Armored cable sheath defect detection method and system based on computer vision

Abstract

The present invention relates to the technical field of image feature analysis, and in particular to a method and system for detecting armored cable sheath defects based on computer vision. The method performs segmentation according to the edge of a raised area to obtain a local cable area. The crack edge is screened out according to the characteristics of the crack edge and the degree of cracking is evaluated. The grayscale change edge is screened out and the grayscale change value is determined, and the degree of looseness is determined in combination with the damage change value characterized by the crack edge. The degree of defect can be accurately quantified by combining the degree of looseness and the degree of cracking. The present invention screens out multiple types of edges through the characteristics of different types of edges and performs corresponding processing, thereby avoiding the influence of complex edges on defect detection and realizing accurate quantification of the degree of defects in the armored cable protective sheath.

Description

Technical Field

[0001] The present invention relates to the technical field of image feature analysis, and in particular to a method and system for detecting armored cable sheath defects based on computer vision. Background Art

[0002] The armored cable protective cover is an outer covering used to protect the internal structure of the cable and the electrical conductor. It can provide mechanical protection, waterproofing, corrosion resistance, and electromagnetic shielding. The inner layer of the protective cover is the armored cable. The armored cable is a flexible tube or steel belt made of a metal material wrapped around the cable to enhance the mechanical strength of the cable. During transportation and use, the armored cable protective cover may be damaged on the surface due to factors such as extrusion, stretching, and impact. These damages will reduce the protective ability of the protective cover. Therefore, during the production and use of the armored cable protective cover, it is necessary to perform defect detection on it to quickly and effectively identify defects and quantify the degree of defects.

[0003] In the existing technology, computer vision technology can be used to check product defects through image information. However, since the armor layer of the armored cable has obviously distributed raised areas, the edge information on the cable is relatively complex. The defective edge information generated by the protective cover will be confused with other edge information. Directly using edge information for defect detection cannot obtain accurate defect quantification results. Summary of the invention

[0004] In order to solve the technical problem that the existing defect edge detection method cannot accurately detect defects in the armored cable protective sheath due to the rich edge information on the armored cable surface, the purpose of the present invention is to provide an armored cable sheath defect detection method and system based on computer vision. The technical solution adopted is as follows:

[0005] The present invention proposes a method for detecting armored cable sheath defects based on computer vision, the method comprising:

[0006] Acquire a surface grayscale image of an armored cable installed with a protective cover; in the surface grayscale image, obtain a number of local cable areas according to the edges of the raised areas on the protective cover;

[0007] Screening out crack edges according to the grayscale value and distribution direction of the edges in the local cable area, and obtaining the degree of cracking in the local cable area according to the morphological distribution and grayscale value of the crack edges;

[0008] Screen out the grayscale change edges in the local cable area; the grayscale change edges are non-crack edges extending into the local cable area based on the two side boundaries corresponding to the cable direction; obtain the regional grayscale change value according to the grayscale difference between the grayscale change edge and the two side boundaries, and obtain the damage change value according to the relative distance between the distribution area of ​​the crack edge and the two side boundaries; obtain the looseness degree of the local cable area according to the non-correlation between the regional grayscale change value and the damage change value, and the regional grayscale change value;

[0009] The defect degree of the local cable area is obtained according to the crack degree and the looseness degree.

[0010] Furthermore, the method for obtaining the local cable area includes:

[0011] Performing straight line fitting on each edge in the surface grayscale image, and obtaining the fitting deviation between each edge and the fitting straight line; taking the angle between the fitting straight line and the direction perpendicular to the cable direction as the deviation angle;

[0012] Obtaining the preferred degree of each edge according to the fitting deviation and the deviation angle; screening out suspected convex region edges according to the preferred degree; taking the distance between adjacent suspected convex region edges as the interval distance; taking the mode of the interval distance as the convex region interval, and taking the adjacent suspected convex region edges whose interval distance is equal to the convex region interval as a set of true convex region edge pairs;

[0013] Taking any real raised area edge pair as the center, the cable area in the surface grayscale image is divided according to the raised area intervals to obtain a number of local cable areas.

