Steel wire rope surface defect detection method and system based on machine vision

Abstract

The invention discloses a steel wire rope surface defect detection method and system based on machine vision, and relates to the technical field of steel wire rope surface defect detection.The method comprises the following steps that image collection is conducted on the surface of a steel wire rope, and a steel wire rope image is obtained; carrying out edge extraction on the steel wire rope image to obtain an edge contour map of the steel wire rope image, and carrying out region segmentation on the steel wire rope based on the edge contour map to obtain a single-strand steel wire region; linear features of the single-strand steel wire area are extracted; analyzing a normal range of linear characteristics based on the linear characteristics of different single-strand steel wire areas of the defect-free steel wire rope, and detecting surface defects of the steel wire rope; the method is used for solving the problem that misjudgment is easy to occur during steel wire rope surface defect detection due to the fact that an analysis method is not reliable enough in an existing steel wire rope surface defect detection technology.

Description

Technical Field

[0001] The present invention relates to the technical field of steel wire rope surface defect detection, and specifically to a steel wire rope surface defect detection method and system based on machine vision. Background Technique

[0002] The steel wire rope surface defect detection technology refers to a technology for detecting surface defects and damages of steel wire ropes, aiming to ensure the safety, reliability and service life of steel wire ropes. Steel wire ropes are widely used in hoisting, transportation and other industrial fields, and surface defects may affect the structural integrity and performance of steel wire ropes.

[0003] Existing steel wire rope surface defect detection technologies usually detect whether there are defects on the steel wire rope surface by analyzing the gray-scale changes on the surface of the steel wire rope. Since the steel wire rope is composed of many steel wires twisted together, the color changes on its surface are irregular, and the surface color of the steel wire rope is not uniform. The color of the steel wires inside the steel wire rope will change to a certain extent during the deformation process, but there is no rule. Therefore, the reliability of detecting whether there are defects on the steel wire rope surface by analyzing gray-scale changes is insufficient. For example, in the patent application with the publication number CN112270658A, a steel wire rope detection method for elevators based on machine vision is disclosed. This solution detects whether there are defects on the steel wire rope surface by analyzing the gray-scale values of the pixel points in the steel wire rope image. Since the color changes on the steel wire rope surface are irregular, the reliability of the analysis method based on gray-scale changes is insufficient. Existing steel wire rope surface defect detection technologies also have the problem that the analysis method is not reliable enough, resulting in easy misjudgment when detecting steel wire rope surface defects. Summary of the Invention

[0004] The present invention aims to solve at least one of the technical problems in the prior art to some extent. By collecting images of the steel wire rope surface to obtain a steel wire rope image, then using the OpenCV edge detection algorithm to extract the edges of the steel wire rope image to obtain an edge contour map, then repairing the boundary lines in the steel wire rope based on the edge contour map to obtain a repaired contour map, then performing region segmentation on the steel wire rope based on the repaired contour map to obtain the single-strand steel wire regions therein, extracting the linear features of the single-strand steel wire regions, selecting a steel wire rope without defects as a sample to analyze the normal range of the linear features, and finally detecting the surface defects of the steel wire rope based on the normal range of the linear features, so as to solve the problem that the existing steel wire rope surface defect detection technology has an unreliable analysis method, resulting in easy misjudgment when detecting steel wire rope surface defects.

[0005] To achieve the above object, in the first aspect, the present application provides a steel wire rope surface defect detection method based on machine vision, including the following steps: Collect images of the steel wire rope surface to obtain a steel wire rope image; Perform edge extraction on the wire rope image to obtain the edge contour map of the wire rope image. Based on the edge contour map, perform region segmentation on the wire rope to obtain the single-strand wire region therein; Extract the linear features of the single-strand wire region; Analyze the normal range of the linear features based on the linear features of different single-strand wire regions of the defect-free wire rope and detect the surface defects of the wire rope.

[0006] Furthermore, perform image acquisition on the surface of the wire rope. Obtaining the wire rope image includes the following sub-steps: Use a pure white background board as the shooting background, place the wire rope in front of the pure white background board and take a picture; Use a high-definition camera to take pictures of the wire rope inside the pure white background board. At the same time, by rotating the wire rope, take pictures of the same section of the wire rope for the first number of times, each time rotating 360° / N1, where N1 is the first number, to obtain the wire rope image.

[0007] Furthermore, perform edge extraction on the wire rope image to obtain the edge contour map of the wire rope image. Based on the edge contour map, perform region segmentation on the wire rope to obtain the single-strand wire region therein, including the following sub-steps: Perform edge extraction on the wire rope image through the OpenCV edge detection algorithm to obtain the edge contour map; Repair the boundary lines in the wire rope based on the edge contour map to obtain the repaired contour map; Perform region segmentation on the wire rope based on the repaired contour map to obtain the single-strand wire region therein.

[0008] Furthermore, repairing the boundary lines in the wire rope based on the edge contour map to obtain the repaired contour map includes the following sub-steps: Perform rectangular selection on the edge contour map to obtain the selected contour. The selected contour is rectangular, including a long side and a wide side. Connect the midpoints of the two wide sides, and mark the connected straight line as the contour central axis; Number the pixel points in the selected contour in the order from bottom left to top right, named pixel number, represented by the symbol P(n,m), where both n and m are non-zero natural numbers and (n,m) is the serial number of P. P(n,m) represents the pixel point at the nth pixel point in the horizontal direction and the mth pixel point in the vertical direction within the selected contour; The width of the contour central axis is the width of one pixel point. Mark the pure black pixel points in the selected contour as contour points, and obtain the contour points on the contour central axis, marked as direction pending points; For any direction of the point to be determined, obtain the contour points within the eight-neighborhood of the point to be determined in that direction, mark them as direction extension points, and then use the direction extension points as the points to be determined in the direction to extract direction extension points again until the first quantity of direction extension points is extracted; Obtain the pixel numbers of the direction extension points, extract the m among them, mark it as the vertical point position, obtain the maximum value and the minimum value of the vertical point position, mark them as the vertical highest point position and the vertical lowest point position respectively, and calculate the absolute value of the difference between the vertical lowest point position and the vertical highest point position, mark it as the vertical span; Compare the vertical span with the first span threshold. If the vertical span is greater than or equal to the first span threshold, output a boundary signal; otherwise, output a non-boundary signal; If a boundary signal is output, mark the line formed by the direction extension points as a boundary line, and mark the direction extension points on the boundary line as boundary points; Establish a plane rectangular coordinate system with n as the X-axis and m as the Y-axis, name it the boundary trend map, and for any boundary line, record the n and m of the pixel numbers of the boundary points therein into the boundary trend map; Conduct a linear regression analysis on the boundary trend map to obtain a boundary regression line, place the boundary regression line into the selected contour according to its relative position with the boundary points, mark all the pixel points passed by the boundary regression line as boundary points, and after analyzing each boundary line, obtain a repaired contour map.

