Cable shielding layer braiding defect detection method and system

Abstract

The invention belongs to the technical field of image processing, and particularly relates to a cable shielding layer braiding defect detection method and system, and the method comprises the steps: obtaining a to-be-detected texture image, the total column number of image blocks, a local texture angle, a local braiding frequency, a global dominant angle, a global angle dispersion, and a global average frequency; calculating a knitting disorder index; obtaining texture feature data of the image sub-blocks, and calculating background statistical data of the image sub-blocks; calculating a light and shadow energy abnormal factor; obtaining a suspected defect area and attribute data thereof; calculating the shielding effectiveness damage degree; determining a final defect; physical marking is performed on the final defect and a digitized quality file is generated. According to the method, the technical problems that the texture features of the cable weaving defects are difficult to extract from complex shadow noise and the damage degree of the weaving defects is not evaluated in the prior art are effectively solved.

Description

Technical Field

[0001] This invention relates to the field of image processing technology. More specifically, this invention relates to a method and system for detecting defects in the braiding of cable shielding layers. Background Technology

[0002] In cable manufacturing, the metal braided shielding layer is a key structure for ensuring the cable's electromagnetic interference resistance. The braided layer is typically made of multiple strands of tinned copper wire or aluminum-magnesium alloy wire, cross-woven at specific angles and densities around the insulated core wire. Defects that often occur during the braiding process, such as broken wires, loose strands, uneven braiding density, and knotting and pilling, can severely damage the integrity of the shielding layer, leading to electromagnetic leakage or signal interference during cable use.

[0003] Existing detection methods mainly rely on manual visual inspection or basic machine vision technology. Manual inspection suffers from high labor intensity, visual fatigue leading to missed detections, and is unsuitable for high-speed production lines. Existing machine vision methods mostly employ grayscale threshold segmentation or standard template matching algorithms. However, due to the highly reflective properties of the metal braided layer and its cylindrical curved surface distribution, light shining on the metal wire surface creates complex random highlights and shadows, resulting in an extremely low image signal-to-noise ratio.

[0004] The visual interference caused by the reflection of metal wires and the curved geometry directly leads to the failure of traditional algorithms. Simple grayscale thresholds cannot distinguish between normal metal wire reflections and bright spots caused by broken wires, and fixed template matching cannot adapt to the stretching and deformation of the braided texture caused by slight vibrations or tension changes during cable movement. This makes the detection system either overly sensitive to noise and generate a large number of false alarms, or unable to identify minute structural defects. Therefore, existing technologies cannot effectively extract deep feature information that can characterize the integrity of the braided structure from complex metal texture backgrounds, thus hindering the accurate control of cable shielding quality. Summary of the Invention

[0005] To address the problem that existing machine technologies struggle to extract texture features of cable braiding defects from complex light and shadow noise when inspecting cable braided shielding layers with high reflectivity and cylindrical curve characteristics, and lack the technical ability to assess the degree of damage caused by braiding defects, the present invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a method for detecting defects in the braiding of a cable shielding layer, comprising: Cable images are acquired and unfolded using an industrial linear array camera. After the images are segmented, the geometric distribution features of the braided layer surface are extracted to obtain the texture image to be tested, the total number of columns in the image segment, the local texture angle, the local braid frequency, the global dominant angle, the global angular dispersion, and the global average frequency. The braid disorder index is calculated based on the deviation and abrupt change of the local texture angle and local braid frequency relative to the global statistical benchmark, respectively. Based on the second moment of the gray-level co-occurrence matrix angle of each image sub-block, the texture feature data of the image sub-block is obtained. The background statistics of the image sub-block are calculated through the pixel gray-level value distribution of the image sub-block. The light and shadow energy anomaly factor is calculated based on the relative intensity of the texture and the degree of background noise represented by the texture feature data and the background statistics of the image sub-block. Suspected abnormal sub-blocks are obtained, and the suspected defect areas and their attribute data are obtained based on the spatial adjacency relationship and feature statistics of the sub-blocks. Based on the attribute data of the suspected defect areas, the shielding effectiveness damage degree is calculated according to the comprehensive severity of the defect and the influence of the spatial scale. The suspected defect areas are sorted according to the magnitude of the shielding effectiveness damage degree to determine the final defect. The final defect is physically marked and a digital quality file is generated.

[0007] This invention addresses detection interference caused by cable surface reflection and unique shapes. Through information collection and analysis, it accurately extracts effective information, eliminates irrelevant interference, and reduces misjudgments and omissions due to reflection or shape issues. To address the difficulty of distinguishing between normal deformation and defects using traditional methods, it compares local and overall features and constructs reasonable judgment criteria to accurately identify true defects, avoiding misjudging normal changes as defects. For subtle defects that are difficult to detect or whose extent is difficult to define, comprehensive feature analysis and region merging techniques can capture subtle problems and reconstruct the full picture of the defect. Simultaneously, it solves the problem of traditional detection lacking assessment of defect severity. By comprehensively calculating the severity and extent of defects, it provides a scientific hazard assessment, making quality control more targeted. Furthermore, through physical marking and quality archives, it achieves a complete closed loop from detection to processing and traceability, further changing the current situation of low efficiency, poor accuracy, and insufficient practicality of traditional detection, making cable defect detection more accurate, efficient, and reliable.