[0014] Furthermore, the method for obtaining the degree of preference includes:

[0015] The fitting deviation and the sum of the deviation angles are negatively correlated to obtain the degree of optimization.

[0016] Furthermore, the method for obtaining the crack edge includes:

[0017] In the local cable area, the difference between the average grayscale value of each edge and the average grayscale value of the local cable area is taken as the crack visibility; if the crack visibility of the edge is greater than the preset crack visibility threshold and the deviation angle is less than the preset angle threshold, the edge is taken as the crack edge.

[0018] Furthermore, the method for obtaining the degree of cracking includes:

[0019] In the local cable area, the crack area of ​​each crack edge is obtained according to the width and length of each crack edge; the local aggregation degree of each crack edge is obtained according to the average distance between each crack edge and other crack edges in a preset neighborhood range; the local aggregation degree of each crack edge is used as a weight to weight the crack conspicuity degree of each crack edge to obtain the crack degree;

[0020] The crack degree and the crack area of ​​each crack edge are multiplied to obtain the initial crack degree of each crack edge; the accumulated value of the initial crack degrees of all crack edges in the local cable area is normalized to obtain the crack degree.

[0021] Furthermore, the method for obtaining the regional grayscale change value includes:

[0022] The two side boundaries include a first boundary and a second boundary;

[0023] Obtain a first difference between the average grayscale value of each grayscale change edge value and the average grayscale value of the first boundary, and obtain a second difference between the average grayscale value of each grayscale change edge and the average grayscale value of the second boundary; and use the average first difference and the average second difference of all grayscale change edges as the regional grayscale change value.

[0024] Furthermore, the method for obtaining the damage change value includes:

[0025] The minimum circumscribed circle is fitted for all the areas formed by the crack edges, and the center of the minimum circumscribed circle is taken as the crack center; in the two side boundaries, the first distance between the crack center and the first boundary, and the second distance between the crack center and the second boundary are taken as the damage change values.

[0026] Furthermore, the non-correlation acquisition method includes:

[0027] The ratio of the first distance to the second distance is taken as the first ratio, the ratio of the average first difference to the average second difference is taken as the second ratio; and the difference between the first ratio and the second ratio is taken as the non-correlation.

[0028] Furthermore, the method for obtaining the degree of looseness includes:

[0029] A negative correlation mapping is performed on the sum of the average first difference and the average second difference to obtain an initial looseness degree, and a normalization process is performed on the product of the initial looseness degree and the non-correlation to obtain the looseness degree.

[0030] The present invention also proposes a computer vision-based armored cable sheath defect detection system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the computer vision-based armored cable sheath defect detection method are implemented.

[0031] The present invention has the following beneficial effects:

[0032] The embodiment of the present invention takes into account that the edge information on the armored cable protective cover is relatively rich. First, the edge of the raised area is segmented to obtain the local cable area. In the subsequent processing, each local cable area is analyzed in a targeted manner, and the influence of irrelevant edges is preliminarily reduced. Considering that the edge of the crack of the armored cable is usually perpendicular to the cable direction and has obvious grayscale characteristics during use, the edge of the crack is accurately screened out based on this to avoid the influence of the edge of the light-dark dividing line generated by light, and then the degree of cracking can be determined according to the edge of the crack. Further considering that if the protective cover is loose, the degree of concave-convexity of the armored cable raised area will change, and then the light and dark characteristics of the boundary position of the local cable area will change. Therefore, the grayscale change edge is screened out and the regional grayscale change value is determined, and the degree of looseness is further determined in combination with the damage change value characterized by the crack edge. The degree of looseness and the degree of cracking can be combined to accurately quantify the degree of defect. The embodiment of the present invention uses the characteristics of different types of edges to screen out multiple types of edges and perform corresponding processing, avoiding the influence of complex edges on defect detection, and realizing accurate quantification of the degree of defect of the armored cable protective cover. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0034] Figure 1 A flowchart of a method for detecting armored cable sheath defects based on computer vision provided by one embodiment of the present invention;

[0035] Figure 2 A schematic diagram of a surface grayscale image provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0036] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a computer vision-based armored cable sheath defect detection method and system proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

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

[0038] The following is a detailed description of a computer vision-based armored cable sheath defect detection method and system provided by the present invention in conjunction with the accompanying drawings.

