Cable crimping detection method and system based on image processing

Abstract

The invention relates to the technical field of cable crimping detection, in particular to a cable crimping detection method and device based on image processing. The method comprises the following steps: acquiring a multi-angle high-resolution monitoring image of a cable crimping point; performing dynamic background elimination processing on the multi-angle high-resolution monitoring image to generate a background-eliminated cable crimping point monitoring image; carrying out image attitude jitter analysis on the background elimination cable crimping point monitoring image one by one angle, and carrying out attitude deviation correction processing to construct an attitude correction monitoring image; cable crimping area feature point identification is carried out on the attitude correction monitoring image, and three-dimensional point cloud modeling is carried out, so that a crimping point three-dimensional registration optimization model is obtained; and carrying out crimping point defect visual detection on the crimping point three-dimensional registration optimization model, carrying out visual rendering, and constructing a crimping point defect form rendering model. According to the invention, high-efficiency and accurate cable crimping detection is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of cable crimping detection, and in particular to a cable crimping detection method and system based on image processing. Background Art

[0002] In modern industrial automation and intelligent manufacturing, the quality of cable crimping has an important impact on the stability and safety of equipment. As a vital part of electrical connection, cable crimping is widely used in various electronic equipment, mechanical devices, and automated production lines. The quality of cable crimping is directly related to the transmission efficiency of electrical signals and the normal operation of equipment. Any poor crimping or unstable contact may lead to equipment failure, performance degradation, or even safety accidents.

[0003] With the continuous development of automation technology, traditional manual inspection methods can no longer meet the needs of efficient and accurate quality control. Traditional cable crimping detection usually relies on manual visual inspection and manual measurement. This method is not only inefficient, but also has large human errors, making it difficult to fully monitor cable crimping in large-scale production. At the same time, since the crimping process may be affected by many factors, such as the technical level of the operator, the instability of equipment parameters, etc., the quality of cable crimping is difficult to guarantee, and traditional inspection methods usually cause irreversible damage when problems are found. Therefore, with the development of industrial automation and intelligence, cable crimping detection methods based on image processing technology have emerged. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a cable crimping detection method and system based on image processing to solve at least one of the above technical problems.

[0005] To achieve the above object, the present invention provides a cable crimping detection method based on image processing, comprising the following steps: Step S1: Acquire a multi-angle high-resolution monitoring image of a cable crimping point; perform dynamic background elimination processing on the multi-angle high-resolution monitoring image to generate a background-eliminated cable crimping point monitoring image; Step S2: performing posture jitter analysis of each angle of the background-eliminated cable crimping point monitoring image, and performing posture deviation correction processing to construct a posture correction monitoring image; Step S3: identifying feature points of the cable crimping area on the posture correction monitoring image, and performing three-dimensional point cloud modeling, thereby obtaining a three-dimensional registration optimization model of the crimping point; Step S4: performing visual inspection of crimping point defects on the crimping point three-dimensional registration optimization model, and performing visual rendering to construct a crimping point defect morphology rendering model; Step S5: quantifying the degree of deformation of the defect area of ​​the crimping point defect morphology rendering model and performing a comprehensive crimping quality assessment to generate a cable crimping quality assessment report; Step S6: Make adaptive correction decisions based on the cable crimping quality assessment report, perform dynamic migration optimization, and build a cable crimping defect optimization model to perform cable crimping detection and diagnosis optimization operations.

