Surface defect detection method of low-smoke halogen-free cables based on binocular vision
Through depth map analysis based on binocular vision, screening and evaluating scratches and expansion cracks on the surface of low-smoke halogen-free cables, the problem of insufficient defect detection accuracy in the prior art is solved, and the low-smoke halogen-free cables are realized Accurate evaluation and prediction of surface defects.
Patent Information
- Application Number
- CN202411125725.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-16
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2044-08-16
AI Technical Summary
The surface defect detection methods for low-smoke halogen-free cables in the prior art mainly rely on visual inspection, and cannot accurately evaluate the degree of harm and future deterioration of defective areas, resulting in insufficient detection accuracy.
Using a binocular vision-based method, the depth value difference of each pixel point is obtained through depth map analysis, the scratch and expansion crack areas are screened out, the probability that the branch line of each scratch belongs to the expansion crack is calculated, and the expansion of the scratch is judged. possibilities, and then assess the degree of harm of the defective area.
It improves the accuracy of surface defects of low-smoke halogen-free cables, can more accurately determine whether scratches may deteriorate in the future, and promptly evaluate and report and handle them, improving the accuracy of detection.
Smart Images

Figure CN118657760B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of image data processing, and in particular to a method for detecting surface defects of low-smoke halogen-free cables based on binocular vision. Background Art
[0002] Low smoke halogen-free cable is a specially designed cable. The outer sheath material of low smoke halogen-free cable does not contain halogen compounds, such as chlorine, bromine, etc. In the event of a fire, no halogen compounds will be released, which can effectively reduce the generation of smoke and toxic gases. This cable not only reduces pollution to the environment, but also reduces the content of toxic gases produced by the cable in a fire. Low smoke halogen-free cable is usually used in indoor wiring systems, such as data centers, office buildings, public transportation systems, etc. At present, for the detection of defects on the surface of low smoke halogen-free cables, visual-based detection methods are usually used; however, conventional defect detection methods can only detect defective areas and reflect the degree of hazard by area. This method cannot reflect the situation where the defective area continues to deteriorate, and the assessment of the degree of hazard in the defective area is inaccurate, which reduces the accuracy of surface defect detection of low smoke halogen-free cables. Summary of the invention
[0003] The present invention provides a low-smoke halogen-free cable surface defect detection method based on binocular vision to solve the problem that the current low-smoke halogen-free cable surface defect detection usually adopts a vision-based detection method; however, the conventional defect detection method can only detect the defect area and reflect the degree of harm by the area. This method cannot reflect the situation where the defect area continues to deteriorate, and the evaluation of the degree of harm of the defect area is inaccurate.
[0004] The low-smoke halogen-free cable surface defect detection method based on binocular vision of the present invention adopts the following technical solutions:
[0005] An embodiment of the present invention provides a method for detecting surface defects of low-smoke halogen-free cables based on binocular vision, the method comprising the following steps:
[0006] Obtain the depth map of the low-smoke halogen-free cable, the depth value of each pixel, and several suspected scratch areas;
[0007] According to the difference between the depth values of each pixel in the low-smoke halogen-free cable depth map, the abnormality degree of each pixel is obtained;
[0008] According to the abnormal degree of the pixel points in each suspected scratch area, a number of scratches and a number of suspected expansion crack areas of each scratch are screened out from the suspected scratch area;
[0009] According to each scratch and the suspected expansion crack area of each scratch, a number of branch lines of each scratch are obtained;
[0010] According to the difference between the abnormal degrees of the pixels of each branch line of each scratch, the probability that each branch line of each scratch belongs to an expansion crack is obtained; according to the probability that each branch line of each scratch belongs to an expansion crack, a number of expansion cracks of each scratch are obtained;
[0011] According to the expansion crack of each scratch and the abnormal degree of the pixel point of the expansion crack, the possibility of each scratch expansion is obtained; according to the possibility of each scratch expansion, it is judged whether each scratch is likely to deteriorate in the future.
[0012] Preferably, the method of obtaining the abnormality degree of each pixel point according to the difference between the depth values of each pixel point in the low-smoke halogen-free cable depth map includes the following specific steps:
[0013] Get a baseline in the LSZH cable depth map;
[0014] According to the depth value of each pixel in the low-smoke halogen-free cable depth map and the depth value of the pixel on the line parallel to the baseline, the abnormality degree of each pixel is obtained.
