defects include:
[0005] The manual inspection method involves visually inspecting the pantograph sliding plate area for defects, which is not only inefficient but also poses safety hazards for workers working at heights.
[0006] Patent document CN111812101A proposes a foreign object detection method based on stereo matching and SVM model classification. This invention belongs to a passive detection method. Although it has a certain effect on foreign object detection, it requires the joint calculation of three-dimensional information of pantograph sliding plate from multiple images, making the implementation principle relatively complex and difficult to implement.
[0007] Patent document CN115018804A proposes a method for identifying pantograph defects using neuromorphic technology. This invention features small-sample learning and high-speed dynamic response, but the method requires training a deep learning model based on a large amount of data, making the feasibility of implementation questionable. Summary of the Invention
[0008] To address the aforementioned technical problems, this invention provides a method for detecting surface defects on pantograph sliding plates. Based on the straight-line characteristic of the upper and lower edges of the sliding plate, a straight-line contour detection method is used to filter out straight-line contours with slopes and intercepts similar to the upper and lower edges of the sliding plate. A selection box is then constructed based on the selected straight lines to define the sliding plate area, effectively suppressing interference from the surrounding background and surface debris, and accurately locating the sliding plate area. Area and contrast filtering conditions are set to filter connected components within the sliding plate area, ensuring that only qualified connected components are selected to identify the true defect area, while filtering out false defect areas such as small scratches, dust, mud spots, and light spots. This method offers higher precision and accuracy, and is faster, applicable to the detection of surface defects on various types of pantograph sliding plates, meeting the requirements for accuracy and time in pantograph defect detection during visual inspection in rail transit.
[0009] Therefore, the technical solution of the present invention is as follows:
[0010] A method for detecting surface defects on a pantograph contactor plate involves fixing a camera to the top of a train to acquire images of the upper surface of the pantograph contactor plate. Before starting the detection, prior information is obtained using the following methods:
[0011] The camera captures an image of the skateboard without wear, which is recorded as the standard image. In this image, the upper or lower edge line of the skateboard is extracted. The allowable range of the slope and intercept is set according to the slope and intercept of the upper or lower edge line and the tolerance value.
[0012] In the standard image, the length direction of the skateboard is marked as feature axis I, and the width direction is marked as feature axis II; select the starting coordinates A and the ending coordinates B of the skateboard on feature axis I; draw a straight line I parallel to feature axis II through the starting coordinates A and a straight line II parallel to feature axis II through the ending coordinates B;
[0013] Calculate the area of the quadrilateral formed by the upper edge line, the lower edge line, line I, and line II, and set the allowable range of the area value using the area value and area tolerance.
[0014] During inspection, the camera captures an image of the current pantograph sliding plate surface, and the following steps are used to detect whether there are defects in the image:
[0015] 1) Extract edges from the entire image, use the obtained edge points to extract straight lines, and obtain the equations of the lines corresponding to multiple straight lines;
[0016] Based on the slope and intercept in the equation of a line, the lines are filtered, and the lines that fall within the allowable interval of the slope and intercept are stored in the line set Q.
[0017] 2) In the set of lines Q, any pair of lines can be combined to obtain multiple pairs of lines;
[0018] Combine multiple pairs of lines with lines I and II respectively to form quadrilaterals, and calculate the area of each quadrilateral;
[0019] Store quadrilaterals whose area values fall within the allowable range of the area values into the quadrilateral set P;
[0020] 3) Join all the quadrilaterals in the quadrilateral set P together, and denote the joined shape as the selection box;
[0021] The image area within the selection box is designated as the skateboard area;
[0022] 4) Extract connected components in the skateboard area, calculate the area of the connected components, the contrast between the connected components and the surrounding areas, and the grayscale fluctuation. When the area value, contrast, and grayscale fluctuation all meet their respective threshold requirements, mark the connected components as defect areas to complete defect detection.
[0023] Furthermore, in step 4), the contrast and grayscale fluctuation between the connected component and its surrounding area are used to determine whether the contrast and grayscale fluctuation meet the threshold requirements.