[0039] See also Figure 1 , which shows a flow chart of a method for detecting armored cable sheath defects based on computer vision provided by an embodiment of the present invention, the method comprising:

[0040] Step S1: obtaining a surface grayscale image of an armored cable installed with a protective cover; in the surface grayscale image, obtaining a number of local cable areas according to the edges of the raised areas on the protective cover.

[0041] In one embodiment of the present invention, the armored cable to be inspected is placed on a horizontal inspection platform, and an industrial high-resolution camera for collecting surface grayscale images is deployed above the inspection platform. Figure 2 , which shows a schematic diagram of a surface grayscale image provided by an embodiment of the present invention. Figure 2 The surface of the armored cable protective cover shown can be considered to be composed of multiple raised areas. Because the raised areas of the armor layer also have obvious edge information, in order to avoid the influence of such edges on defect detection, the cable area in the surface grayscale image should be segmented according to the edges of the raised areas on the protective cover to obtain several local cable areas. In the subsequent image processing process, targeted analysis is performed on each local cable area to eliminate the influence of irrelevant edges.

[0042] It should be noted that the grayscale method used in the embodiment of the present invention is weighted average grayscale, and the specific operation is a technical means well known to those skilled in the art, which will not be described in detail here.

[0043] In the embodiment of the present invention, the canny edge detection algorithm is used to identify the edge in the image, and continuous edge pixels form an edge.

[0044] Optionally, in an embodiment of the present invention, because the raised areas on the surface of the armored cable protective cover are evenly distributed, the armored cable can be placed in a fixed position during the image acquisition stage, and the image field of view is fixed, so that the raised areas on the protective cover can be directly divided according to the preset areas, thereby determining the local cable area.

[0045] Preferably, in one embodiment of the present invention, considering that the edges of the raised areas on the protective cover are relatively vertically distributed relative to the cable direction and are approximately a straight line, the embodiment of the present invention performs straight line fitting on each edge in the surface grayscale image and obtains the fitting deviation between each edge and the fitted straight line; the angle between the fitted straight line and the perpendicular direction of the cable direction is taken as the deviation angle. That is, the smaller the fitting deviation, the closer the edge is to a straight line, and the smaller the deviation angle, the more perpendicular the direction of the edge is to the cable direction. Therefore, the preferred degree of each edge can be obtained according to the fitting deviation and the deviation angle, and the edges of the suspected raised areas can be screened out according to the preferred degree. Because there may be some cracks that are also a straight line perpendicular to the cable direction, only the edges of the suspected raised areas are screened out according to the preferred degree, and further analysis is required. Because the raised areas on the protective cover are evenly distributed, that is, the intervals between the edges of adjacent raised areas should be close to the same value, the distance between the edges of adjacent suspected raised areas is taken as the interval distance; the mode of the interval distance is taken as the raised area interval, and the edges of adjacent suspected raised areas with an interval distance equal to the raised area interval are taken as a set of true raised area edge pairs. Taking any true raised area edge pair as the center, the cable area in the surface grayscale image is divided according to the raised area interval to obtain several local cable areas. That is, along the center to both sides, the division is performed according to the raised area interval as the step length until the cable area is completely divided to obtain the local cable area.

[0046] In one embodiment of the present invention, the linear fitting method adopts the least square method for fitting, and the fitting deviation is obtained by: in the image coordinate system, the distance between each pixel point on the edge and the fitting straight line is accumulated as the fitting deviation. The deviation angle is an acute angle.

[0047] In one embodiment of the present invention, Figure 2 As shown, there is an obvious grayscale difference between the armored cable and the plane of the detection platform. Therefore, the cable area can be identified and used for segmentation using a conventional image segmentation algorithm or a semantic segmentation network. The specific operation is a technical means well known to those skilled in the art and will not be described in detail here.