[0008] Preferably, the texture image to be tested, the total number of columns in the image block, the local texture angle, the local weave frequency, the global dominant angle, the global angle dispersion, and the global average frequency are obtained, including: A two-dimensional planar image of the cable is acquired and denoted as the texture image to be tested. The texture image to be tested is divided into multiple non-overlapping image sub-blocks, and the number of image sub-blocks in the horizontal direction is recorded to obtain the total number of columns of the image blocks. The sparse orientation histogram of each image sub-block is calculated, and the angle corresponding to the maximum gradient direction is extracted to obtain the local texture angle. At the same time, Fourier transform is performed on the image sub-blocks to obtain the frequency probability of the highest energy in the local area, which is the local weaving frequency. The local texture angles of all image sub-blocks are counted and the mode is taken to obtain the global dominant angle. The standard deviation of all local texture angles is calculated to obtain the global angle dispersion. The mean of the local weaving frequency of all sub-blocks is calculated to obtain the global average frequency.

[0009] Preferably, the weaving disorder index satisfies the following expression: ; In the formula, Indicates the first The weaving disorder index of each image sub-block, dimensionless; For the first Local texture angles of image sub-blocks, in degrees; The dominant angle globally; The global angular dispersion; For the first Local weaving frequency of an image sub-block, in lines / mm; The global average frequency; It is an absolute value function; It is the natural logarithm function; It is the hyperbolic tangent function; , It is a very small positive number, and the denominator is guaranteed to be non-zero.

[0010] This invention calculates a braiding disorder index, which can accurately distinguish between normal morphological changes and genuine defects in cables. By comparing key features of local and overall components, it effectively filters out interference caused by slight cable movement or bending, preventing normal changes from being misjudged as defects. Simultaneously, it can keenly capture and amplify relevant signals for significant deviations caused by structural anomalies, making defect characteristics more prominent, reducing misjudgments and omissions, and making defect assessment more targeted and accurate.

[0011] Preferably, based on the second moment of the gray-level co-occurrence matrix of each image sub-block, texture feature data of the image sub-block is obtained, and background statistics of the image sub-block are calculated through the pixel gray-level value distribution of the image sub-block, including: Calculate the gray-level co-occurrence matrix of each image sub-block, extract the contrast features of the matrix based on texture analysis, and denot it as local contrast; extract the second angular moment of the matrix, denoted as local energy value, and calculate the arithmetic mean of the local energy values ​​of all sub-blocks in the whole image, denoted as global average energy; calculate the average pixel gray-level value in each image sub-block of the texture image under test, denoted as local gray-level mean; statistically analyze the overall pixel gray-level distribution of the texture image under test, and calculate the pixel gray-level mean and standard deviation of the whole image, denoted as global gray-level mean and global gray-level standard deviation, respectively.

[0012] Preferably, the light and shadow energy anomaly factor satisfies the following expression: ; In the formula, Indicates the first The light and shadow energy anomaly factor of each image sub-block is dimensionless. For the first Local contrast of image sub-blocks; For the first Local energy values ​​of image sub-blocks; The global average energy; For the first The local grayscale mean of each image sub-block; The global grayscale mean; The global grayscale standard deviation; It is an absolute value function; , For a very small positive number, the denominator must not be 0; It is an exponential function with the natural constant e as its base.

[0013] This invention helps eliminate some interference from reflections on cable surfaces, improving the accuracy of identifying true defects. By combining the degree of texture irregularity and brightness differences, it reduces the probability of misjudging normal reflective points as defects, while also minimizing the chance of missing actual problems due to reflections masking the underlying issues. This method can distinguish between normal reflections and defective reflections, improving the reliability of detection results.

[0014] Preferably, obtaining suspected defect areas and their attribute data includes: Calculate the mean value of the light and shadow energy anomaly factor of all image sub-blocks in the texture image to be tested. Screen out image sub-blocks whose light and shadow energy anomaly factor exceeds the mean value of the light and shadow energy anomaly factor and record them as suspected anomaly sub-blocks. Use a connectivity analysis algorithm to cluster and merge them to obtain suspected defect regions. Count the total number of image sub-blocks contained in each suspected defect region and record it as the number of region coverage blocks. Calculate the weaving disorder index and the mean value of the light and shadow energy anomaly factor of all image sub-blocks in a single suspected defect region and record them as the region average disorder degree and the region average anomaly degree.