[0024] Calculate the gray-level variance S and gray-level mean M within the connected component;
[0025] Calculate the gray-level variance S' and gray-level mean M' within the surrounding region of the connected component;
[0026] Determine if the contrast meets the following threshold requirements:
[0027] |M-M'|>Mt
[0028] Where Mt is the contrast threshold, with a value of 30-50;
[0029] Determine whether the grayscale fluctuation meets the following threshold requirements:
[0030] S>S'
[0031] Smin≤S≤Smax
[0032] Wherein, Smin is the minimum threshold for grayscale variance, with a value of 0.01-0.05, and Smax is the maximum threshold for grayscale variance, with a value of 0.1-0.2.
[0033] Preferably, the surrounding region of the connected component is obtained in the following way: the connected component is expanded, and the expanded region is denoted as region C. The region containing the connected component is removed from region C, and the remaining region is denoted as the surrounding region of the connected component.
[0034] Furthermore, in step 4), the method for calculating the area of the connected components and determining whether the area value meets the threshold requirement is as follows:
[0035] Calculate the area of the connected component and denote it as the area value G;
[0036] Calculate the area of the smallest bounding rectangle of the connected component, denoted as the area value G';
[0037]
[0038] Among them G min Minimum area threshold of connected components, G min = Minimum area of defect to be detected / Camera pixel equivalent.
[0039] To reduce computation and improve detection speed, preferably, in step 4), the area of the connected region is first calculated, and it is determined whether the area value of the connected region meets the threshold requirement. Connected regions that do not meet the threshold requirement are filtered out. Then, the contrast and grayscale fluctuation of the connected regions that meet the requirements and the surrounding areas are calculated, and it is determined whether the contrast and grayscale fluctuation both meet their respective threshold requirements.
[0040] Furthermore, in step 1), the method for extracting the straight line is the LSD method, LBD method, FLD method, EDLine method, or Hough transform method.
[0041] Preferably, the method for setting the allowable range of slope and intercept based on the slope and intercept of the upper or lower edge line and the tolerance value is as follows:
[0042] Let the slope of the upper edge line be k1 and the intercept be b1;
[0043] The tolerance value Δk of the slope is 0.3 to 0.7, and the allowable range of the slope is set (k1-Δk, k1+Δk);
[0044] The tolerance value Δb of the intercept is s×W, where s is a preset coefficient with a value of 01 to 0.2, and W is the total number of pixels on feature axis II. The allowable range of the intercept is set to (b1-Δb, b1+Δb).
[0045] Record the slope of the edge line as k2 and the intercept as b2;
[0046] The tolerance value Δk of the slope is 0.3 to 0.7, and the allowable range of the slope is set (k2-Δk, k2+Δk);
[0047] The tolerance value Δb of the intercept is s×W, and the allowable interval of the intercept is set as (b2-Δb, b2+Δb).
[0048] Preferably, let Q be the area of the quadrilateral formed by the upper edge line, the lower edge line, line I, and line II in the standard image, let ΔQ be the area tolerance, and let (Q-ΔQ, Q+ΔQ) be the allowable range of the area value; wherein, the area tolerance ΔQ is 0.1 times the area value Q to 0.3 times the area value Q.
[0049] Preferably, the method for edge extraction of the entire image in step 1) is either the Canny detection method or the EdgeDrawing detection method.
[0050] Because the pantograph is installed above the train and operates outdoors, a camera needs to be fixed above the train during measurement to periodically capture images of the pantograph's upper surface (including the pantograph slide plate area) from top to bottom. By monitoring the image quality, the presence of wear defects in the pantograph slide plate area can be analyzed. This method presents the following technical challenges: low contrast between the pantograph slide plate area and the surrounding area in the image makes it difficult to effectively distinguish the boundary; the pantograph slide plate area is interfered with by cluttered information such as contact wires, return lines, and pantograph supports.
[0051] Therefore, common contour localization methods, such as template matching methods, suffer from problems such as low computational efficiency, slow speed, high noise, and inaccurate localization, which affect the effectiveness of the detection results.
[0052] In the field of rail transit inspection, minor scratches and dust on the skateboard surface are not considered defects; only pits with a depth greater than the required size are considered defects. Therefore, when inspecting for defects within the skateboard area, it is necessary to assess whether the area and contrast of connected regions meet the requirements. For example, the minimum area of connected regions must be ≥10×10mm.