[0009] Further, based on the repaired contour map, perform regional segmentation on the steel wire rope. The steps for obtaining the single-strand steel wire region include the following sub-steps: Perform regional division on the selected contour with the boundary regression line as the boundary, and mark the region between the two boundary regression lines as the single-strand steel wire region; The single-strand steel wire region does not include the boundary points passed by the boundary regression line; Obtain the number of contour points within the single-strand steel wire region, mark it as the pixel valid number, compare the pixel valid number with the second quantity. If the pixel valid number is less than the second quantity, output a signal of insufficient valid pixels; otherwise, output a signal of sufficient valid pixels; Judge whether both boundary regression lines of the single-strand steel wire region have intersections with the two long sides of the selected contour. If so, output a signal of complete region; otherwise, output a signal of incomplete region; If a signal of insufficient valid pixels or a signal of incomplete region is output, remove the single-strand steel wire region, and only retain the single-strand steel wire regions that simultaneously output a signal of sufficient valid pixels and a signal of complete region.

[0010] Further, the steps for extracting the linear features of the single-strand steel wire region include the following sub-steps: For any single-strand steel wire region, rename the contour points therein as region points; For any regional point, mark it as a target point, search for the regional points adjacent to the target point, mark them as neighborhood points, then use the neighborhood points as target points to search for neighborhood points in turn until there are no neighborhood points. At this time, the target points and neighborhood points obtained form a continuous line composed of pixel points, which is marked as a characteristic line. Then, analyze with any regional point as the target point, and the regional points that have been included in the characteristic line do not participate in the analysis. Repeat the execution until all regional points within the single-strand steel wire area are included in different characteristic lines; For any characteristic line, uniformly name the target points and neighborhood points therein as line points, obtain the pixel numbers of the line points, obtain the difference between the minimum value and the maximum value of n among them and add 1, which is marked as the horizontal span, and obtain the difference between the minimum value and the maximum value of m among them and add 1, which is marked as the vertical span; Number the characteristic lines in the order from top to bottom, and represent them by the symbol S i where i is a non-zero natural number and i is the serial number of S; Taking i as the horizontal axis, and taking the horizontal span and the vertical span as the vertical axes respectively to establish a plane rectangular coordinate system, which is named the horizontal characteristic map and the vertical characteristic map. Enter the horizontal span of S i into the horizontal characteristic map, and enter the vertical span into the vertical characteristic map. Mark the coordinate points in the horizontal characteristic map as horizontal characteristic points, and mark the coordinate points in the vertical characteristic map as vertical characteristic points; Connect adjacent horizontal characteristic points with a smooth curve to obtain a horizontal linear characteristic, and connect adjacent vertical characteristic points with a smooth curve to obtain a vertical linear characteristic. The horizontal linear characteristic and the vertical linear characteristic are collectively referred to as the linear characteristic.

[0011] Furthermore, analyzing the normal range of the linear characteristic based on the linear characteristics of different single-strand steel wire areas of the defect-free wire rope and detecting the surface defects of the wire rope include the following sub-steps: Select a defect-free wire rope as a sample to analyze the normal range of the linear characteristic; Detect the surface defects of the wire rope based on the normal range of the linear characteristic.

[0012] Furthermore, selecting a defect-free wire rope as a sample to analyze the normal range of the linear characteristic includes the following sub-steps: Select a defect-free wire rope as a sample, mark it as a sample wire, and mark the linear characteristic extracted from the sample wire as a sample characteristic; Summarize the sample characteristics analyzed from the wire rope images taken at different angles in the same area as the same-area characteristics, place the sample characteristics within the same-area characteristics in the same horizontal characteristic map and vertical characteristic map, and mark them as the horizontal analysis map and the vertical analysis map respectively; Mark the horizontal linear features in the horizontal analysis diagram as horizontal analysis lines, and mark the vertical linear features in the vertical analysis diagram as vertical analysis lines; For any value of the horizontal axis, obtain the maximum value of the difference in the vertical axis values in the horizontal analysis line, and mark it as the horizontal difference feature; for any value of the horizontal axis, obtain the maximum value of the difference in the vertical axis values in the vertical analysis line, and mark it as the vertical difference feature; For different same-region features, find the maximum values of the horizontal difference feature and the vertical difference feature, and mark them as the horizontal maximum difference and the vertical maximum difference respectively. The normal range includes the horizontal range and the vertical range. The horizontal range is [0, horizontal maximum difference], and the vertical range is [0, vertical maximum difference].

[0013] Further, detecting the surface defects of the steel wire rope based on the normal range of the linear features includes the following sub-steps: Mark the steel wire rope to be subjected to surface defect detection as the wire to be inspected; Extract the horizontal difference feature and the vertical difference feature of the same-region feature of the wire to be inspected, and mark them as the horizontal feature to be inspected and the vertical feature to be inspected respectively; If the horizontal feature to be inspected is within the horizontal range, output a horizontal normal signal, otherwise output a horizontal defect signal; if the vertical feature to be inspected is within the vertical range, output a vertical normal signal, otherwise output a vertical defect signal; If both a horizontal normal signal and a vertical normal signal are output simultaneously, mark that there are no defects on the surface of the wire to be inspected corresponding to the same-region feature, otherwise mark that there are defects on the surface of the wire to be inspected corresponding to the same-region feature.