[0015] Preferably, the degree of shielding effectiveness impairment satisfies the following expression: ; In the formula, Indicates the first The degree of shielding effectiveness impairment in a suspected defective area is dimensionless. For the first The average disorder level of the suspected defective areas; For the first The average anomaly of the suspected defective areas; For the first Number of area coverage blocks for each suspected defective area; This represents the total number of columns in the image blocks. For a very small positive number, the denominator must not be 0; It is the natural logarithm function.

[0016] This invention helps assess the impact of defects on cable performance. It not only focuses on the severity of the defect itself but also considers its distribution and scale, making the assessment results more comprehensive and objective. This assessment method can distinguish the severity of different defects, avoiding overemphasis on minor, harmless issues while reducing the neglect of serious defects with a wide range and significant impact. It provides a basis for prioritizing key issues and helps improve the rationality of quality control.

[0017] Preferably, determining the final defect includes: All suspected defect areas in the texture image to be tested are placed into a candidate list and sorted from largest to smallest shielding effectiveness damage. An alarm threshold curve is generated by piecewise threshold fitting. The candidate list is traversed, and suspected defect areas with shielding effectiveness damage greater than the alarm threshold curve are marked as final defects. Their coordinate positions in the texture image to be tested and their corresponding shielding effectiveness damage are recorded.

[0018] This invention determines the final defects through sorting and threshold filtering, which helps to improve the accuracy and efficiency of defect identification. It sorts defects according to their severity, prioritizing the identification of critical defects with greater impact, while filtering out irrelevant interference through reasonable thresholds, reducing processing chaos caused by an excessive number of test results. This approach allows staff to focus on key issues more quickly, reducing time spent sifting through invalid information, thus improving detection efficiency while ensuring the accuracy of defect identification, balancing detection effectiveness and work efficiency.

[0019] Preferably, physical marking is performed on the final defects and a digital quality profile is generated, including: Based on the coordinates of the final defect in the texture image to be tested, combined with the cable movement speed and the physical distance from the camera to the marking machine, the delayed trigger time is calculated, and a trigger signal is sent to the inkjet printer. When the cable passes the marking point, inkjet printing or labeling is performed on the physical location of the final defect. The statistical data of all final defects in this batch of inspections are summarized to generate a quality inspection report. The report includes a defect distribution map, defect type statistics, maximum shielding effectiveness damage degree, and detailed coordinates of each defect.

[0020] Secondly, the present invention provides a cable shield braiding defect detection system, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned cable shield braiding defect detection method is implemented.

[0021] By adopting the above technical solution, a computer program for detecting defects in cable shielding layer braiding is generated and stored in a memory for loading and execution by a processor. This allows for the creation of a terminal device based on the memory and processor, facilitating its use.

[0022] The beneficial effects of this invention are as follows: Based on the actual needs of industrial production, this invention provides a relatively comprehensive and efficient solution for cable quality control, possessing significant macro-level value. In terms of production efficiency, it reduces reliance on manual inspection, promotes the automation and precision of inspection, helps improve inspection speed, can adapt to the needs of some high-speed production lines, reduces waiting time in the production process, and contributes to improving overall production efficiency. In terms of quality control, by accurately identifying defects and scientifically assessing hazards, it can filter out some unqualified products, reducing the risk of cables with quality problems entering the market, helping to improve the overall quality level of products, and positively impacting product competitiveness in the market. In terms of industry development, it provides ideas for upgrading cable testing technology, offers a reference example for quality control in related industries, and helps promote the overall improvement of industry quality levels. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating a method for detecting defects in the braiding of a cable shielding layer according to the present invention; Figure 2 This is a schematic thermal distribution diagram illustrating the braiding disorder index of a cable shielding layer braiding defect detection method according to the present invention; Figure 3 This is a schematic diagram illustrating the thermal distribution of the light and shadow energy anomaly factor in a cable shield braiding defect detection method according to the present invention; Figure 4This diagram schematically illustrates the evaluation and screening results of shielding effectiveness impairment in suspected defect areas of a cable shielding layer braiding defect detection method according to the present invention. Detailed Implementation

[0024] This invention discloses a method for detecting defects in the braiding of cable shielding layers, referring to... Figure 1 This includes steps S1-S4: S1: Acquire and unfold cable images using an industrial linear array camera, divide the image into blocks and extract the geometric distribution features of the braided layer surface to obtain the texture image to be tested, the total number of columns in the image blocks, the local texture angle, the local braid frequency, the global dominant angle, the global angle dispersion, and the global average frequency.

[0025] It should be noted that in cable braiding production, the cables are in high-speed axial motion accompanied by random radial rotation. Traditional area array cameras struggle to capture complete cylindrical textures, and due to the cylindrical geometry of the wires, reflected light from the edges is severely distorted. Directly processing the raw image leads to an extremely high false detection rate. From a signal processing perspective, the braided layer is essentially a highly periodic texture structure. Any physical defects, such as broken wires or loose strands, will disrupt this periodicity in both the frequency and spatial domains. Therefore, simply acquiring the image is insufficient; the visual information of the physical space must be mapped to the feature space. By extracting texture angles through histograms of directional gradients, the direction of the braided wires can be captured. By extracting frequencies through Fourier transforms, the density of the braid can be calculated. This angle + frequency feature extraction can filter out low-frequency noise caused by uneven lighting, directly locking onto the skeletal features of the braided structure and providing a clean data benchmark for subsequent differential analysis.