[0048] Furthermore, in one embodiment of the present invention, a method for obtaining the degree of preference includes:

[0049] The sum of the fitting deviation and the deviation angle is negatively correlated to obtain the degree of preference. As an example, the degree of preference is expressed by the formula: ;in is the preference degree of the ith edge, is the deviation angle of the ith edge, is the fitting deviation of the ith edge. The positive integer 1 in the denominator is used to prevent the denominator from being zero.

[0050] In one embodiment of the present invention, the preference degree is normalized, a preference degree threshold is set to 0.8, and an edge with a preference degree greater than the preference degree threshold is regarded as an edge of a suspected convex region.

[0051] Step S2: According to the gray value and distribution direction of the edge in the local cable area, the crack edge is screened out, and the crack degree of the local cable area is obtained according to the morphological distribution and gray value of the crack edge.

[0052] During the use of cables, fatigue damage will occur to the protective cover due to excessive use such as bending and stretching, and fatigue cracks will occur on its surface. Due to the special shape of the cable, the cracks in the protective cover will be perpendicular to the direction of the cable under the influence of bending stress, and the cracks produced are in the shape of a tortoise shell, with multiple small cracks concentrated together. In addition, there is a significant difference between the grayscale information of the cracks and the grayscale of the protective cover itself. If a shallow crack occurs, the crack edge will be darker than the protective cover itself; if a deeper crack occurs, the metal armor layer of the cable will be exposed, showing brighter grayscale information. Therefore, based on these features, the crack edge can be screened out according to the grayscale value size and distribution direction of the edge in the local cable area.

[0053] Preferably, in one embodiment of the present invention, the method for obtaining the crack edge includes:

[0054] In the local cable area, the difference between the average grayscale value of each edge and the average grayscale value of the local cable area is used as the crack visibility. The greater the crack visibility, the more likely the edge is a crack edge. Further considering that the smaller the deviation angle, the more perpendicular the edge is to the cable direction, if the crack visibility of the edge is greater than the preset crack visibility threshold and the deviation angle is less than the preset angle threshold, the edge is regarded as a crack edge. In one embodiment of the present invention, after the crack visibility is normalized, the crack visibility threshold is set to 0.8 and the angle threshold is set to 10.

[0055] In the local cable area, the more crack edges there are, the wider the distribution range is, and the closer the adjacent crack edges are distributed, the more obvious the cracks are, which means that the degree of cracking at the corresponding position of the protective cover is greater. Therefore, the degree of cracking in the local cable area can be obtained based on the morphological distribution of the crack edges and the gray value size.

[0056] Preferably, in one embodiment of the present invention, the method for obtaining the degree of cracking includes:

[0057] In the local cable area, the crack area of ​​each crack edge is obtained according to the width and length of each crack edge. The larger the crack area, the greater the impact of the crack edge on the protective cover, and the greater the degree of cracking.

[0058] For a crack edge, the closer it is to other crack edges in the neighborhood, the more concentrated the crack distribution is, and the greater the degree of cracking is. Therefore, the local concentration degree of each crack edge is obtained according to the average distance between each crack edge and other crack edges in the preset neighborhood. The local concentration degree of each crack edge is further used as a weight to weight the degree of crack visibility of each crack edge to obtain the crack degree, that is, the greater the crack degree is, the greater the impact of the crack edge on the protective cover is, and the greater the degree of cracking is.

[0059] The crack degree and crack area of ​​each crack edge are multiplied to obtain the initial crack degree of each crack edge; the accumulated values ​​of the initial crack degrees of all crack edges in the local cable area are normalized to obtain the crack degree.

[0060] It should be noted that, in one embodiment of the present invention, the preset neighborhood range of the crack edge is the K neighborhood of the crack edge, that is, the neighborhood range composed of the K other crack edges that are closest to each other. In the embodiment of the present invention, the K value is set to 3.