[0015] Preferably, the obtaining of the baseline in the low-smoke halogen-free cable depth map comprises the following specific steps:
[0016] In the low-smoke halogen-free cable depth map, the Canny edge detection algorithm is used to process the cable depth map to obtain the edge detection result of the cable surface; the Hough transform line detection algorithm is used to extract the edge lines on both sides of the cable from the edge detection results of the cable surface; and the longer edge line on both sides is used as the baseline.
[0017] Preferably, the abnormality degree of each pixel is obtained according to the depth value of each pixel in the low-smoke halogen-free cable depth map and the depth value of the pixel on the line parallel to the reference line, and the specific formula included is as follows:
[0018]
[0019] In the formula, Indicates the depth diagram of low smoke zero halogen cable The abnormality degree of each pixel; Indicates the depth diagram of low smoke zero halogen cable The depth value of each pixel; In the depth diagram of low smoke zero halogen cable, The pixel points on the line parallel to the reference line are Pixels outside The depth value of each pixel; In the depth diagram of low smoke zero halogen cable, The pixel points on the line parallel to the reference line are The number of all pixels except pixels; is the normalization function.
[0020] Preferably, the method of screening out a plurality of scratches and a plurality of suspected expansion crack regions of each scratch from the suspected scratch region according to the abnormality degree of the pixel points in each suspected scratch region comprises the following specific steps:
[0021] The average value of the abnormality of all pixels in each suspected scratch area is recorded as the scratch possibility of each suspected scratch area;
[0022] The scratch probability is greater than the preset threshold The suspected scratch area is recorded as a scratch;
[0023] On the boundary of any scratch, if the degree of abnormality in the eight neighborhoods of any pixel point is greater than the preset threshold The eight neighborhoods are recorded as the suspected expansion crack areas of any scratch.
[0024] Preferably, the method of obtaining a plurality of branch lines of each scratch according to each scratch and the suspected expansion crack region of each scratch comprises the following specific steps:
[0025] A morphological refinement operation is performed on each scratch and the area formed by all suspected expansion crack areas to obtain the skeleton line corresponding to each scratch; the medial axis extraction algorithm is used to extract the medial axis of the skeleton line corresponding to each scratch to obtain a medial axis and several branch lines.
[0026] Preferably, the method of obtaining the probability that each branch line of each scratch belongs to an expansion crack according to the difference between the abnormal degrees of the pixel points of each branch line of each scratch comprises the following specific steps:
[0027] In the The first scratch On each branch line, the abnormality degree of each pixel point is counted in turn to obtain the abnormality degree sequence;
[0028] In the The first scratch On each branch line, the chain code value of each pixel point is counted in turn to obtain the chain code sequence;
[0029]
[0030] In the formula, Indicates The first scratch The probability that a branch line belongs to an expansion crack; Indicates The first scratch On the branch line The abnormality degree of each pixel; Indicates The first scratch On the branch line The abnormality degree of each pixel; Indicates The first scratch The number of all pixels on a branch line; Indicates The first scratch The variance of all data in the first-order difference sequence of the abnormality degree sequence corresponding to the branch line; Indicates The first scratch The variance of all data in the chain code sequence corresponding to each branch line; is the normalization function; is an exponential function with a natural constant as its base.
[0031] Preferably, the method of obtaining a plurality of expansion cracks of each scratch according to the probability that each branch line of each scratch belongs to an expansion crack comprises the following specific steps:
[0032] When The first scratch The probability that a branch line belongs to an expansion crack is greater than the preset threshold When The first scratch The branch line is A scratch expansion crack.
[0033] Preferably, the method of obtaining the possibility of each scratch expansion according to the expansion crack of each scratch and the abnormality degree of the pixel point of the expansion crack includes the following specific steps:
[0034] For The scratches are detected by using the contour detection algorithm based on connectivity. The contour of each propagating crack in the scratch is The first scratch The Euclidean distance between adjacent sampling points is calculated, and the distances between all adjacent sampling points are accumulated to obtain the The first scratch The length of a propagating crack;
[0035] No. The probability of a scratch expanding is calculated as follows:
[0036]
[0037] In the formula, Indicates The possibility of scratch expansion; Indicates The first scratch The length of a propagating crack; Indicates The number of all propagating cracks in a scratch; Indicates The first scratch The first expansion crack The abnormality degree of each pixel; Indicates The first scratch The number of all pixels of a propagating crack; is the normalization function.