[0026] Specifically, an industrial linear array camera is used to acquire and unfold cable images. After the images are divided into blocks, the geometric distribution features of the braided layer surface are extracted to obtain the texture image to be tested, the total number of columns in the image blocks, the local texture angle, the local braid frequency, the global dominant angle, the global angle dispersion, and the global average frequency, including: A high-frequency linear array camera is used to scan and acquire cable images. The cable boundary is located by edge detection to obtain a two-dimensional planar image of the cable, which is denoted as the texture image to be tested. The texture image to be tested is divided into multiple non-overlapping image sub-blocks, and the number of image sub-blocks in the horizontal direction is recorded to obtain the total number of columns of the image blocks. The sparse orientation histogram of each image sub-block is calculated, and the angle corresponding to the maximum gradient direction is extracted to obtain the local texture angle. At the same time, Fourier transform is performed on the image sub-blocks to obtain the frequency probability of the highest energy in the local area, which is the local weaving frequency. The local texture angles of all image sub-blocks are counted and the mode is taken to obtain the global dominant angle. The standard deviation of all local texture angles is calculated to obtain the global angle dispersion. The mean of the local weaving frequency of all sub-blocks is calculated to obtain the global average frequency.

[0027] Thus, the test texture image, total number of columns in the image block, local texture angle, local braid frequency, global dominant angle, global angle dispersion, and global average frequency of the cable under test were obtained.

[0028] S2: Calculate the weaving disorder index based on the deviation and abrupt change of the local texture angle and local weaving frequency relative to the global statistical benchmark, respectively; obtain the texture feature data of the image sub-block based on the second moment of the gray-level co-occurrence matrix angle of each image sub-block; calculate the background statistics of the image sub-block through the pixel gray-level value distribution of the image sub-block; calculate the light and shadow energy anomaly factor based on the relative intensity of texture and the degree of background noise represented by the texture feature data and background statistics of the image sub-block.

[0029] It should be noted that slight bending and jitter are unavoidable physical phenomena during the transmission of flexible cables. This leads to natural geometric deflections of local texture angles. However, true manufacturing defects, such as strand breakage and stacking, not only manifest as abnormalities in local texture angles but also, more fundamentally, as abrupt changes in local material density. If only the angle difference of local texture angles is used for judgment, any normal cable sway will trigger false alarms; if only local material density is used for judgment, defects with normal density but incorrect braiding direction will be missed. From a topological perspective, normal bending involves a change in distance while maintaining the local density; defects, on the other hand, remain constant but are accompanied by drastic fluctuations in energy density. Therefore, this invention constructs a nonlinear model that uses density abrupt changes as a judgment condition. Only when the density changes significantly will the angle deviation be recognized as a defect signal. This effectively decouples flexible deformation from structural damage and effectively solves the problem of high false alarm rates in dynamic scenarios.

[0030] Specifically, the weaving disorder index is calculated based on the degree of deviation and abrupt change of the local texture angle and local weaving frequency relative to the global statistical benchmark, including: The knitting disorder index satisfies the following expression: ; In the formula, Indicates the first The weaving disorder index of each image sub-block, dimensionless; For the first Local texture angles of image sub-blocks, in degrees; The dominant angle globally; The global angular dispersion; For the first Local weaving frequency of an image sub-block, in lines / mm; The global average frequency; It is an absolute value function; It is the natural logarithm function; It is the hyperbolic tangent function; , It is a very small positive number, and the denominator is guaranteed to be non-zero.

[0031] In the formula, An adaptive baseline was constructed using global statistical features. This term only increases significantly when the local texture angle deviates from the global dominant angle and the degree of deviation exceeds the natural fluctuation range of the overall image and the global angle dispersion. This automatically filters out the overall angle changes caused by the flexible bending of the cable. It characterizes the degree of abrupt change in local density; The density abrupt change is normalized to a weighting factor between 1 and 2 using the hyperbolic tangent function, and the angle deviation term of the local texture angle is nonlinearly amplified. Using local density mutations as amplifiers, when a region experiences both angular distortion and geometric deformation, along with significant density mutations and uneven physical distribution, the exponent is nonlinearly amplified.

[0032] For example, , , Scenario 1: The cable is bent normally. But the first The local weaving frequency of each image sub-block is normal, that is... At this point, the density term is 1, then ; Scenario 2 exists, where the defect of scattered shares occurs. , At this point, the density term is close to 1.66, then ; , All values ​​are rounded to two decimal places.