[0053] The pantograph skateboard surface defect detection method provided by this invention is based on the characteristic that the upper and lower edges of the skateboard are straight lines. It adopts straight line segment contour detection to screen out straight line contours that are similar to the slope and intercept of the upper and lower edges of the skateboard. Then, a selection box is constructed based on the screened straight lines to select the skateboard area. Compared with traditional contour extraction methods, it can effectively suppress the interference of complex backgrounds and debris on the skateboard surface in the image and accurately locate the skateboard area.
[0054] The filter criteria of area and contrast are set to filter the connected regions within the skateboard area, so that the connected regions that meet the criteria are filtered out to obtain the real defect areas and filter out false defect areas such as small scratches.
[0055] This method offers higher precision and accuracy, and is faster (processing images with a resolution of 4096×3000 takes less than 100ms). It is suitable for detecting surface defects on various types of pantograph sliding plates, meeting the requirements for pantograph defect detection accuracy and detection time in the visual inspection process of rail transit. Attached Figure Description
[0056] Figure 1 This is an image of a portion of the upper surface of the pantograph;
[0057] Figure 2 Image of a skateboard with dents / defects;
[0058] Figure 3 Images of skateboards with shadow interference and irregular scratches;
[0059] Figure 4 Image of a skateboard with visible sides and minor scratches. Detailed Implementation
[0060] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.
[0061] The pantograph is installed above the train and operates outdoors. Therefore, a camera needs to be fixed above the train during measurement to periodically collect images of the pantograph's upper surface (including the sliding plate area) from top to bottom. By monitoring the image quality, it is possible to analyze whether there are any defects in the sliding plate area.
[0062] like Figure 1 As shown, the low contrast between the skateboard area and the surrounding area in the image makes it difficult to effectively distinguish the boundary, and the skateboard area is interfered with by cluttered information such as contact lines, return lines, and pantograph supports.
[0063] This embodiment provides a method for detecting surface defects on a pantograph sliding plate, which aims to effectively locate the sliding plate area and detect defective areas within the sliding plate area.
[0064] The specific plan is as follows:
[0065] The camera is fixed to the top of the train to acquire images of the upper surface of the pantograph contact plate (the surface in contact with the overhead contact line). Prior information is obtained using the following methods before the inspection begins:
[0066] The camera captures an image of the skateboard without wear, which is recorded as the standard image. In this image, the upper or lower edge line of the skateboard is extracted. The allowable range of the slope and intercept is set according to the slope and intercept of the upper or lower edge line and the tolerance value.
[0067] In the standard image, the length direction of the skateboard is marked as feature axis I, and the width direction is marked as feature axis II; select the starting coordinates A and the ending coordinates B of the skateboard on feature axis I; draw a straight line I parallel to feature axis II through the starting coordinates A and a straight line II parallel to feature axis II through the ending coordinates B;
[0068] In practice, when the skateboard area is distributed horizontally, its length direction is characteristic axis I, i.e., the x-axis, and its width direction is characteristic axis II, i.e., the y-axis. Conversely, when the skateboard area is distributed vertically, its length direction is characteristic axis I, i.e., the y-axis, and its width direction is characteristic axis II, i.e., the x-axis.
[0069] When the skateboard area runs horizontally through the image, the starting point coordinates A and the ending point coordinates B are located at the left and right edges of the image, respectively; when the skateboard area runs vertically through the image, the starting point coordinates A and the ending point coordinates B are located at the top and bottom edges of the image, respectively.
[0070] When the skateboard area does not extend through the image, search for the starting point coordinates A and the ending point coordinates B in the image.
[0071] like Figure 2 As shown, calculate the area of the quadrilateral formed by the upper edge line, the lower edge line, line I, and line II, and set the allowable range of the area value using the area value and area tolerance.
[0072] During inspection, the camera captures an image of the current pantograph slide surface, and the following steps are used to detect whether there are defects in the image:
[0073] 1) Extract edges from the entire image, use the obtained edge points to extract straight lines, and obtain the equations of the lines corresponding to multiple straight lines;
[0074] Specifically, the methods for edge extraction of the entire image are: Canny detection method or EdgeDrawing detection method.