[0038] Preferably, judging whether each scratch is likely to deteriorate in the future based on the possibility of each scratch expanding comprises the following specific steps:
[0039] When The probability of scratch expansion is greater than the preset threshold When A scratch may worsen over time.
[0040] The beneficial effects of the technical solution of the present invention are: according to the difference between the abnormal degree of the pixel points of each branch line of each scratch, the probability that each branch line of each scratch belongs to an expansion crack is obtained; according to the probability that each branch line of each scratch belongs to an expansion crack, a number of expansion cracks of each scratch are obtained; the accuracy of judging the expansion cracks around the scratch can be improved, which is conducive to detecting the surface defects of low-smoke halogen-free cables. According to the expansion crack of each scratch and the abnormal degree of the pixel points of the expansion crack, the possibility of each scratch expansion is obtained; according to the possibility of each scratch expansion, it is judged whether each scratch is likely to deteriorate in the future; according to the degree of harm of each scratch, it is judged whether it needs to be repaired; the degree of harm of the defective area is evaluated more accurately, which improves the accuracy of low-smoke halogen-free cable surface defect detection. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be 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.
[0042] Figure 1 The present invention is a flowchart of the steps of the low-smoke halogen-free cable surface defect detection method based on binocular vision. DETAILED DESCRIPTION
[0043] 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 low-smoke halogen-free cable surface defect detection method based on binocular vision proposed by the present invention, its specific implementation method, structure, characteristics and effects, 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.
[0044] 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.
[0045] The specific scheme of the low-smoke halogen-free cable surface defect detection method based on binocular vision provided by the present invention is described in detail below with reference to the accompanying drawings.
[0046] See also Figure 1 , which shows a flowchart of a method for detecting surface defects of low-smoke halogen-free cables based on binocular vision according to an embodiment of the present invention, the method comprising the following steps:
[0047] Step S001: Obtain a depth map of a low-smoke halogen-free cable, a depth value of each pixel point, and a number of suspected scratch areas.
[0048] It should be noted that binocular stereo vision measurement technology uses two cameras to observe the same scene or object from slightly different angles, and obtains the three-dimensional information of the object by calculating the disparity (position deviation of corresponding points in the image). First, feature points such as corner points, edge points or regional features are extracted from each image, and then, these feature points are matched in the two images to establish a correspondence between them. This step is usually implemented using feature descriptors (such as SIFT, SURF) and matching algorithms (such as matching based on minimum distance). Once the correspondence between the feature points is established, the disparity between these points is calculated. The disparity represents the pixel offset of the same object point in the two images. The larger the disparity, the closer the object point is to the camera; the smaller the disparity, the farther the object point is from the camera.
[0049] The image of the low-smoke halogen-free cable to be tested is obtained by taking pictures with a binocular camera, and the image of the low-smoke halogen-free cable to be tested is gray-scaled to obtain a gray-scale image of the low-smoke halogen-free cable;
[0050] It should be noted that the focal length of the binocular camera is obtained by consulting the camera specification table; the actual distance between the left and right cameras is the baseline distance;
[0051] Get the focal length and baseline distance of the binocular camera;
[0052] Input the grayscale image of low-smoke halogen-free cable into the semi-global stereo matching algorithm to obtain the disparity value corresponding to each pixel;
[0053] According to the normalized focal length, baseline distance and disparity value of the binocular camera, the depth map of the low-smoke halogen-free cable is obtained; according to the conversion formula between disparity and depth, the depth value of each pixel in the depth map of the low-smoke halogen-free cable is obtained;
[0054] It should be noted that, using known camera parameters (such as focal length, viewing angle, camera internal and external parameters, etc.) and parallax information, the three-dimensional coordinates of each corresponding point on the surface of the object can be calculated through geometric triangulation or more advanced visual algorithms. It should be noted that what is obtained is the three-dimensional information of each pixel point, that is, the horizontal and vertical coordinates and depth. This embodiment uses the minimum and maximum norm method for normalization, which is a well-known technology.
[0055] Among them, the semi-global stereo matching algorithm and the conversion formula between disparity and depth are well-known technologies, and the specific methods are not introduced here.