[0006] The present invention can obtain detailed images of cable crimping points from different viewing angles through multi-angle and high-resolution monitoring image acquisition, ensuring that all features of the crimping points are fully captured. This method can provide more comprehensive crimping point information and avoid missing crimping defects due to viewing angle problems. Background elimination technology effectively removes background noise and interference in the image to ensure that the image only contains real data of the cable crimping points. This processing can greatly improve the accuracy of subsequent image analysis, avoid the influence of the background environment on image recognition, and ensure the recognition accuracy of feature points and defects. Due to the image posture jitter caused by factors such as camera position deviation or object vibration during multi-angle image acquisition, these errors can be identified by dynamically analyzing the image posture. And enhance image stability and improve the accuracy of subsequent processing. After image correction, the shape of the crimping point is more accurate and stable, which can eliminate the error caused by angle deviation, ensure the standardization of image data, and provide accurate basis for subsequent feature point recognition and 3D modeling, which is crucial to improving the accuracy and reliability of defect detection. Through accurate feature point recognition, the key points of the cable crimping area can be located, providing important reference for subsequent 3D modeling. The extraction of feature points helps to improve the detail recognition ability of the crimping point, especially for the detection of tiny defects. 3D modeling can truly reproduce the 3D structure of the crimping point, providing more intuitive data support for defect detection. 3D point cloud data can accurately represent the shape of the crimping point in space. This allows for a better assessment of its geometry and deformation, providing a more sophisticated model for subsequent defect analysis and quality assessment. Defect detection is performed based on the 3D registration optimization model, so that every subtle defect at the crimping point can be accurately identified. The 3D model can help the algorithm better understand the geometric features of the crimping area, thereby improving the sensitivity of defect detection. Through visualization technology, the defects and morphology of the crimping points can be displayed, which can provide intuitive defect information to inspectors. This visualization method not only improves the accuracy of detection, but also facilitates subsequent quality analysis and decision support. Through quantitative analysis, the degree of deformation of the crimping point can be accurately assessed to determine whether it meets the product standard requirements. The quantification of the degree of deformation provides a quantitative basis for evaluating the crimping quality. The basis makes the test results more objective and scientific. The generated quality assessment report can provide operators and quality management personnel with a comprehensive crimping quality analysis. The report covers the status of the crimping points, defects and potential risks, provides data support for decision-making, and helps to implement effective quality control. Through the adaptive correction algorithm, the crimping process parameters can be dynamically adjusted according to the actual test results. If quality problems or potential defects are detected, the system can adjust the operating procedures in real time and automatically optimize the crimping parameters, thereby improving the stability of the production process and the quality control level. The dynamic optimization model can be adjusted according to real-time data to further optimize the detection algorithm and quality control strategy. This intelligent dynamic adjustment can cope with different production environments and product requirements.Significantly improve production efficiency and product consistency. Through optimized defect detection and diagnosis operations, crimping quality problems can be more effectively identified and repaired. The system can continuously learn and optimize, gradually improve detection accuracy, reduce human intervention, and improve the level of detection automation.

[0007] Preferably, step S1 comprises the following steps: Step S11: Acquire a multi-angle high-resolution monitoring image of the cable crimping point; Step S12: Calculate the light intensity at each angle on the multi-angle high-resolution monitoring image to extract the light intensity parameters at each angle; Step S13: performing multi-angle image brightness difference analysis on the multi-angle high-resolution monitoring image to generate image brightness difference values ​​between multiple angles; Step S14: performing dynamic light source deviation compensation calculation on the image brightness difference values ​​between multiple angles based on the illumination intensity parameter of each angle to generate a multi-angle dynamic light source deviation compensation value; Step S15: optimizing the light source error of the multi-angle high-resolution monitoring image based on the multi-angle dynamic light source deviation compensation value, thereby obtaining a light source error optimized monitoring image; Step S16: Perform dynamic background elimination processing on the light source error optimization monitoring image to generate a background eliminated cable crimping point monitoring image.

[0008] The present invention can obtain detailed information of the crimping point from different viewing angles by acquiring high-resolution monitoring images from multiple angles, ensuring that all key features of the crimping point are captured. Multi-angle image acquisition can avoid blind spots caused by a single angle, thereby improving the comprehensiveness of image information. High-resolution images can provide clearer details, making it easier to accurately identify and analyze tiny defects, deformations, and crimping quality of the crimping point, which is crucial for high-precision detection requirements. In images from multiple angles, due to differences in the angle and intensity of the light source, uneven brightness in the image may occur. By calculating the light intensity parameters at each angle, these differences can be quantified, helping subsequent steps to better adjust and compensate for light source errors and improve the quality of the light. By analyzing and extracting the intensity, we can more accurately understand the lighting conditions at each angle and provide data support for subsequent light source error compensation. This ensures that the subsequent analysis is not affected by uneven lighting and improves the overall image processing quality. Through brightness difference analysis, we can find the brightness inconsistency caused by different light source angles and intensities, which is very critical for further compensation and optimization processing. It avoids unstable image quality caused by different lighting conditions. The brightness difference value provides a quantitative basis for subsequent light source error compensation, ensuring that the brightness differences between different angles can be accurately eliminated in the subsequent optimization process, further improving the consistency and quality of the image. By dynamically calculating and compensating based on the light intensity parameters at each angle, The deviation caused by the light source can significantly reduce the brightness inconsistency between different angles. This processing step can effectively avoid the impact of lighting differences on image quality and ensure the uniformity of the image at all angles. Dynamic compensation can make adjustments based on real-time lighting changes, avoiding the limitations of traditional static light source compensation methods, and ensuring that the brightness differences of the image are effectively corrected under different production environments and lighting conditions. After dynamic light source deviation compensation, the lighting error of the image is effectively corrected, making the image at each angle more uniform in brightness and contrast, which can improve the overall quality of the image and provide more stable and accurate data input for subsequent image processing and defect detection. The image after light source error optimization It can eliminate the interference caused by uneven light sources, making the details of the crimping points clearer and easier to identify, which is crucial to improving the accuracy and robustness of defect detection, especially for the detection and quality assessment of subtle defects. Dynamic background removal technology can effectively remove environmental interference in the image, such as background noise, external light source interference, etc., and only retain the effective information of the crimping points. This step further improves the image quality and reduces the background interference on the recognition of crimping points. The image after background removal is clearer and cleaner, and the features of the crimping points are more prominent, which is easy for subsequent feature extraction and defect detection. This background removal not only improves the image quality, but also ensures that the various features of the crimping points can be more accurately identified and analyzed.