[0014] In a second aspect, the present application provides a steel wire rope surface defect detection system based on machine vision, including an image acquisition module, a region division module, a feature extraction module, and a defect detection module; the image acquisition module, the region division module, and the defect detection module are respectively connected to the feature extraction module for data connection; The image acquisition module is used to collect an image of the surface of the steel wire rope to obtain a steel wire rope image; The region division module is used to extract the edge of the steel wire rope image to obtain an edge contour map of the steel wire rope image, and based on the edge contour map, perform region segmentation on the steel wire rope to obtain the single-strand wire region therein; The feature extraction module is used to extract the linear features of the single-strand wire region; The defect detection module is used to analyze the normal range of the linear features based on the linear features of different single-strand wire regions of the defect-free steel wire rope and detect the surface defects of the steel wire rope.

[0015] Advantages of the present invention: By collecting images of the surface of the wire rope, the wire rope images are obtained. Then, through the OpenCV edge detection algorithm, the edges of the wire rope images are extracted to obtain the edge contour map. Based on the edge contour map, the boundary lines in the wire rope are repaired to obtain the repaired contour map. Then, based on the repaired contour map, the wire rope is segmented into regions to obtain the single-strand wire regions. The advantage is that there are several steel wires in the same wire rope. When analyzing the whole, the amount of data is large. If there are small defects, the influence on the overall characteristics is small and it is not easy to accurately detect, especially for thicker wire ropes. And thicker wire ropes are composed of multiple thinner wire ropes. By dividing the regions of the thinner wire ropes and disassembling the wire rope, the influence of surface defects on the linear characteristics can be effectively improved, and the accuracy and effectiveness of the analysis of surface defects of the wire rope are improved; The present invention extracts the linear characteristics of the single-strand wire regions, selects the wire ropes without defects as samples to analyze the normal range of the linear characteristics, and finally detects the surface defects of the wire rope based on the normal range of the linear characteristics. The advantage is that the linear characteristics reflect the characteristics of each steel wire in the single-strand wire region. If there are surface defects such as fractures and cracks, there must be significant differences in their linear characteristics. And by analyzing the wire rope images at different angles of the same cross-section of the same wire rope, the influence of the color difference on the surface of the wire rope on defect detection can be reduced, because the steel wires in the same cross-section of the same wire rope have the same manufacturing process and the same stress when twisted into the wire rope, and the color difference on their surfaces is the smallest, improving the accuracy and rationality of the detection of surface defects of the wire rope. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is the principle block diagram of the system of the present invention; Figure 2 is the wire rope image of the system of the present invention; Figure 3 is the edge contour map of the system of the present invention; Figure 4 is the selected contour and the central axis of the contour of the system of the present invention; Figure 5 is the direction pending point of the system of the present invention; Figure 6 is the direction extension point of the system of the present invention; Figure 7 is the boundary trend map of the system of the present invention; Figure 8 is the repaired contour map of the system of the present invention; Figure 9 is the schematic diagram of the remaining single-strand wire regions of the system of the present invention; Figure 10A characteristic line within the single-strand steel wire area of the system of the present invention; Figure 11 A horizontal characteristic diagram of the system of the present invention; Figure 12 A horizontal analysis diagram of the system of the present invention; Figure 13 A step flow chart of the method of the present invention. Detailed implementation manners

[0017] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0018] Embodiment 1, please refer to Figure 1 As shown, the present application provides a steel wire rope surface defect detection system based on machine vision, including an image acquisition module, a region division module, a feature extraction module, and a defect detection module; the image acquisition module, the region division module, and the defect detection module are respectively connected to the feature extraction module for data connection; The image acquisition module is used to acquire an image of the surface of the steel wire rope to obtain a steel wire rope image; The image acquisition module is configured with an image acquisition strategy, and the image acquisition strategy includes: Using a pure white background board as the shooting background, placing the steel wire rope in front of the pure white background board for shooting; Please refer to Figure 2 As shown, the steel wire rope within the pure white background board is photographed by a high-definition camera. At the same time, by rotating the steel wire rope, the same section of the steel wire rope is photographed for the first number of times, each time rotating 360° / N1, where N1 is the first number, to obtain a steel wire rope image; In practical applications, the first number N1 is set to 3, that is, each time it rotates 120°, the same cross-section of the steel wire rope is comprehensively photographed to obtain 3 steel wire rope images. Then move the steel wire rope to photograph the next area, so as to obtain images of all surfaces of the steel wire rope. The steel wire rope images in this embodiment are as Figure 2 shown.