[0056] It should be noted that when performing Canny edge detection in the depth map of the low-smoke halogen-free cable, the scratch area can be preliminarily determined because of the high contrast of the scratch area, which contrasts with the surrounding smooth surface; and the scratches usually exist continuously within a certain range, rather than intermittent edges.
[0057] Perform Canny edge detection on the depth map of the low-smoke halogen-free cable to obtain several edge lines, and perform edge connection on all edge lines to form a closed area, thereby obtaining several suspected scratch areas of the cable depth map;
[0058] Obtain the depth map of the low-smoke zero-halogen cable, the depth value of each pixel, and several suspected scratch areas.
[0059] Step S002: Obtain the abnormality degree of each pixel point according to the difference between the depth values of each pixel point in the low-smoke halogen-free cable depth map.
[0060] It should be noted that some minor scratches on the surface of low-smoke halogen-free cables will not affect the normal use of the cables if the scratches do not have an obvious expansion trend. However, if some scratches have an obvious expansion trend, the defective area is more harmful and needs to be repaired as soon as possible. The judgment of whether the scratch has an obvious expansion trend can be judged by the characteristics of the small cracks around the scratches. These small cracks are difficult to extract under threshold segmentation. Therefore, it is necessary to use an intelligent inspection robot equipped with a binocular camera to segment the suspected expansion cracks of the scratches for further judgment.
[0061] It should be noted that when a certain cable pixel is a suspected scratch expansion crack pixel, the depth value of the pixel is greatly different from the depth value of the pixel under normal circumstances. The greater the difference, the greater the abnormality of the pixel. For the calculation of the depth value of a certain cable pixel under normal circumstances, the depth distribution of the cable pixel is analyzed for prediction. Since the cable is cylindrical, in the depth map of a normal cable, the depth value of the pixel sequence in the extension direction of the cable changes little, and the depth value changes greatly in the extension direction perpendicular to the cable. Therefore, the depth value of a certain pixel under normal circumstances can be predicted by the mean of the depth values of the pixel sequence passing through the pixel and parallel to the extension direction of the cable.
[0062] In the low-smoke halogen-free cable depth map, the Canny edge detection algorithm is used to process the cable depth map to obtain the edge detection result of the cable surface; the Hough transform line detection algorithm is used to extract the edge lines on both sides of the cable from the edge detection result of the cable surface; the longer edge line on both sides is used as the reference line;
[0063] Among them, the Canny edge detection algorithm and the Hough transform line detection algorithm are well-known technologies, and the specific methods are not introduced here.
[0064] Low smoke zero halogen cable depth chart The calculation method of the abnormality degree of each pixel is as follows:
[0065]
[0066] In the formula, Indicates the depth diagram of low smoke zero halogen cable The abnormality degree of each pixel; Indicates the depth diagram of low smoke zero halogen cable The depth value of each pixel; In the depth diagram of low smoke zero halogen cable, The pixel points on the line parallel to the reference line are Pixels outside The depth value of each pixel; In the depth diagram of low smoke zero halogen cable, The pixel points on the line parallel to the reference line are The number of all pixels except pixels; is the normalization function.
[0067] It should be noted that Expressed the The pixel points on the line parallel to the reference line are The average depth value of all pixels except the pixels, which means that under normal circumstances, the The depth value of each pixel; Indicates The difference between the actual depth value of the pixel and the normal depth value is greater. The greater the abnormality of the pixel point.
[0068] At this point, the abnormality degree of each pixel in the low-smoke halogen-free cable depth map is obtained.
[0069] Step S003: According to the abnormal degree of pixel points in each suspected scratch area, several scratches and several suspected expansion crack areas of each scratch are screened out from the suspected scratch area; according to each scratch and the suspected expansion crack area of each scratch, several branch lines of each scratch are obtained; according to the difference between the abnormal degrees of pixel points of each branch line of each scratch, the probability that each branch line of each scratch belongs to an expansion crack is obtained; according to the probability that each branch line of each scratch belongs to an expansion crack, several expansion cracks of each scratch are obtained.
[0070] It should be noted that since the expansion crack always appears on the edge of the scratch, the pixel sequence on the boundary of the suspected scratch area is extracted, and crack extraction is performed with each pixel point as the starting point. The abnormality of all pixels in its eight neighborhoods is calculated with a certain edge pixel point on the boundary as the center, and a smaller threshold is set. This is because when the threshold is set to a large value, the complete expansion crack may not be extracted; the pixel point with an abnormality greater than the preset threshold is selected. Pixels (excluding the selected pixels and the pixels in the scratch area) are selected, and the lines from the central pixel to these pixels may be the extension direction of the expansion crack; based on this, the small crack area attached to the scratch is the expansion crack.