[0061] As an example, the degree of cracking is expressed as:

[0062] ;in is the cracking degree of the jth local cable area, is the number of crack edges in the jth local cable region, is the width of the hth crack edge, is the length of the hth crack edge, is the average gray value of the hth crack edge, is the average gray value of the jth local cable area, is the average distance between the hth crack edge and other crack edges within the preset neighborhood, Hyperbolic tangent function.

[0063] In the cracking degree formula, It indicates the degree of local aggregation, that is, the average distance is negatively correlated by the inverse. The smaller the distance, the more concentrated the distribution of the crack edges in the neighborhood. The multiplication method is used to measure the degree of obviousness of the cracks. The weighting is performed, and the greater the local aggregation degree, the more obvious the crack is, and the greater the crack degree is. The initial crack degrees of all crack edges are combined through the accumulation operation, that is, the more crack edges there are and the greater the initial crack degree of the crack edges, the greater the crack degree of the local cable area. The final result is further normalized by the hyperbolic tangent function.

[0064] Step S3: Filter out the grayscale change edges in the local cable area; the grayscale change edges are non-crack edges extending into the local cable area based on the two side boundaries corresponding to the cable direction; obtain the regional grayscale change value according to the grayscale difference between the grayscale change edge and the two side boundaries, and obtain the damage change value according to the relative distance between the distribution area of ​​the crack edge and the two side boundaries; obtain the looseness degree of the local cable area according to the non-correlation between the regional grayscale change value and the damage change value, and the regional grayscale change value.

[0065] For the local cable area, because it is divided according to the edge of the raised area on the protective cover, and the raised area on the protective cover has a fixed height, the effect of light on the raised area is consistent. After being irradiated by light, there will be grayscale change edges on the local cable area, which are edges extending from the boundaries on both sides to the inside. Figure 2 Taking a horizontally placed cable as an example, the edge on the local cable area is a grayscale change edge as long as one end is on the left and right boundaries of the local cable area and the other end is a non-crack edge extending inward. The greater the height of the raised area, the grayscale change edge will change to a lower grayscale due to the influence of the raised area. If the corresponding position of the local cable area is loose, it will be equivalent to the raised area becoming lower, and the grayscale value on the grayscale change edge will become larger. Therefore, the regional grayscale change value can be obtained based on the grayscale difference between the grayscale change edge and the two sides of the boundary, that is, the smaller the regional grayscale change value, the greater the degree of looseness of the corresponding position. Further considering that there may be a cracked area composed of crack edges in the local cable area, the crack edges in the cracked area have certain grayscale characteristics, so the crack edges will affect the evaluation results of the regional grayscale change value. The closer the cracked area is to one side boundary, the greater the impact on one side boundary. First, according to the relative distance between the distribution area of ​​the crack edge and the two side boundaries, the damage change value is obtained, and then the non-correlation between the regional grayscale change value and the damage change value is obtained. The non-correlation can be used as the confidence of the regional grayscale change value. The smaller the non-correlation, the greater the correlation between the current regional grayscale change value characteristics and the damage change value characteristics. The grayscale information presented by the regional grayscale change value is more likely to be caused by the crack edge. The smaller the reference of the regional grayscale change value, the lower the confidence should be.

[0066] Preferably, in one embodiment of the present invention, the method for obtaining the regional grayscale change value includes:

[0067] The two side boundaries include the first boundary and the second boundary;

[0068] A first difference between the average grayscale value of each grayscale change edge and the average grayscale value of the first boundary is obtained, and a second difference between the average grayscale value of each grayscale change edge and the average grayscale value of the second boundary is obtained; because there are multiple grayscale change edges, the average first difference and the average second difference of all grayscale change edges are used as the regional grayscale change value.

[0069] Furthermore, in one embodiment of the present invention, the two side boundaries are also divided into a first boundary and a second boundary, and corresponding to the regional grayscale change value, a method for obtaining the damage change value includes:

[0070] The minimum circumscribed circle is fitted for the areas formed by all crack edges, and the center of the minimum circumscribed circle is taken as the crack center; on both sides of the boundary, the first distance between the crack center and the first boundary, and the second distance between the crack center and the second boundary are taken as the damage change values.