[0075] The methods for line extraction include LSD, LBD, FLD, EDLine, or Hough transform.
[0076] Based on the slope and intercept in the linear equation, lines are filtered, and those falling within the allowable range of slope and intercept are stored in the line set Q. This step can filter out the contour interference of surrounding background objects, such as... Figure 3 Line equations 4 and 5 (shaded outline) are given.
[0077] 2) In the set of lines Q, any pair of lines can be combined to obtain multiple pairs of lines;
[0078] Combine multiple pairs of lines with lines I and II respectively to form quadrilaterals, and calculate the area of each quadrilateral;
[0079] Store quadrilaterals whose area values fall within the allowable range of their area values into the quadrilateral set P;
[0080] This step can further filter out background clutter, such as... Figure 3 The equation of the straight line in the image is 1 (the outline of the shaded area). Figure 4 Equation 3 of the straight line (side edge line of the pantograph).
[0081] by Figure 3 Taking equation 1 of the line as an example: Equation 1 of the line and... Figure 3When the line equations 2 in the equation are combined, the area of the quadrilateral formed by the line pair and line I and line II is much smaller than the allowable range of area values, so it will not be stored in the quadrilateral set P.
[0082] 3) Join all the quadrilaterals in the quadrilateral set P together, and denote the joined shape as the selection box;
[0083] The image area within the selection box is designated as the skateboard area;
[0084] The above process analyzes the slope intercept of the lines and the area of the quadrilaterals in the current image to filter out the selection box that is closest to the skateboard area, thereby effectively locating the skateboard area.
[0085] 4) Extract connected components in the skateboard area, calculate the area of the connected components, the contrast between the connected components and the surrounding areas, and the grayscale fluctuation. When the area value, contrast, and grayscale fluctuation all meet their respective threshold requirements, mark the connected components as defect areas to complete defect detection.
[0086] Among them, the contrast and grayscale fluctuation between the connected component and the surrounding area are used to determine whether the contrast and grayscale fluctuation meet the threshold requirements.
[0087] Calculate the gray-level variance S and gray-level mean M within the connected component;
[0088] Calculate the gray-level variance S' and gray-level mean M' within the surrounding region of the connected component;
[0089] like Figure 2 As shown, the defective area is contrasting with the surrounding area; in this embodiment, it is determined whether the contrast meets the following threshold requirements:
[0090] |M-M'|>Mt
[0091] Where Mt is the contrast threshold, with a value of 30-50;
[0092] Determine whether the grayscale fluctuation meets the following threshold requirements:
[0093] S>S'
[0094] Smin≤S≤Smax
[0095] Wherein, Smin is the minimum threshold for grayscale variance, with a value of 0.01-0.05, and Smax is the maximum threshold for grayscale variance, with a value of 0.1-0.2.
[0096] The defective area exhibits greater grayscale fluctuations compared to the surrounding area. Determining whether these grayscale fluctuations meet a threshold requirement allows for the filtering out of defects such as… Figure 3 The irregular scratches shown. And as... Figure 4The slight scratches caused by friction shown are not considered defects in the pantograph because they do not significantly affect its operation. They are considered pseudo-defects. However, if the scratches worsen and form pits (such as...), they will become more noticeable. Figure 2 If the grayscale fluctuation, contrast and area value of a certain area meet the threshold requirements, it will be identified as a defective area of the skateboard.
[0097] More specifically, the surrounding region of a connected component is obtained as follows: the connected component is expanded, and the expanded region is denoted as region C. The region containing the connected component is removed from region C, and the remaining region is denoted as the surrounding region of the connected component.
[0098] In step 4), the area of the connected components is calculated, and the method to determine whether the area value meets the threshold requirement is as follows:
[0099] Calculate the area of the connected component, denoted as the area value G; the method for calculating the area of the connected component is a conventional method.
[0100] Calculate the area of the minimum bounding rectangle of the connected component, denoted as the area value G'; the area of the minimum bounding rectangle is the product of its length and width.
[0101]
[0102] Among them G min Minimum area threshold of connected components, G min = Minimum area of defect to be detected / Camera pixel equivalent. By comparing area thresholds, areas that are too small can be filtered out, such as salt-and-pepper noise, dust, small mud spots, and other stains, as well as light spot noise caused by outdoor light.