[0019] The region division module is used to extract the edges of the steel wire rope image to obtain an edge contour diagram of the steel wire rope image, and based on the edge contour diagram, the steel wire rope is regionally segmented to obtain the single-strand steel wire area therein; the region division module includes an edge detection unit, a contour repair unit, and a region division unit; Please refer to Figure 3As shown, the edge detection unit is used to extract the edges of the wire rope image through the OpenCV edge detection algorithm to obtain an edge contour map; In practical applications, the contour of the wire rope image is extracted through the existing OpenCV edge detection algorithm, and the edge contour map is obtained as Figure 3 shown; The contour repair unit is used to repair the boundary lines in the wire rope based on the edge contour map to obtain a repaired contour map; The contour repair unit is configured with a contour repair strategy, and the contour repair strategy includes: Please refer to Figure 4 shown, the edge contour map is rectangularly framed to obtain a framed contour. The framed contour is rectangular, including a long side and a wide side. The midpoints of the two wide sides are connected, and the connected straight line is marked as the contour central axis; In practical applications, the existing rectangular framing technology is used to frame the rectangular area where the wire rope is located, and the framed contour is obtained as Figure 4 shown, Figure 4 The contour central axis is also shown. For easy observation, Figure 4 the fade-in degree is adjusted to highlight the contour central axis; Since the wire rope as a whole can be roughly regarded as a cylinder and its surface is uneven, the boundaries of each thin wire rope are not easy to define on the upper and lower sides and cannot be shown in the edge contour map. Therefore, it needs to be repaired. The boundary at the contour central axis is the clearest and must exist. Therefore, the pixel points on the contour central axis are used as the starting point for analysis to find the pixel points belonging to the boundary line; The pixel points in the framed contour are numbered in the order from bottom left to top right, named pixel numbers, and represented by the symbol P(n,m). Among them, both n and m are non-zero natural numbers and (n,m) is the serial number of P. P(n,m) represents the pixel point that is the nth pixel point in the horizontal direction and the mth pixel point in the vertical direction within the framed contour; Please refer to Figure 5 shown, the width of the contour central axis is the width of one pixel point. The pure black pixel points in the framed contour are marked as contour points, and the contour points on the contour central axis are obtained and marked as direction pending points; Please refer to Figure 6 shown, for any direction pending point, the contour points within the eight-neighborhood of the direction pending point are obtained and marked as direction extension points. Then, the direction extension points are used as direction pending points again to extract direction extension points until the first quantity of direction extension points is extracted; In practical applications, in this embodiment, Figure 5Taking the direction pending point in it as an example, the process of repairing the boundary line is illustrated. The analysis processes of the remaining direction pending points will not be specifically explained in this embodiment; the eight-neighborhood refers to the contour points adjacent to the direction pending point in the upper left, directly above, upper right, directly left, directly right, lower left, directly below, and lower right directions of the direction pending point. The contour points are the Figure 5 pure black pixel points in it. Taking the Figure 5 gray direction pending point in it as an example, among which there are contour points in both the upper left and lower right in the eight-neighborhood. Mark them as direction extension points, and then take the direction extension points as direction pending points. Taking the upper left direction extension point as an example, take it as a direction pending point and search for the eight-neighborhood again. Among them, there are contour points in both the directly left and directly right directions, and the direction pending points and direction extension points do not participate in the analysis. Therefore, only the direction extension points in the directly left direction are retained, and so on, continuously expanding the direction extension points; in this embodiment, the first quantity is set to 20. The meaning of setting the first quantity is to find a certain number of contour points and judge whether the contour points belong to the boundary according to their distribution characteristics. From Figure 2 it can be seen that the boundary of each thin steel wire rope is oblique, so the span in its longitudinal direction is relatively large, and the boundary points can be accurately found by this judgment; for the Figure 5 direction pending point, the analyzed direction extension points are as shown in Figure 6 ; Figure 6 the gray contour points in it are the direction extension points; Obtain the pixel numbers of the direction extension points, extract the m among them, mark it as the vertical point position, obtain the maximum value and minimum value of the vertical point position, and mark them as the vertical highest point position and vertical lowest point position respectively. Calculate the absolute value of the difference between the vertical lowest point position and the vertical highest point position, and mark it as the vertical span; Compare the vertical span with the first span threshold. If the vertical span is greater than or equal to the first span threshold, output a boundary signal; otherwise, output a non-boundary signal; If a boundary signal is output, mark the line composed of the direction extension points as the boundary line, and mark the direction extension points on the boundary line as boundary points; Please refer to Figure 7 As shown, establish a plane rectangular coordinate system with n as the X-axis and m as the Y-axis, named the boundary trend map. For any boundary line, enter the n and m of the pixel numbers of the boundary points in it into the boundary trend map; Please refer to Figure 8 As shown, perform a linear regression analysis on the boundary trend map to obtain a boundary regression line. Place the boundary regression line into the framed contour according to its relative position with the boundary points, mark all the pixel points passed by the boundary regression line as boundary points. After analyzing each boundary line, obtain the repaired contour map; In practical applications, the first span threshold is set to 4. The highest vertical point is obtained as 57 and the lowest vertical point is 51. The vertical span is calculated to be 6. By comparison, the vertical span is greater than the first span threshold, and a boundary signal is output, that is Figure 6 The direction extension points in it form a boundary line, and the boundary trend map is constructed as shown in Figure 7 shown, Figure 7 also shows the boundary regression line. Then all the boundary regression lines are placed in the selected outline as shown in Figure 8 shown. When placing them in the selected outline, extend the boundary regression line until both ends intersect with the long side or the wide side; The area division unit is used to divide the wire rope based on the repaired contour map to obtain the single-strand wire area therein; The area division unit is configured with an area division strategy, and the area division strategy includes: Divide the selected outline with the boundary regression line as the boundary, and mark the area between the two boundary regression lines as the single-strand wire area; The single-strand wire area does not include the boundary points passed by the boundary regression line; Obtain the number of inner contour points in the single-strand wire area, mark it as the pixel valid number, compare the pixel valid number with the second number. If the pixel valid number is less than the second number, output a signal of insufficient valid pixels, otherwise output a signal of sufficient valid pixels; Judge whether both boundary regression lines of the single-strand wire area intersect with the two long sides of the selected outline. If so, output a signal of complete area, otherwise output a signal of incomplete area; Please refer to Figure 9 shown. If a signal of insufficient valid pixels or a signal of incomplete area is output, the single-strand wire area will be excluded, and only the single-strand wire areas that output both a signal of sufficient valid pixels and a signal of complete area will be retained; In practical applications, the second number is set to 100. Usually, the number of contour points in the thin wire rope is much larger than the second number, while the number of contour points in the interval between two thin wire ropes is much less than 100. And only when both boundary regression lines of the single-strand wire area intersect with the two long sides of the selected outline and do not intersect with the wide side, can its complete linear feature be extracted. After exclusion, the remaining single-strand wire areas are as shown in Figure 9 shown, Figure 9 The shaded part in it is the remaining single-strand wire area; In this embodiment, for the convenience of showing the analysis process through images, only a part of the wire rope image is selected for display. Therefore, only one single-strand wire area can be extracted. In actual use, the long side of the selected contour map is longer, and more single-strand wire areas can be extracted.