[0071] Preset Threshold The preset threshold is 0.05. is 0.7;
[0072] The average value of the abnormality of all pixels in each suspected scratch area is recorded as the scratch possibility of each suspected scratch area;
[0073] The scratch probability is greater than the preset threshold The suspected scratch area is recorded as a scratch.
[0074] On the boundary of any scratch, if the degree of abnormality in the eight neighborhoods of any pixel point is greater than the preset threshold The eight neighborhoods are recorded as the suspected expansion crack area of any scratch;
[0075] According to the above method, the eight neighborhoods of each pixel point on the boundary of the scratch are determined to obtain several suspected expansion crack regions of the scratch;
[0076] According to the above method, several suspected expansion crack regions of each scratch are obtained;
[0077] A morphological refinement operation is performed on each scratch and the area formed by all suspected expansion crack areas to obtain the skeleton line corresponding to each scratch; the medial axis extraction algorithm is used to extract the medial axis of the skeleton line corresponding to each scratch to obtain a medial axis and several branch lines.
[0078] Among them, the morphological thinning operation and the medial axis extraction algorithm are well-known technologies, and the specific methods are not introduced here.
[0079] It should be noted that in the previous step, several branch lines of each scratch were extracted, but many interference areas also have the above characteristics, so further investigation is needed.
[0080] It should be further explained that for scratch expansion cracks, when the crack starts to extend from the edge of the scratch, the initial depth may be large, and when the crack extends to the end, the depth of the crack is shallow. During the entire extension process of the expansion crack, the abnormal degree of its pixel point may decrease gradually. When the branch line of the scratch meets this feature, the probability that the branch line belongs to the expansion crack of the scratch is greater. And because the expansion direction of the expansion crack cannot be controlled during the expansion process, the change of the branch line is more tortuous, so the greater the degree of change of the expansion crack, the greater the probability that the branch line belongs to the expansion crack. Specifically, the degree of change of the branch line can be calculated using the chain code for any branch line.
[0081] In the The first scratch On each branch line, the abnormality degree of each pixel point is counted in turn to obtain the abnormality degree sequence;
[0082] Calculate the first-order difference sequence of the abnormality degree series;
[0083] In the The first scratch On each branch line, the chain code value of each pixel is counted in turn to obtain the chain code sequence.
[0084] It should be noted that the starting point of each branch line is the endpoint connected to the central axis;
[0085] The chain code value of the pixel point is calculated using the 8-connected chain code algorithm;
[0086] Among them, the first-order difference and 8-connected chain code algorithms are well-known technologies, and the specific methods are not introduced here.
[0087] No. The first scratch The probability that a branch line belongs to a propagating crack is calculated as follows:
[0088]
[0089] In the formula, Indicates The first scratch The probability that a branch line belongs to an expansion crack; Indicates The first scratch On the branch line The abnormality degree of each pixel; Indicates The first scratch On the branch line The abnormality degree of each pixel; Indicates The first scratch The number of all pixels on a branch line; Indicates The first scratch The variance of all data in the first-order difference sequence of the abnormality degree sequence corresponding to the branch line; Indicates The first scratch The variance of all data in the chain code sequence corresponding to each branch line; is the normalization function; is an exponential function with a natural constant as its base.
[0090] What needs to be explained is that Any term in the first-order difference sequence representing the degree of abnormality. It represents the sum of the first-order difference sequence. The smaller the sum of the first-order difference sequence, the smaller the The first scratch The more the abnormality of the branch line satisfies the characteristic of gradual decrease, the The greater the probability that a branch line belongs to an expansion crack. Indicates The first scratch The variance of all data in the first-order difference sequence of the abnormal degree sequence corresponding to the branch line. The smaller the variance value, the higher the variance of the first-order difference sequence. The first scratch The more the abnormal degree corresponding to the branch line satisfies the uniform change, the smaller the gradient of the expansion crack becomes along the extension direction. The first scratch The greater the probability that a branch line belongs to an expansion crack. Indicates The first scratch The variance of all data in the chain code sequence corresponding to the branch line, the larger the variance is, the more it satisfies the irregular change characteristics of the expansion crack, and the greater the probability of belonging to the expansion crack.