[0033] It should be noted that at the micro-texture level, simple geometric features are insufficient to describe all defect types, especially for subtle imperfections such as pilling or burrs. These imperfections may not significantly change the weaving angle or density macroscopically, but they leave traces in the roughness of the micro-texture and the pixel neighborhood relationship. The gray-level co-occurrence matrix, as a second-order statistical tool, can calculate the fineness and roughness of the texture from the perspective of pixel pairs. Although normal metal weave layers are reflective, their texture has a high degree of repeatability and uniformity. However, in defective areas, due to the breakage or disordered stacking of metal wires, this uniformity is broken, manifesting as drastic fluctuations in gray-level differences between neighboring pixels. Therefore, this invention constructs a set of micro-features based on the optical properties of the material by calculating the features of the gray-level co-occurrence matrix and the global gray-level distribution, providing data support for subsequently separating real physical damage from complex light and shadow.

[0034] Preferably, based on the gray-level co-occurrence matrix and angular second moment of each image sub-block, texture feature data of the image sub-block is obtained, and background statistics of the image sub-block are calculated through the pixel gray-level value distribution of the image sub-block, including: Calculate the gray-level co-occurrence matrix of each image sub-block, extract the contrast features of the matrix based on texture analysis, and denot it as local contrast; extract the second angular moment of the matrix, denoted as local energy value, and calculate the arithmetic mean of the local energy values ​​of all sub-blocks in the whole image, denoted as global average energy; calculate the average pixel gray-level value in each image sub-block of the texture image under test, denoted as local gray-level mean; statistically analyze the overall pixel gray-level distribution of the texture image under test, and calculate the pixel gray-level mean and standard deviation of the whole image, denoted as global gray-level mean and global gray-level standard deviation, respectively.

[0035] Thus, the local contrast, local energy value, global average energy, local gray-level mean, global gray-level mean, and global gray-level standard deviation of each image sub-block are obtained.

[0036] It should be noted that visual inspection of metal surfaces often faces the challenge of specular interference. A high-brightness pixel could be a specular reflection from a perfectly intact metal wire at a specific angle, or it could be a strongly reflective point formed by a broken, warped metal wire. From a physical optics perspective, while normal specular points are bright, their surrounding texture structure is ordered; however, specular points caused by broken wires are often accompanied by a rough scattering surface at the break, leading to an increase in local texture entropy. Therefore, this invention introduces a light and shadow energy anomaly factor, using microscopic texture disorder as a weighting term to modulate macroscopic brightness deviation. The purpose of calculating the light and shadow energy anomaly factor expression is to ensure that only when an area is abnormally bright and has a significant degree of texture disorder will it be judged as a defect, thus effectively solving the false detection problem caused by metal reflection.

[0037] Specifically, based on the texture feature data of image sub-blocks and the relative intensity of texture and background noise level represented by background statistics, a light and shadow energy anomaly factor is calculated, including: The light and shadow energy anomaly factor satisfies the following expression: ; In the formula, Indicates the first The light and shadow energy anomaly factor of each image sub-block is dimensionless. For the first Local contrast of image sub-blocks; For the first Local energy values ​​of image sub-blocks; The global average energy; For the first The local grayscale mean of each image sub-block; The global grayscale mean; The global grayscale standard deviation; It is an absolute value function; , For a very small positive number, the denominator must not be 0; It is an exponential function with the natural constant e as its base.

[0038] In the formula, In the middle, molecules This describes the microscopic complexity of the texture; the broken edges of broken fibers are typically rough and reflective, leading to an increased contrast-energy product; the denominator... Microscopic features are normalized using global average energy; exponential term As a gain coefficient, this term increases exponentially when the brightness of a local area deviates significantly from the background distribution, such as an extremely bright broken wire reflection point. The physical meaning is that only when the texture of a region is extremely messy and its brightness is out of place with its surrounding background is it considered a broken wire defect.

[0039] For example, Scenario 1: Normal metallic highlights. It is 2, but the texture is regular. ,but Scenario 2: Reflective spots caused by broken wires. The value is 2, but the texture at the fracture surface is messy. ,but ; Round to two decimal places. Round to one decimal place.

[0040] Thus, the light and shadow energy anomaly factor of the image sub-block was obtained.

[0041] S3: Obtain suspected abnormal sub-blocks, and based on the spatial adjacency relationship and characteristic statistical values ​​of the sub-blocks, obtain suspected defect areas and their attribute data; based on the attribute data of the suspected defect areas, calculate the shielding effectiveness impairment degree according to the comprehensive severity of the defect and the influence of spatial scale.