[0071] Furthermore, based on the above-mentioned method for obtaining the regional grayscale change value and the fragmentation change value, the non-correlation obtaining method includes:

[0072] The ratio of the first distance to the second distance is taken as the first ratio, the ratio of the average first difference to the average second difference is taken as the second ratio, and the difference between the first ratio and the second ratio is taken as the non-correlation. That is, the characteristics of the regional grayscale change value and the characteristics of the fragmentation change value are quantified in the form of ratios. The greater the difference between the two ratios, the stronger the non-correlation.

[0073] Preferably, in one embodiment of the present invention, the method for obtaining the degree of looseness includes:

[0074] The sum of the average first difference and the average second difference is negatively correlated to obtain the initial looseness, and the product of the initial looseness and the non-correlation is normalized to obtain the looseness. As an example, the looseness is expressed by the formula:

[0075] ;in is the looseness of the jth local cable area, is the average first difference of the jth local cable area, is the average second difference of the jth local cable area, is the first distance of the jth local cable area, The second distance of the jth local cable area, is the hyperbolic tangent function used for normalization, is the absolute value symbol.

[0076] In the looseness formula, it should be noted that due to the reflective characteristics of the raised area of ​​the armored cable protective sheath, the regional grayscale change value and the fragmentation change value do not equal 0, that is, the denominator in the looseness formula does not equal 0, so the negative correlation mapping is achieved in the form of an inverse.

[0077] Step S4: Obtain the defect degree of the local cable area according to the crack degree and the looseness degree.

[0078] In one embodiment of the present invention, the product of the degree of cracking and the degree of looseness is used as the defect degree of the local cable area to quantify the defect degree of the local area of ​​the armored cable protective cover. By quantifying the degree of defects, it is convenient for staff to refer to the processing of the current protective cover. For example, in one embodiment of the present invention, the degree of defects is normalized. If the degree of defects is less than 0.3, it means that the current protective cover has minor damage, which does not affect the use at that time and can be repaired with a suitable adhesive; for areas with a defect degree greater than or equal to 0.3 and less than 0.8, you can consider using a patch or applying appropriate touch-up paint to repair the surface of the protective cover; for areas with a defect degree greater than or equal to 0.8, you need to consider replacing a new protective cover. It should be noted that when repairing the protective cover, it is necessary to select the repair material according to the material of the protective cover to ensure that the repair material is compatible with the cable material and can withstand the normal stress and environmental conditions of the armored cable. For example, it is necessary to select adhesives and patches suitable for the material and stress environment.

[0079] In summary, the embodiment of the present invention performs segmentation according to the edge of the raised area to obtain a local cable area. The crack edge is screened out according to the characteristics of the crack edge and the degree of cracking is evaluated. The grayscale change edge is screened out and the grayscale change value is determined, and the degree of looseness is determined in combination with the damage change value characterized by the crack edge. The degree of defect can be accurately quantified by combining the degree of looseness and the degree of cracking. The embodiment of the present invention uses the characteristics of different types of edges to screen out multiple types of edges and perform corresponding processing, thereby avoiding the influence of complex edges on defect detection and achieving accurate quantification of the degree of defects in the armored cable protective sheath.

[0080] The present invention also proposes a computer vision-based armored cable sheath defect detection system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any one of the steps of a computer vision-based armored cable sheath defect detection method.

[0081] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0082] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A method for detecting armored cable sheath defects based on computer vision, characterized in that: The method comprises: Acquire a surface grayscale image of an armored cable installed with a protective cover; in the surface grayscale image, obtain a number of local cable areas according to the edges of the raised areas on the protective cover; Screening out crack edges according to the grayscale value and distribution direction of the edges in the local cable area, and obtaining the degree of cracking in the local cable area according to the morphological distribution and grayscale value of the crack edges; Screen out the grayscale change edges in the local cable area; the grayscale change edges are non-crack edges extending into the local cable area based on the two side boundaries corresponding to the cable direction; obtain the regional grayscale change value according to the grayscale difference between the grayscale change edge and the two side boundaries, and obtain the damage change value according to the relative distance between the distribution area of ​​the crack edge and the two side boundaries; obtain the looseness degree of the local cable area according to the non-correlation between the regional grayscale change value and the damage change value, and the regional grayscale change value; The defect degree of the local cable area is obtained according to the crack degree and the looseness degree.