[0091] Preset Threshold for , when The first scratch The probability that a branch line belongs to an expansion crack is greater than the preset threshold When The first scratch The branch line is An expansion crack from a scratch.
[0092] At this point, several expansion cracks are obtained for each scratch.
[0093] Step S004: according to the expansion crack of each scratch and the abnormal degree of the pixel point of the expansion crack, the possibility of each scratch expansion is obtained; according to the possibility of each scratch expansion, whether each scratch is likely to deteriorate in the future is determined.
[0094] It should be noted that in the previous step, all tiny expansion cracks around a scratch are extracted, and the possibility of the scratch expanding is calculated based on the expansion crack characteristics. For the expansion cracks around the scratch, the longer the crack length, the more obvious the tendency of the scratch to expand; and the deeper the depth of the suspected expansion crack area, the greater the possibility of the scratch expanding.
[0095] No. The probability of a scratch extending is calculated as follows:
[0096]
[0097] In the formula, Indicates The possibility of scratch expansion; Indicates The first scratch The length of a propagating crack; Indicates The number of all propagating cracks in a scratch; Indicates The first scratch The first expansion crack The abnormality degree of each pixel; Indicates The first scratch The number of all pixels of a propagating crack; is the normalization function.
[0098] What needs to be explained is that Indicates The sum of the lengths of all the expansion cracks of the scratch. The larger the sum of the lengths, the longer the expansion crack. The greater the possibility of a scratch expanding. Indicates The first scratch The sum of the abnormality levels of all pixels of a propagating crack; Indicates The sum of the abnormality of all the extended crack pixels of the scratch. The larger the value, the The more serious the expansion crack of the first scratch, the The greater the possibility of a scratch expanding.
[0099] Preset Threshold is 0.8, when The probability of scratch expansion is greater than the preset threshold When A scratch may worsen over time.
[0100] What needs to be explained is that the scratch may worsen in the future, which means that the scratch is more harmful and needs to be repaired in advance to avoid causing greater losses.
[0101] It should be noted that: in this embodiment, when the denominator in the formula is 0, the denominator is set to 1 to ensure that the formula is valid, and this example is used for description.
[0102] At this point, this embodiment is completed.
[0103] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A low-smoke halogen-free cable surface defect detection method based on binocular vision, characterized in that: The method comprises the following steps: Obtain the depth map of the low-smoke halogen-free cable, the depth value of each pixel, and several suspected scratch areas; According to the difference between the depth values of each pixel in the low-smoke halogen-free cable depth map, the abnormality degree of each pixel is obtained; According to the abnormal degree of the pixel points in each suspected scratch area, a number of scratches and a number of suspected expansion crack areas of each scratch are screened out from the suspected scratch area; According to each scratch and the suspected expansion crack area of each scratch, a number of branch lines of each scratch are obtained; According to the difference between the abnormal degrees of the pixels of each branch line of each scratch, the probability that each branch line of each scratch belongs to an expansion crack is obtained; according to the probability that each branch line of each scratch belongs to an expansion crack, a number of expansion cracks of each scratch are obtained; According to the expansion crack of each scratch and the abnormal degree of the pixel point of the expansion crack, the possibility of each scratch expansion is obtained; according to the possibility of each scratch expansion, it is judged whether each scratch is likely to deteriorate in the future; The method of obtaining the probability that each branch line of each scratch belongs to an expansion crack according to the difference between the abnormal degrees of the pixel points of each branch line of each scratch includes the following specific steps: In the The first scratch On each branch line, the abnormality degree of each pixel point is counted in turn to obtain the abnormality degree sequence; In the The first scratch On each branch line, the chain code value of each pixel point is counted in turn to obtain the chain code sequence; In the formula, Indicates The first scratch The probability that a branch line belongs to an expansion crack; Indicates The first scratch On the branch line The abnormality degree of each pixel; Indicates The first scratch On the branch line The abnormality degree of each pixel; Indicates The first scratch The number of all pixels on a branch line; Indicates The first scratch The variance of all data in the first-order difference sequence of the abnormality degree sequence corresponding to the branch line; Indicates The first scratch The variance of all data in the chain code sequence corresponding to each branch line; is the normalization function; is an exponential function with a natural constant as its base.