[0020] The feature extraction module is used to extract the linear features of the single-strand steel wire area; The feature extraction module is configured with feature extraction strategies, which include: For any single-strand steel wire area, rename the contour points therein as area points; See also Figure 10 As shown, for any regional point, mark it as a target point, find the regional points adjacent to the target point, mark them as neighborhood points, and then use the neighborhood points as target points to find the neighborhood points in turn until there are no neighborhood points. At this time, the target point and the neighborhood points obtained form a line composed of continuous pixel points, which is marked as a feature line. Then, use any regional point as the target point for analysis, and the regional points that have been included in the feature line will not participate in the analysis. Repeat the process until all regional points in the single-strand steel wire area are included in different feature lines. For any feature line, the target point and the neighborhood point are uniformly named as line points, the pixel number of the line point is obtained, the difference between the minimum and maximum values ​​of n is obtained and added by one, marked as the horizontal span, and the difference between the minimum and maximum values ​​of m is obtained and added by one, marked as the vertical span; In practical applications, the purpose of searching for neighboring points is to combine adjacent area points into a characteristic line. This will not be described in detail in this embodiment. In this embodiment, a characteristic line in a single-strand steel wire area is as follows: Figure 10 As shown in the figure, the lines composed of gray pixels are the feature lines. Figure 10 It is easy to know that the horizontal span is 68 and the vertical span is 4. In actual use, the feature lines with both horizontal and vertical spans less than or equal to 20 need to be removed. This step is to remove the shorter feature lines. The shorter feature lines usually have no reference value, so they need to be removed. The characteristic lines are numbered from top to bottom, with the symbol S i Represents, where i is a non-zero natural number and i is the sequence number of S; See also Figure 11 As shown in the figure, with i as the horizontal axis, and the horizontal span and vertical span as the vertical axis, a plane rectangular coordinate system is established, named as the horizontal feature map and the vertical feature map, and S i The horizontal span is recorded into the horizontal feature map, and the vertical span is recorded into the vertical feature map, and the coordinate points in the horizontal feature map are marked as horizontal feature points, and the coordinate points in the vertical feature map are marked as vertical feature points; Connecting adjacent transverse feature points through a smooth curve to obtain transverse linear features, and connecting adjacent longitudinal feature points through a smooth curve to obtain longitudinal linear features. The transverse linear features and longitudinal linear features are collectively referred to as linear features. In practical applications, since the analysis processes of the horizontal feature map and the vertical feature map are the same and only represent features in different directions, this embodiment only takes the horizontal feature map as an example for illustration. The constructed horizontal feature map is as follows Figure 11 As shown, since the coordinate points are too dense to be specifically displayed, Figure 11 only the horizontal linear features are shown in

[0021] The defect detection module is used to analyze the normal range of the linear features based on the linear features of different single-strand wire areas of the defect-free wire rope and detect the surface defects of the wire rope; the defect detection module includes a normal range analysis unit and a surface defect detection unit; The normal range analysis unit is used to select the defect-free wire rope as a sample to analyze the normal range of the linear features; The normal range analysis unit is configured with a normal range analysis strategy, and the normal range analysis strategy includes: Select the defect-free wire rope as a sample, mark it as the sample wire, and mark the linear features extracted from the sample wire as the sample features; Please refer to Figure 12 As shown, the sample features analyzed from the wire rope images taken at different angles in the same area are summarized as the same-area features. The sample features within the same-area features are placed in the same horizontal feature map and vertical feature map, and are respectively marked as the horizontal analysis map and the vertical analysis map; Mark the horizontal linear features in the horizontal analysis map as the horizontal analysis lines, and mark the vertical linear features in the vertical analysis map as the vertical analysis lines; For any value of the horizontal axis, obtain the maximum value of the difference in the vertical axis values in the horizontal analysis line, and mark it as the horizontal difference feature; for any value of the horizontal axis, obtain the maximum value of the difference in the vertical axis values in the vertical analysis line, and mark it as the vertical difference feature; For different same-area features, find the maximum values of the horizontal difference feature and the vertical difference feature, and mark them as the horizontal maximum difference and the vertical maximum difference respectively. The normal range includes the horizontal range and the vertical range. The horizontal range is [0, horizontal maximum difference], and the vertical range is [0, vertical maximum difference]; In practical applications, there are wire rope images at three angles for the same cross-section of the wire rope. Integrate their sample features into the same-area features. Taking the horizontal analysis map as an example, the integrated horizontal analysis map is as follows Figure 12 As shown, the obtained horizontal difference feature is 19. Then, based on the analysis of all the same-area features of all the sample wires, find the maximum value of the horizontal difference feature among them, and obtain the horizontal maximum difference as 21, and obtain the horizontal range as [0, 21]. Similarly, the vertical maximum difference is [0, 2]; The surface defect detection unit is used to detect the surface defects of the wire rope based on the normal range of the linear features; The surface defect detection unit is configured with a surface defect detection strategy, and the surface defect detection strategy includes: Mark the steel wire rope that needs to be detected for surface defects as the wire to be detected; Extract the lateral difference feature and the longitudinal difference feature of the same-region feature of the wire to be detected, and mark them as the lateral feature to be detected and the longitudinal feature to be detected respectively; If the lateral feature to be detected is within the lateral range, output a lateral normal signal, otherwise output a lateral defect signal; if the longitudinal feature to be detected is within the longitudinal range, output a longitudinal normal signal, otherwise output a longitudinal defect signal; If both the lateral normal signal and the longitudinal normal signal are output simultaneously, mark that there are no defects on the surface corresponding to the same-region feature of the wire to be detected, otherwise mark that there are defects on the surface corresponding to the same-region feature of the wire to be detected; In practical applications, the detection process of the surface defects of the steel wire rope in the surface defect detection unit is clear enough, and no further explanation will be given in this embodiment.