[0103] To reduce computational load and improve detection speed, as a preferred implementation method, the area of the connected components is first calculated, and it is determined whether the area value of the connected components meets the threshold requirement. Connected components that do not meet the threshold requirement are filtered out. Then, the contrast and grayscale fluctuation of the connected components that meet the requirements and the surrounding area are calculated, and it is determined whether the contrast and grayscale fluctuation both meet their respective threshold requirements.
[0104] In this embodiment, the method for setting the allowable range of slope and intercept based on the slope and intercept of the upper or lower edge straight line and the tolerance value is as follows:
[0105] Let the slope of the upper edge line be k1 and the intercept be b1;
[0106] The tolerance value Δk of the slope is 0.3 to 0.7, and the allowable range of the slope is set (k1-Δk, k1+Δk);
[0107] The tolerance value Δb of the intercept is s×W, where s is a preset coefficient with a value of 01 to 0.2, and W is the total number of pixels on feature axis II. The allowable range of the intercept is set to (b1-Δb, b1+Δb).
[0108] Record the slope of the edge line as k2 and the intercept as b2;
[0109] The tolerance value Δk of the slope is 0.3 to 0.7, and the allowable range of the slope is set (k2-Δk, k2+Δk);
[0110] The tolerance value Δb of the intercept is s×W, and the allowable interval of the intercept is set as (b2-Δb, b2+Δb).
[0111] Let Q be the area of the quadrilateral formed by the upper edge line, the lower edge line, line I, and line II in the standard image, let ΔQ be the area tolerance, and let (Q-ΔQ, Q+ΔQ) be the allowable range of the area value. Among them, the area tolerance ΔQ takes the value of 0.1 times the area value Q to 0.3 times the area value Q.
[0112] The foregoing description of specific exemplary embodiments of the present invention is for illustrative and descriptive purposes. It is not intended to be exhaustive, nor to limit the invention to the precise forms disclosed; obviously, many changes and variations are possible in accordance with the foregoing teachings. The exemplary embodiments were chosen and described to explain the specific principles of the invention and its practical application, thereby enabling others skilled in the art to implement and utilize various exemplary embodiments of the invention, as well as their different alternatives and modifications. The scope of the invention is intended to be defined by the appended claims and their equivalents.
Claims
1. A method for detecting surface defects on a pantograph sliding plate, characterized in that: A camera is fixed to the top of the train to acquire images of the upper surface of the pantograph's sliding plate. Prior information is obtained using the following methods before the inspection begins: The camera captures an image of the skateboard without wear, which is recorded as the standard image. In this image, the upper or lower edge line of the skateboard is extracted. The allowable range of the slope and intercept is set according to the slope and intercept of the upper or lower edge line and the tolerance value. In the standard image, the length direction of the marked skateboard is characteristic axis I, and the width direction is characteristic axis II; Select the starting coordinates A and ending coordinates B of the skateboard on characteristic axis I; draw a straight line I parallel to characteristic axis II through the starting coordinates A and a straight line II parallel to characteristic axis II through the ending coordinates B; Calculate the area of the quadrilateral formed by the upper edge line, the lower edge line, line I, and line II, and set the allowable range of the area value using the area value and area tolerance. During inspection, the camera captures an image of the current pantograph sliding plate surface, and the following steps are used to detect whether there are defects in the image: 1) Extract edges from the entire image, use the obtained edge points to extract straight lines, and obtain the equations of the lines corresponding to multiple straight lines; Based on the slope and intercept in the equation of a line, the lines are filtered, and the lines that fall within the allowable interval of the slope and intercept are stored in the line set Q. 2) In the set of lines Q, any pair of lines can be combined to obtain multiple pairs of lines; Combine multiple pairs of lines with lines I and II respectively to form quadrilaterals, and calculate the area of each quadrilateral; Store quadrilaterals whose area values fall within the allowable range of the area values into the quadrilateral set P; 3) Join all the quadrilaterals in the quadrilateral set P together, and denote the joined shape as the selection box; The image area within the selection box is designated as the skateboard area; 4) Extract connected components in the skateboard area, calculate the area of the connected components, the contrast between the connected components and the surrounding areas, and the grayscale fluctuation. When the area value, contrast, and grayscale fluctuation all meet their respective threshold requirements, mark the connected components as defect areas to complete defect detection.