2. According to claim 1, the method for detecting surface defects of low-smoke halogen-free cables based on binocular vision is characterized in that: The specific steps of obtaining the abnormality degree of each pixel point according to the difference between the depth values of each pixel point in the low-smoke halogen-free cable depth map are as follows: Get a baseline in the LSZH cable depth map; According to the depth value of each pixel in the low-smoke halogen-free cable depth map and the depth value of the pixel on the line parallel to the baseline, the abnormality degree of each pixel is obtained.
3. According to claim 2, the method for detecting surface defects of low-smoke halogen-free cables based on binocular vision is characterized in that: The specific steps of obtaining the baseline in the low-smoke zero-halogen cable depth map are as follows: In the low-smoke halogen-free cable depth map, the Canny edge detection algorithm is used to process the cable depth map to obtain the edge detection result of the cable surface; the Hough transform line detection algorithm is used to extract the edge lines on both sides of the cable from the edge detection results of the cable surface; and the longer edge line on both sides is used as the baseline.
4. According to claim 2, the method for detecting surface defects of low-smoke halogen-free cables based on binocular vision is characterized in that: The specific formula for obtaining the abnormality degree of each pixel point according to the depth value of each pixel point in the low-smoke halogen-free cable depth map and the depth value of the pixel point on the line parallel to the baseline is as follows: In the formula, Indicates the depth diagram of low smoke zero halogen cable The abnormality degree of each pixel; Indicates the depth diagram of low smoke zero halogen cable The depth value of each pixel; In the depth diagram of low smoke zero halogen cable, The pixel points on the line parallel to the reference line are Pixels outside The depth value of each pixel; In the depth diagram of low smoke zero halogen cable, The pixel points on the line parallel to the reference line are The number of all pixels except pixels; is the normalization function.
5. According to claim 1, the method for detecting surface defects of low-smoke halogen-free cables based on binocular vision is characterized in that: The method of selecting a plurality of scratches and a plurality of suspected expansion crack regions of each scratch from the suspected scratch region according to the abnormality of the pixel points in each suspected scratch region includes the following specific steps: The average value of the abnormality of all pixels in each suspected scratch area is recorded as the scratch possibility of each suspected scratch area; The scratch probability is greater than the preset threshold The suspected scratch area is recorded as a scratch; On the boundary of any scratch, if the degree of abnormality in the eight neighborhoods of any pixel point is greater than the preset threshold The eight neighborhoods are recorded as the suspected expansion crack areas of any scratch.
6. The method for detecting surface defects of low-smoke halogen-free cables based on binocular vision according to claim 1, characterized in that: The specific steps of obtaining a plurality of branch lines of each scratch according to each scratch and the suspected expansion crack area of each scratch are as follows: A morphological refinement operation is performed on each scratch and the area formed by all suspected expansion crack areas to obtain the skeleton line corresponding to each scratch; the medial axis extraction algorithm is used to extract the medial axis of the skeleton line corresponding to each scratch to obtain a medial axis and several branch lines.
7. The method for detecting surface defects of low-smoke halogen-free cables based on binocular vision according to claim 1, characterized in that: The method of obtaining a plurality of expansion cracks of each scratch according to the probability that each branch line of each scratch belongs to an expansion crack comprises the following specific steps: When The first scratch The probability that a branch line belongs to an expansion crack is greater than the preset threshold When The first scratch The branch line is A scratch expansion crack.
8. The method for detecting surface defects of low-smoke halogen-free cables based on binocular vision according to claim 1, characterized in that: The possibility of each scratch expansion is obtained according to the expansion crack of each scratch and the abnormal degree of the pixel point of the expansion crack, and the specific formula included is as follows: In the formula, Indicates The possibility of scratch expansion; Indicates The first scratch The length of a propagating crack; Indicates The number of all propagating cracks in a scratch; Indicates The first scratch The first expansion crack The abnormality degree of each pixel; Indicates The first scratch The number of all pixels of a propagating crack; is the normalization function.
9. The method for detecting surface defects of low-smoke halogen-free cables based on binocular vision according to claim 1, characterized in that: The method of judging whether each scratch is likely to deteriorate in the future based on the possibility of each scratch expanding includes the following specific steps: When The probability of scratch expansion is greater than the preset threshold When A scratch may worsen over time.