[0022] Embodiment 2, please refer to Figure 13 As shown, the present application provides a method for detecting surface defects of a steel wire rope based on machine vision, including the following steps: Step S1, collect an image of the surface of the steel wire rope to obtain a steel wire rope image; Step S1 includes the following sub-steps: Step S101, use a pure white background board as the shooting background, and place the steel wire rope in front of the pure white background board for shooting; Step S102, shoot the steel wire rope in the pure white background board through a high-definition camera, and at the same time rotate the steel wire rope to take the first number of shots of the same section of the steel wire rope, each time rotating 360° / N1, where N1 is the first number, to obtain a steel wire rope image; Step S2, extract the edges of the steel wire rope image to obtain an edge contour map of the steel wire rope image, and perform region segmentation on the steel wire rope based on the edge contour map to obtain the single-strand wire region therein; Step S2 includes the following sub-steps: Step S201, extract the edges of the steel wire rope image through the OpenCV edge detection algorithm to obtain an edge contour map; Step S202, repair the boundary lines in the steel wire rope based on the edge contour map to obtain a repaired contour map; Step S202 includes the following sub-steps: Step S2021, perform rectangular selection on the edge contour map to obtain a selected contour. The selected contour is a rectangle, which includes a long side and a wide side. Connect the midpoints of the two wide sides, and mark the connected straight line as the contour central axis; In step S2022, the pixel points in the selected contour are numbered in the order from the lower left to the upper right, named pixel numbers, and represented by the symbol P(n,m), where both n and m are non-zero natural numbers and (n,m) is the serial number of P. P(n,m) represents the pixel point that is the nth pixel point in the horizontal direction and the mth pixel point in the vertical direction within the selected contour. In step S2023, the width of the contour central axis is the width of one pixel point. The pure black pixel points in the selected contour are marked as contour points, and the contour points on the contour central axis are obtained and marked as direction pending points. In step S2024, for any direction pending point, the contour points within the eight-neighborhood of the direction pending point are obtained and marked as direction extension points. Then, using the direction extension points as direction pending points, direction extension points are extracted again until the first quantity of direction extension points is extracted. In step S2025, the pixel numbers of the direction extension points are obtained, and the m among them is extracted and marked as the vertical position. The maximum value and the minimum value of the vertical position are obtained and marked as the vertical highest position and the vertical lowest position respectively. The absolute value of the difference between the vertical lowest position and the vertical highest position is calculated and marked as the vertical span. In step S2026, the vertical span is compared with the first span threshold. If the vertical span is greater than or equal to the first span threshold, a boundary signal is output; otherwise, a non-boundary signal is output. In step S2027, if a boundary signal is output, the line formed by the direction extension points is marked as a boundary line, and the direction extension points on the boundary line are marked as boundary points. In step S2028, a plane rectangular coordinate system is established with n as the X-axis and m as the Y-axis, named the boundary trend map. For any boundary line, the n and m of the pixel numbers of the boundary points among them are entered into the boundary trend map. In step S2029, linear regression analysis is performed on the boundary trend map to obtain a boundary regression line. The boundary regression line is placed into the selected contour according to its relative position with respect to the boundary points, and all the pixel points passed by the boundary regression line are marked as boundary points. After analyzing each boundary line, a repaired contour map is obtained. In step S203, based on the repaired contour map, the steel wire rope is regionally segmented to obtain the single-strand steel wire regions therein. Step S203 includes the following sub-steps: In step S2031, the selected contour is regionally divided with the boundary regression line as the boundary, and the region between two boundary regression lines is marked as the single-strand steel wire region. In step S2032, the single-strand steel wire region does not include the boundary points passed by the boundary regression line. Step S2033: Obtain the number of inner contour points in the single-strand steel wire area, mark it as the pixel valid number, compare the pixel valid number with the second number. If the pixel valid number is less than the second number, output a signal indicating insufficient valid pixels; otherwise, output a signal indicating sufficient valid pixels. Step S2034: Determine whether both of the two boundary regression lines of the single-strand steel wire area have intersections with the two long sides of the framed contour. If so, output a signal indicating a complete area; otherwise, output a signal indicating an incomplete area. Step S2035: If a signal indicating insufficient valid pixels or an incomplete area is output, eliminate the single-strand steel wire area, and only retain the single-strand steel wire areas that simultaneously output a signal indicating sufficient valid pixels and a signal indicating a complete area. Step S3: Extract the linear features of the single-strand steel wire area. Step S3 includes the following sub-steps: Step S301: For any single-strand steel wire area, rename the contour points therein as area points. Step S302: For any area point, mark it as a target point, find the area points adjacent to the target point, mark them as neighborhood points, and then take the neighborhood points as target points to sequentially find neighborhood points until there are no more neighborhood points. At this time, the obtained target points and neighborhood points form a line composed of continuous pixel points, marked as a feature line. Then, analyze with any area point as the target point. The area points that have been included in the feature line do not participate in the analysis. Repeat the execution until all area points in the single-strand steel wire area are included in different feature lines. Step S303: For any feature line, uniformly name the target points and neighborhood points therein as line points, obtain the pixel numbers of the line points, obtain the difference between the minimum value and the maximum value of n therein and add 1, mark it as the horizontal span, obtain the difference between the minimum value and the maximum value of m therein and add 1, mark it as the vertical span. Step S304: Number the feature lines in the order from top to bottom, represented by the symbol S i where i is a non-zero natural number and i is the serial number of S. Step S305: Establish a plane rectangular coordinate system with i as the horizontal axis and the horizontal span and the vertical span as the vertical axes respectively, named the horizontal feature map and the vertical feature map. Enter the horizontal span of S i into the horizontal feature map and the vertical span into the vertical feature map. Mark the coordinate points in the horizontal feature map as horizontal feature points and the coordinate points in the vertical feature map as vertical feature points. Step S306: Connect adjacent horizontal feature points with a smooth curve to obtain the horizontal linear feature, connect adjacent vertical feature points with a smooth curve to obtain the vertical linear feature. The horizontal linear feature and the vertical linear feature are collectively called linear features. Step S4: Analyze the normal range of the linear features based on the linear features of different single-strand wire areas of the defect-free wire rope and detect the surface defects of the wire rope. Step S4 includes the following sub-steps: Step S401: Select a defect-free wire rope as a sample to analyze the normal range of the linear features. Step S401 includes the following sub-steps: Step S4011: Select a defect-free wire rope as a sample, mark it as the sample wire, and mark the linear features extracted from the sample wire as the sample features. Step S4012: Summarize the sample features analyzed from the wire rope images taken at different angles in the same area as the same-area features, place the sample features within the same-area features in the same horizontal feature map and vertical feature map, and mark them as the horizontal analysis map and vertical analysis map respectively. Step S4013: Mark the horizontal linear features in the horizontal analysis map as the horizontal analysis lines, and mark the vertical linear features in the vertical analysis map as the vertical analysis lines. Step S4014: For any value of the horizontal axis, obtain the maximum value of the difference in the vertical axis values in the horizontal analysis line, and mark it as the horizontal difference feature; for any value of the horizontal axis, obtain the maximum value of the difference in the vertical axis values in the vertical analysis line, and mark it as the vertical difference feature. Step S4015: For different same-area features, find the maximum values of the horizontal difference feature and the vertical difference feature, and mark them as the horizontal maximum difference and the vertical maximum difference respectively. The normal range includes the horizontal range and the vertical range. The horizontal range is [0, horizontal maximum difference], and the vertical range is [0, vertical maximum difference]. Step S402: Detect the surface defects of the wire rope based on the normal range of the linear features. Step S402 includes the following sub-steps: Step S4021: Mark the wire rope that needs to be detected for surface defects as the wire to be inspected. Step S4022: Extract the horizontal difference feature and the vertical difference feature of the same-area features of the wire to be inspected, and mark them as the horizontal feature to be inspected and the vertical feature to be inspected respectively. Step S4023: If the horizontal feature to be inspected is within the horizontal range, output a horizontal normal signal; otherwise, output a horizontal defect signal; if the vertical feature to be inspected is within the vertical range, output a vertical normal signal; otherwise, output a vertical defect signal. Step S4024: If both the horizontal normal signal and the vertical normal signal are output, mark that there are no defects on the surface of the wire to be inspected corresponding to the same-area features; otherwise, mark that there are defects on the surface of the wire to be inspected corresponding to the same-area features.