[0042] It should be noted that in discretized image processing, defects are often segmented into isolated pixel blocks or sub-blocks. However, in actual industrial evaluation systems, a defect is a continuous physical object, not discrete data points. For example, a continuous scratch might be identified by an algorithm as a series of discontinuous anomalies. If only a single point is evaluated, it might be ignored because its feature value does not reach the peak value. Therefore, this invention introduces connected component clustering analysis to restore the complete shape of the defect, restoring data features to physical entities. Through neighborhood connectivity analysis, spatially adjacent suspected anomaly blocks are merged into a whole, and the average attribute is calculated using this whole as the unit. This not only smooths out randomly generated isolated noise points but also lays a data foundation for subsequent hazard assessment based on defect size.

[0043] Preferably, suspected abnormal sub-blocks are obtained, and based on the spatial adjacency relationship and characteristic statistical values ​​of the sub-blocks, suspected defective regions and their attribute data are obtained, including: Calculate the mean value of the light and shadow energy anomaly factor of all image sub-blocks in the texture image to be tested. Screen out image sub-blocks whose light and shadow energy anomaly factor exceeds the mean value of the light and shadow energy anomaly factor and record them as suspected anomaly sub-blocks. Use a connectivity analysis algorithm to cluster and merge them to obtain suspected defect regions. Count the total number of image sub-blocks contained in each suspected defect region and record it as the number of region coverage blocks. Calculate the weaving disorder index and the mean value of the light and shadow energy anomaly factor of all image sub-blocks in a single suspected defect region and record them as the region average disorder degree and the region average anomaly degree.

[0044] Thus, we have obtained the suspected defective areas, as well as the attribute data of the suspected defective areas, including the number of regional coverage blocks, the average disorder degree of the region, and the average anomaly of the region.

[0045] It should be noted that the core function of cable shielding is electromagnetic compatibility protection. Its failure modes are closely related to the geometry of defects. Small but densely packed holes have limited impact on high-frequency signal leakage, while narrow, elongated gaps can cause severe electromagnetic leakage, much like an antenna. That is, a defect that, although locally textured, is very long may be far more harmful than a localized, highly disordered but small-area noise point. Therefore, the evaluation criteria for cable defects cannot be limited to visual resemblance to defects, but need to be elevated to whether they affect shielding effectiveness. This invention introduces a shielding effectiveness impairment degree, which uses a logarithmic function to model defect size. This amplifies the weight of long-distance defects while avoiding numerical overflow caused by ultra-large area defects, thus outputting a comprehensive index that truly reflects the risk of process quality.

[0046] Preferably, based on the attribute data of the suspected defective area, the shielding effectiveness impairment is calculated according to the overall severity of the defect and its spatial scale, including: The degree of shielding effectiveness impairment satisfies the following expression: ; In the formula, Indicates the first The degree of shielding effectiveness impairment in a suspected defective area is dimensionless. For the first The average disorder level of the suspected defective areas; For the first The average anomaly of the suspected defective areas; For the first Number of area coverage blocks for each suspected defective area; This represents the total number of columns in the image blocks. For a very small positive number, the denominator must not be 0; It is the natural logarithm function.

[0047] In the formula, By using vector magnitude, macroscopic structural defects and microscopic material flaws are integrated to ensure that regardless of the severity of any type of defect, it will be reflected as an increase in the basic risk value. It reflects the relative proportion of defects in the direction perpendicular to the cable axis, that is, how much of the cable the defects surround. As a spatial gain factor, it indicates that the longer and larger the defect, the higher the risk, but its growth conforms to the marginal effect. The system focuses not only on the severity of the defect but also on its magnitude; even minor structural deformations, if continuous, can lead to… Increase.

[0048] For example, Scenario 1: An isolated, very intense highlight noise. However, there is only 1 point. ,but Scenario 2: A long stretch of slightly uneven cabling with low strength. However, the continuous length reaches 50 points. ,but As can be seen, this index effectively suppresses isolated noise and amplifies real process defects with spatial continuity. Round to three decimal places. Round to two decimal places.

[0049] S4: Sort suspected defective areas according to the degree of damage to shielding effectiveness, determine the final defects; perform physical marking on the final defects and generate digital quality files.

[0050] It should be noted that in large-scale continuous industrial production, absolute zero defects are often uneconomical and unrealistic. Production managers are more concerned with critical defects that could lead to product recalls or safety accidents. If the detection system indiscriminately alarms for all minor deviations, it will cause frequent production line downtime, severely impacting production efficiency. Furthermore, different application scenarios have different requirements for shielding effectiveness. Therefore, this invention cannot simply output a binary judgment of whether a defect exists; it also needs to provide a risk ranking. By sorting all detected suspected defect areas in descending order of value and truncating them using a set threshold, the system can achieve a priority ranking and intelligent screening mechanism. This mechanism ensures that, with limited maintenance resources, the most dangerous defects are addressed first, achieving an effective balance between quality control and production efficiency.

[0051] Specifically, suspected defective areas are sorted according to the degree of damage to shielding effectiveness to determine the final defects, including: All suspected defect areas in the texture image to be tested are placed into a candidate list and sorted from largest to smallest shielding effectiveness damage. An alarm threshold curve is generated by piecewise threshold fitting. The candidate list is traversed, and suspected defect areas with shielding effectiveness damage greater than the alarm threshold curve are marked as final defects. Their coordinate positions in the texture image to be tested and their corresponding shielding effectiveness damage are recorded.