2. The method for detecting surface defects of the pantograph sliding plate as described in claim 1, characterized in that: In step 4), the contrast and grayscale fluctuation between the connected component and its surrounding area are used to determine whether the contrast and grayscale fluctuation meet the threshold requirements. Calculate the gray-level variance S and gray-level mean M within the connected component; Calculate the gray-level variance S' and gray-level mean M' within the surrounding region of the connected component; Determine if the contrast meets the following threshold requirements: Where Mt is the contrast threshold, with a value of 30-50; Determine whether the grayscale fluctuation meets the following threshold requirements: ' Wherein, Smin is the minimum threshold for grayscale variance, with a value of 0.01-0.05, and Smax is the maximum threshold for grayscale variance, with a value of 0.1-0.
2.
3. The method for detecting surface defects of the pantograph sliding plate as described in claim 2, characterized in that: The surrounding region of a connected component is obtained as follows: the connected component is expanded, and the expanded region is denoted as region C. The region containing the connected component is removed from region C, and the remaining region is denoted as the surrounding region of the connected component.
4. The method for detecting surface defects of the pantograph sliding plate as described in claim 1, characterized in that: In step 4), the area of the connected components is calculated, and the method to determine whether the area value meets the threshold requirement is as follows: Calculate the area of the connected component and denote it as the area value G; Calculate the area of the smallest bounding rectangle of the connected component, denoted as the area value G'; Among them G min Minimum area threshold of connected components, G min =Minimum area of defect to be detected / Camera pixel equivalent.
5. The method for detecting surface defects of the pantograph sliding plate as described in claim 1, characterized in that: In step 4), first calculate the area of the connected components, determine whether the area value of the connected components meets the threshold requirement, and filter out the connected components that do not meet the threshold requirement; then calculate the contrast and grayscale fluctuation between the connected components that meet the requirements and the surrounding areas, and determine whether the contrast and grayscale fluctuation both meet their respective threshold requirements.
6. The method for detecting surface defects of the pantograph sliding plate as described in claim 1, characterized in that: In step 1), the method for line extraction is the LSD method, LBD method, FLD method, EDLine method, or Hough transform method.
7. The method for detecting surface defects of the pantograph sliding plate as described in claim 1, characterized in that: The method for setting the allowable range of slope and intercept based on the slope and intercept of the upper or lower edge line and the tolerance value is as follows: Let the slope of the upper edge line be k1 and the intercept be b1; The tolerance value Δk of the slope is set to 0.3~0.7, and the allowable range of the slope is set to (k1-Δk, k1+Δk). The tolerance value Δb of the intercept is s×W, where s is a preset coefficient with a value of 01~0.2, and W is the total number of pixels on feature axis II. The allowable range of the intercept is set to (b1-Δb, b1+Δb). Record the slope of the edge line as k2 and the intercept as b2; The tolerance value Δk of the slope is 0.3~0.7, and the allowable range of the slope is set (k2-Δk, k2+Δk). The tolerance value Δb of the intercept is s×W, and the allowable interval of the intercept is set as (b2-Δb, b2+Δb).
8. The method for detecting surface defects of the pantograph sliding plate as described in claim 1, characterized in that: Let Q be the area of the quadrilateral formed by the upper edge line, the lower edge line, line I, and line II in the standard image, let △Q be the area tolerance, and let (Q-△Q, Q+△Q) be the allowable range of the area value. Among them, the area tolerance △Q takes the value of 0.1 times the area value Q to 0.3 times the area value Q.
9. The method for detecting surface defects of the pantograph sliding plate as described in claim 1, characterized in that: The method for edge extraction of the entire image in step 1) is either the Canny detection method or the EdgeDrawing detection method.
Citation Information
Patent Citations
Pantograph slide plate defect detection device and method
CN111812101A
Device for identifying pantograph defects by using brain-like technology
CN115018804A
Online monitoring system for pantograph wear
CN110068593A
ORB algorithm-based pantograph-catenary operation state detection method
CN115331000A