[0023] Embodiment 3. The present application provides an electronic device, which may include: a processor, a communication interface, a memory, and a communication bus. Among them, the processor, the communication interface, and the memory complete mutual communication through the communication bus. The memory stores computer-readable instructions, and the processor can call the instructions in the memory. When the computer-readable instructions are executed by the processor, the steps in the method for detecting surface defects of wire ropes based on machine vision are run to achieve the following functions: performing image acquisition on the surface of the wire rope to obtain a wire rope image; performing edge extraction on the wire rope image to obtain an edge contour map of the wire rope image, and performing region segmentation on the wire rope based on the edge contour map to obtain the single-strand wire region therein; extracting the linear features of the single-strand wire region; analyzing the normal range of the linear features based on the linear features of different single-strand wire regions of the defect-free wire rope and detecting the surface defects of the wire rope.

[0024] In addition, when the logical instructions in the above-mentioned memory are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0025] Embodiment 4. The present application also provides a computer-readable storage medium. The present application provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps in the method for detecting surface defects of wire ropes based on machine vision as described above are run to achieve the following functions: performing image acquisition on the surface of the wire rope to obtain a wire rope image; performing edge extraction on the wire rope image to obtain an edge contour map of the wire rope image, and performing region segmentation on the wire rope based on the edge contour map to obtain the single-strand wire region therein; extracting the linear features of the single-strand wire region; analyzing the normal range of the linear features based on the linear features of different single-strand wire regions of the defect-free wire rope and detecting the surface defects of the wire rope.

[0026] Through the description of the above embodiments, the embodiments of the present invention can be provided as a method, a system or a computer program product. Based on such an understanding, the above technical solution, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0027] In the embodiments provided in the present application, it should be understood that the disclosed system or method can be implemented in other ways. The above-described embodiments are merely illustrative. For example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation. For another example, multiple modules or units can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some communication interfaces. The indirect coupling or communication connection of systems, modules and units can be electrical, mechanical or other forms.

[0028] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A wire rope surface defect detection method based on machine vision, characterized in that: The steps include: Capturing images of the surface of the steel wire rope to obtain steel wire rope images; Extract the edge of the wire rope image to obtain an edge contour map of the wire rope image, segment the wire rope region based on the edge contour map to obtain a single strand steel wire region; Extract the linear features of the single-strand steel wire area; Based on the linear features of different single-strand steel wire areas of a defect-free steel wire rope, the normal range of the linear features is analyzed and the surface defects of the steel wire rope are detected.

2. The method for detecting surface defects of a steel wire rope based on machine vision according to claim 1, characterized in that: The image acquisition of the wire rope surface includes the following sub-steps: Use a pure white background as the shooting background, and place the wire rope in front of the pure white background for shooting; The steel wire rope on the pure white background is photographed by a high-definition camera. At the same time, the steel wire rope is rotated and the same section of the steel wire rope is photographed for the first time. Each rotation is 360° / N1, where N1 is the first time, to obtain the steel wire rope image.

3. The method for detecting surface defects of a steel wire rope based on machine vision according to claim 2, characterized in that: The edge of the wire rope image is extracted to obtain an edge contour map of the wire rope image, and the wire rope is segmented based on the edge contour map to obtain a single strand of steel wire region, which includes the following sub-steps: The edge of the wire rope image is extracted using the OpenCV edge detection algorithm to obtain the edge contour map; The boundary lines in the wire rope are repaired based on the edge contour map to obtain a repair contour map; The steel wire rope is segmented based on the repair contour map to obtain the single-strand steel wire area.

4. The method for detecting surface defects of a steel wire rope based on machine vision according to claim 3 is characterized in that: The boundary lines in the wire rope are repaired based on the edge contour map, and obtaining the repair contour map includes the following sub-steps: Performing a rectangular frame selection on the edge contour image to obtain a frame selection contour, wherein the frame selection contour is a rectangle including a long side and a wide side, connecting the midpoints of the two wide sides, and marking the connecting straight line as the contour centerline; The pixels in the frame selection outline are numbered from the lower left to the upper right, named as pixel numbers, and represented by the symbol P(n,m), where n and m are both non-zero natural numbers and (n,m) is the sequence number of P, and P(n,m) represents the pixel that is the nth pixel in the horizontal direction and the mth pixel in the vertical direction in the frame selection outline; The width of the central axis of the contour is the width of a pixel, the pure black pixel points in the framed contour are marked as contour points, and the contour points on the central axis of the contour are obtained and marked as points to be determined in terms of direction; For any undetermined direction point, contour points in the eight neighborhoods of the undetermined direction point are obtained and marked as direction extension points, and then the direction extension points are extracted again using the direction extension points as the undetermined direction points until the first number of direction extension points are extracted; Get the pixel number of the direction extension point, extract m from it, mark it as the vertical point, get the maximum and minimum values ​​of the vertical point, mark them as the vertical highest point and the vertical lowest point respectively, calculate the absolute value of the difference between the vertical lowest point and the vertical highest point, and mark it as the vertical span; Compare the vertical span with the first span threshold, if the vertical span is greater than or equal to the first span threshold, output a boundary signal, otherwise output a non-boundary signal; If a boundary signal is output, the line formed by the direction extension points is marked as a boundary line, and the direction extension points on the boundary line are marked as boundary points; A plane rectangular coordinate system is established with n as the X-axis and m as the Y-axis, which is named a boundary trend map. For any boundary line, the pixel numbers n and m of the boundary points therein are entered into the boundary trend map; A linear regression analysis is performed on the boundary trend map to obtain a boundary regression line, which is placed in the frame selection contour according to its relative position to the boundary point. All pixel points passed by the boundary regression line are marked as boundary points. After analyzing each boundary line, a repair contour map is obtained.