[0052] At this point, the final list of defects and its sorting results were obtained.

[0053] It's important to note that the ultimate goal of inspection is closed-loop control. Simply displaying defects on the software interface is far from sufficient, because cable production is continuous. By the time a defect appears on the screen, the corresponding physical cable segment may already be deeply wound into the take-up reel, making it difficult to locate. Therefore, it's essential to establish an instantaneous mapping between digital detection and the physical entity. By controlling inkjet printers or marking machines at the end of the production line, markings are printed at the corresponding locations on the physical cable within milliseconds of defect detection. This allows downstream operators to visually identify and reject defective products. Simultaneously, detailed digital quality reports are generated, establishing a digital quality file for each reel of cable. This is crucial for fields with extremely high traceability requirements, such as aviation and nuclear power.

[0054] Preferably, physical marking is performed on the final defects and a digital quality profile is generated, including: Based on the coordinates of the final defect in the texture image to be tested, combined with the cable movement speed and the physical distance from the camera to the marking machine, the delayed trigger time is calculated, and a trigger signal is sent to the inkjet printer. When the cable passes the marking point, inkjet printing or labeling is performed on the physical location of the final defect. The statistical data of all final defects in this batch of inspections are summarized to generate a quality inspection report. The report includes a defect distribution map, defect type statistics, maximum shielding effectiveness damage degree, and detailed coordinates of each defect.

[0055] It should be noted that, Figure 2 This is a heat map of the braiding disorder index. The low grayscale background area represents a normal braiding structure, while the highlighted area represents a highly abnormal braiding structure. The map marks two typical areas: one is an area where the cable is normally bent and deformed, and the other is an area where there is an actual loose strand defect. The difference in their severity is shown by the change in the grayscale level.

[0056] Figure 3 The thermal distribution map represents the abnormal light and shadow energy factors. The map marks the normal metal reflection points and the broken wire reflection points. By displaying the contrast of gray levels, the difference between the two different reflection types in the abnormal manifestation is shown.

[0057] Figure 4 The graph shows the results of the assessment and screening of the shielding effectiveness impairment of suspected defect areas. The horizontal axis represents the spatial scale of the suspected defect area, and the vertical axis represents the local severity of the suspected defect area. The dashed line in the graph is the alarm threshold curve, and the points above the curve represent the final defects that are identified as requiring special attention.

[0058] This completes the detection of defects in the cable shielding layer braid.

[0059] This invention also discloses a cable shield braiding defect detection system, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a cable shield braiding defect detection method according to the present invention.

[0060] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0061] While this specification has shown and described numerous embodiments of the invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in the practice of this invention.

Claims

1. A method for detecting defects in the braiding of a cable shielding layer, characterized in that, include: The cable image is acquired and unfolded using an industrial linear array camera. After the image is divided into blocks, the geometric distribution features of the braided layer surface are extracted to obtain the texture image to be tested, the total number of columns in the image block, the local texture angle, the local braid frequency, the global dominant angle, the global angle dispersion, and the global average frequency. The weaving disorder index is calculated based on the deviation and abrupt change of the local texture angle and local weaving frequency relative to the global statistical benchmark, respectively. The texture feature data of the image sub-block is obtained based on the second moment of the gray-level co-occurrence matrix angle of each image sub-block. The background statistics of the image sub-block are calculated through the pixel gray-level value distribution of the image sub-block. The light and shadow energy anomaly factor is calculated based on the relative texture intensity and background noise level represented by the texture feature data and background statistics of the image sub-block. Obtain suspected abnormal sub-blocks, and based on the spatial adjacency relationship and characteristic statistical values ​​of the sub-blocks, obtain suspected defect areas and their attribute data; based on the attribute data of the suspected defect areas, calculate the shielding effectiveness impairment degree according to the overall severity of the defect and the influence of spatial scale. Suspected defective areas are sorted according to the degree of damage to shielding effectiveness to determine the final defects; Physically mark the final defects and generate a digital quality profile.

2. The method for detecting defects in cable shielding braiding according to claim 1, characterized in that, The obtained texture image to be tested, total number of columns in image blocks, local texture angle, local weave frequency, global dominant angle, global angle dispersion, and global average frequency include: A two-dimensional planar image of the cable is acquired and denoted as the texture image to be tested. The texture image to be tested is divided into multiple non-overlapping image sub-blocks, and the number of image sub-blocks in the horizontal direction is recorded to obtain the total number of columns of the image blocks. The sparse orientation histogram of each image sub-block is calculated, and the angle corresponding to the maximum gradient direction is extracted to obtain the local texture angle. At the same time, Fourier transform is performed on the image sub-blocks to obtain the frequency probability of the highest energy in the local area, which is the local weaving frequency. The local texture angles of all image sub-blocks are counted and the mode is taken to obtain the global dominant angle. The standard deviation of all local texture angles is calculated to obtain the global angle dispersion. The mean of the local weaving frequency of all sub-blocks is calculated to obtain the global average frequency.