2. The method for detecting armored cable sheath defects based on computer vision according to claim 1, characterized in that: The method for obtaining the local cable area includes: Performing straight line fitting on each edge in the surface grayscale image, and obtaining the fitting deviation between each edge and the fitting straight line; taking the angle between the fitting straight line and the direction perpendicular to the cable direction as the deviation angle; Obtaining the preferred degree of each edge according to the fitting deviation and the deviation angle; screening out suspected convex region edges according to the preferred degree; taking the distance between adjacent suspected convex region edges as the interval distance; taking the mode of the interval distance as the convex region interval, and taking the adjacent suspected convex region edges whose interval distance is equal to the convex region interval as a set of true convex region edge pairs; Taking any real raised area edge pair as the center, the cable area in the surface grayscale image is divided according to the raised area intervals to obtain a number of local cable areas.

3. The method for detecting armored cable sheath defects based on computer vision according to claim 2, characterized in that: The method for obtaining the preference degree includes: The fitting deviation and the sum of the deviation angles are negatively correlated to obtain the degree of optimization.

4. The method for detecting armored cable sheath defects based on computer vision according to claim 2, characterized in that: The method for obtaining the edge of the crack comprises: In the local cable area, the difference between the average grayscale value of each edge and the average grayscale value of the local cable area is taken as the crack visibility; if the crack visibility of the edge is greater than the preset crack visibility threshold and the deviation angle is less than the preset angle threshold, the edge is taken as the crack edge.

5. The method for detecting armored cable sheath defects based on computer vision according to claim 4 is characterized in that: The method for obtaining the degree of cracking includes: In the local cable area, the crack area of ​​each crack edge is obtained according to the width and length of each crack edge; the local aggregation degree of each crack edge is obtained according to the average distance between each crack edge and other crack edges in a preset neighborhood range; the local aggregation degree of each crack edge is used as a weight to weight the crack conspicuity degree of each crack edge to obtain the crack degree; The crack degree and the crack area of ​​each crack edge are multiplied to obtain the initial crack degree of each crack edge; the accumulated value of the initial crack degrees of all crack edges in the local cable area is normalized to obtain the crack degree.

6. The method for detecting armored cable sheath defects based on computer vision according to claim 1, characterized in that: The method for obtaining the regional grayscale change value includes: The two side boundaries include a first boundary and a second boundary; Obtain a first difference between the average grayscale value of each grayscale change edge and the average grayscale value of the first boundary, and obtain a second difference between the average grayscale value of each grayscale change edge and the average grayscale value of the second boundary; and use the average first difference and the average second difference of all grayscale change edges as the regional grayscale change value.

7. The method for detecting armored cable sheath defects based on computer vision according to claim 6, characterized in that: The method for obtaining the damage change value includes: The minimum circumscribed circle is fitted for all the areas formed by the crack edges, and the center of the minimum circumscribed circle is taken as the crack center; in the two side boundaries, the first distance between the crack center and the first boundary, and the second distance between the crack center and the second boundary are taken as the damage change values.

8. The method for detecting armored cable sheath defects based on computer vision according to claim 7, characterized in that: The non-correlation acquisition method includes: The ratio of the first distance to the second distance is taken as the first ratio, the ratio of the average first difference to the average second difference is taken as the second ratio; and the difference between the first ratio and the second ratio is taken as the non-correlation.

9. The method for detecting armored cable sheath defects based on computer vision according to claim 8, characterized in that: The method for obtaining the degree of looseness includes: A negative correlation mapping is performed on the sum of the average first difference and the average second difference to obtain an initial looseness degree, and a normalization process is performed on the product of the initial looseness degree and the non-correlation to obtain the looseness degree.

10. A computer vision-based armored cable sheath defect detection system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the armored cable sheath defect detection method based on computer vision as described in any one of claims 1 to 9 are implemented.

Images