5. The method for detecting surface defects of a steel wire rope based on machine vision according to claim 4, characterized in that: The steel wire rope is segmented based on the repair contour map to obtain the single-strand steel wire region, including the following sub-steps: The selected outline is divided into regions using the boundary regression line as the boundary, and the region between the two boundary regression lines is marked as a single-strand steel wire region; The single-strand steel wire region does not include the boundary points through which the boundary regression line passes; Obtain the number of contour points in the single-strand steel wire area, mark it as the effective number of pixels, compare the effective number of pixels with the second number, and if the effective number of pixels is less than the second number, output an effective pixel deficiency signal, otherwise output an effective pixel sufficient signal; Determine whether the two boundary regression lines of the single-strand steel wire area have intersections with the two long sides of the frame selection contour. If so, output a complete area signal, otherwise output an incomplete area signal; If the output signal is insufficient for effective pixels or an incomplete area signal, the single-strand steel wire area is removed, and only the single-strand steel wire area that outputs both sufficient effective pixel signals and complete area signals is retained.

6. The method for detecting surface defects of a steel wire rope based on machine vision according to claim 5, characterized in that: Extracting the linear features of a single steel wire region includes the following sub-steps: For any single-strand steel wire area, rename the contour points therein as area points; For any regional point, mark it as a target point, find the regional points adjacent to the target point, mark them as neighborhood points, and then use the neighborhood point as the target point to find the neighborhood points in turn until there is no neighborhood point. At this time, the target point and the neighborhood point form a line composed of continuous pixel points, which is marked as a feature line. Then, use any regional point as the target point for analysis, and the regional points that have been included in the feature line will not participate in the analysis. Repeat the process until all regional points in the single-strand steel wire area are included in different feature lines. For any feature line, the target point and the neighborhood point are uniformly named as line points, the pixel number of the line point is obtained, the difference between the minimum and maximum values ​​of n is obtained and added by one, marked as the horizontal span, and the difference between the minimum and maximum values ​​of m is obtained and added by one, marked as the vertical span; The characteristic lines are numbered from top to bottom, with the symbol S i Represents, where i is a non-zero natural number and i is the sequence number of S; With i as the horizontal axis, and the horizontal span and vertical span as the vertical axis, a plane rectangular coordinate system is established, named the horizontal feature map and the vertical feature map. i The horizontal span is recorded into the horizontal feature map, and the vertical span is recorded into the vertical feature map, and the coordinate points in the horizontal feature map are marked as horizontal feature points, and the coordinate points in the vertical feature map are marked as vertical feature points; Adjacent transverse feature points are connected by a smooth curve to obtain transverse linear features, and adjacent longitudinal feature points are connected by a smooth curve to obtain longitudinal linear features. The transverse linear features and longitudinal linear features are collectively referred to as linear features.

7. The method for detecting surface defects of a steel wire rope based on machine vision according to claim 6, characterized in that: Analyzing the normal range of linear features based on the linear features of different single-strand steel wire regions of a defect-free steel wire rope and detecting surface defects of the steel wire rope includes the following sub-steps: Select defect-free wire ropes as samples to analyze the normal range of linear characteristics; Surface defects of wire ropes are detected based on the normal range of linear features.

8. The method for detecting surface defects of a steel wire rope based on machine vision according to claim 7, characterized in that: The normal range of linear characteristics of selected defect-free wire ropes as samples for analysis includes the following sub-steps: Select a defect-free steel wire rope as a sample, mark it as a sample steel wire, and mark the linear features extracted from the sample steel wire as sample features; The sample features obtained by analyzing the wire rope images taken at different angles in the same area are summarized as the same-area features, and the sample features in the same-area features are placed in the same horizontal feature map and vertical feature map, which are marked as horizontal analysis map and vertical analysis map respectively; Mark the horizontal linear features in the horizontal analysis graph as horizontal analysis lines, and mark the vertical linear features in the vertical analysis graph as vertical analysis lines; For any value of the horizontal axis, obtain the maximum value of the difference between the vertical axis values ​​in the horizontal analysis line, and mark it as the horizontal difference feature; For any value of the horizontal axis, obtain the maximum value of the difference between the vertical axis values ​​in the vertical analysis line, and mark it as the vertical difference feature; For different same-area features, find the maximum values ​​of the horizontal difference feature and the vertical difference feature, and mark them as the horizontal maximum difference and the vertical maximum difference respectively. The normal range includes the horizontal range and the vertical range. The horizontal range is [0, horizontal maximum difference], and the vertical range is [0, vertical maximum difference].

9. The method for detecting surface defects of a steel wire rope based on machine vision according to claim 8, characterized in that: The detection of surface defects of wire ropes based on the normal range of linear features includes the following sub-steps: Mark the steel wire rope that needs to be inspected for surface defects as the steel wire to be inspected; Extract the transverse difference feature and the longitudinal difference feature of the same area feature of the steel wire to be inspected, and mark them as the transverse feature to be inspected and the longitudinal feature to be inspected respectively; If the horizontal feature to be inspected is within the horizontal range, a horizontal normal signal is output, otherwise a horizontal defect signal is output; if the vertical feature to be inspected is within the vertical range, a vertical normal signal is output, otherwise a vertical defect signal is output; If the transverse normal signal and the longitudinal normal signal are output simultaneously, it is marked that there is no defect on the surface of the steel wire to be inspected corresponding to the characteristics in the same area; otherwise, it is marked that there is a defect on the surface of the steel wire to be inspected corresponding to the characteristics in the same area.

10. A wire rope surface defect detection system based on machine vision, used to implement the wire rope surface defect detection method based on machine vision according to any one of claims 1 to 9, characterized in that: It includes an image acquisition module, a region division module, a feature extraction module and a defect detection module; the image acquisition module, the region division module and the defect detection module are respectively connected to the feature extraction module data; The image acquisition module is used to acquire images of the surface of the steel wire rope to obtain steel wire rope images; The region division module is used to extract the edge of the steel wire rope image to obtain an edge contour map of the steel wire rope image, and to segment the steel wire rope based on the edge contour map to obtain a single strand steel wire region therein; The feature extraction module is used to extract the linear features of the single-strand steel wire area; The defect detection module is used to analyze the normal range of linear features based on the linear features of different single-strand steel wire regions of a defect-free steel wire rope and detect surface defects of the steel wire rope.

Images