3. The method for detecting defects in cable shielding braiding according to claim 1, characterized in that, The weaving disorder index satisfies the following expression: ； In the formula, Indicates the first The weaving disorder index of each image sub-block, dimensionless; For the first Local texture angles of image sub-blocks, in degrees; The dominant angle globally; The global angular dispersion; For the first Local weaving frequency of an image sub-block, in lines / mm; The global average frequency; It is an absolute value function; It is the natural logarithm function; It is the hyperbolic tangent function; , It is a very small positive number, and the denominator is guaranteed to be non-zero.

4. The method for detecting defects in cable shielding braiding according to claim 1, characterized in that, The textural feature data of the image sub-block is obtained based on the second moment of the gray-level co-occurrence matrix of each image sub-block. Background statistics of the image sub-block are calculated based on the pixel gray-level value distribution of the image sub-block, including: Calculate the gray-level co-occurrence matrix of each image sub-block, extract the contrast features of the matrix based on texture analysis, and denot it as local contrast; extract the second angular moment of the matrix, denoted as local energy value, and calculate the arithmetic mean of the local energy values ​​of all sub-blocks in the whole image, denoted as global average energy; calculate the average pixel gray-level value in each image sub-block of the texture image under test, denoted as local gray-level mean; statistically analyze the overall pixel gray-level distribution of the texture image under test, and calculate the pixel gray-level mean and standard deviation of the whole image, denoted as global gray-level mean and global gray-level standard deviation, respectively.

5. The method for detecting defects in cable shielding braiding according to claim 1, characterized in that, The light and shadow energy anomaly factor satisfies the following expression: ； In the formula, Indicates the first The light and shadow energy anomaly factor of each image sub-block is dimensionless. For the first Local contrast of image sub-blocks; For the first Local energy values ​​of image sub-blocks; The global average energy; For the first The local grayscale mean of each image sub-block; The global grayscale mean; The global grayscale standard deviation; It is an absolute value function; , For a very small positive number, the denominator must not be 0; It is an exponential function with the natural constant e as its base.

6. The method for detecting defects in cable shielding braiding according to claim 1, characterized in that, The process of obtaining suspected defective regions and their attribute data includes: Calculate the mean value of the light and shadow energy anomaly factor of all image sub-blocks in the texture image to be tested. Screen out image sub-blocks whose light and shadow energy anomaly factor exceeds the mean value of the light and shadow energy anomaly factor and record them as suspected anomaly sub-blocks. Use a connectivity analysis algorithm to cluster and merge them to obtain suspected defect regions. Count the total number of image sub-blocks contained in each suspected defect region and record it as the number of region coverage blocks. Calculate the weaving disorder index and the mean value of the light and shadow energy anomaly factor of all image sub-blocks in a single suspected defect region and record them as the region average disorder degree and the region average anomaly degree.

7. The method for detecting defects in cable shielding braiding according to claim 1, characterized in that, The degree of shielding effectiveness impairment satisfies the following expression: ； In the formula, Indicates the first The degree of shielding effectiveness impairment in a suspected defective area is dimensionless. For the first The average disorder level of the suspected defective areas; For the first The average anomaly of the suspected defective areas; For the first Number of area coverage blocks for each suspected defective area; This represents the total number of columns in the image blocks. For a very small positive number, the denominator must not be 0; It is the natural logarithm function.

8. The method for detecting defects in cable shielding braiding according to claim 1, characterized in that, The determination of the final defect includes: All suspected defect areas in the texture image to be tested are placed into a candidate list and sorted from largest to smallest shielding effectiveness damage. An alarm threshold curve is generated by piecewise threshold fitting. The candidate list is traversed, and suspected defect areas with shielding effectiveness damage greater than the alarm threshold curve are marked as final defects. Their coordinate positions in the texture image to be tested and their corresponding shielding effectiveness damage are recorded.

9. The method for detecting defects in cable shielding braiding according to claim 1, characterized in that, The process of physically marking the final defects and generating a digital quality profile includes: Based on the coordinates of the final defect in the texture image to be tested, combined with the cable movement speed and the physical distance from the camera to the marking machine, the delayed trigger time is calculated, and a trigger signal is sent to the inkjet printer. When the cable passes the marking point, inkjet printing or labeling is performed on the physical location of the final defect. The statistical data of all final defects in this batch of inspections are summarized to generate a quality inspection report. The report includes a defect distribution map, defect type statistics, maximum shielding effectiveness damage degree, and detailed coordinates of each defect.

10. A cable shielding layer braiding defect detection system, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement a method for detecting defects in cable shielding braids according to any one of claims 1-9.

Images