[0009] Preferably, the specific steps of step S15 are: Perform deep background semantic segmentation on the light source error optimization monitoring image to extract the image background area; Calculating the background texture gradient amplitude of the image background area to generate the background texture gradient amplitude; Analyze the texture direction distribution of the image background area to obtain the background texture direction distribution characteristics; Mining dynamic background changes based on background texture gradient amplitude and background texture direction distribution features to extract dynamic background texture change features; The background area pixels of the light source error optimization monitoring image are eliminated according to the dynamic background texture change characteristics to generate a background-eliminated cable crimping point monitoring image.

[0010] The present invention performs background semantic segmentation through a deep learning model, and can effectively distinguish the crimping point area and the background area in the image. Semantic segmentation technology can automatically identify and separate the crimping point and the background in the image based on a deep neural network, reduce the error of human intervention, and accurately extract the background area. By accurately segmenting the background and the crimping point area, it is possible to focus on processing the background area and avoid unnecessary processing of the important crimping point area, thereby improving the overall image processing efficiency. Deep semantic segmentation can handle complex backgrounds, such as illumination changes, complex texture environments, etc., effectively cope with variable production conditions, and ensure the accuracy of background segmentation. Background texture gradient amplitude calculation can effectively extract texture intensity information of the background area of ​​the image and measure the degree of texture change in the background area. The gradient amplitude reflects the change in pixel intensity in the image, and this indicator can be used to quantify the texture complexity of the background. The texture gradient amplitude of the background provides an effective basis for subsequent texture analysis. By calculating the amplitude, the parts with stronger or weaker textures in the background area can be clearly identified, providing basic data for subsequent texture direction distribution analysis and dynamic background change mining. Texture direction distribution analysis can capture the directional characteristics of textures in the background, such as lines and stripes. This step helps to understand the geometry of the background area, determine which texture directions are typical background textures, and distinguish irrelevant elements. The distribution information of texture direction is very important for distinguishing dynamic background from static background. For example, if the texture direction distribution in some areas changes, it may mean that there are dynamic background changes or lighting changes in the area, which will affect the detection of crimping points. By jointly analyzing the background texture gradient amplitude and texture direction features, the dynamic changes in the background caused by environmental changes (such as light source changes, scene changes, etc.) can be captured. This method can accurately identify the dynamic changes in the background area, thus providing a basis for subsequent background optimization. By mining the dynamic background texture change features, the dynamic background and static background can be effectively distinguished. For example, some areas may change dynamically due to lighting changes or equipment vibrations, while some areas remain static. Mining these differences can improve the accuracy of crimping point detection. According to the dynamic background texture change features, the pixels in the background area are accurately eliminated. This step helps to remove unnecessary parts of the background to avoid affecting the detection and analysis of crimping points, and ensure that only the image areas related to crimping quality are focused. After background removal, the details of the crimping point area are more prominent, making subsequent feature extraction and defect detection more accurate. After removing background noise, the quality of the crimping point can be more clearly displayed, improving the sensitivity and accuracy of defect identification. After removing background interference, the visual effect of the image is optimized, making the status of the crimping point clearer and more specific, which is convenient for subsequent analysis, diagnosis and decision-making.

[0011] Preferably, step S2 specifically comprises the following steps: Step S21: defining a unified acquisition timestamp; performing time difference analysis on the background-eliminated cable crimping point monitoring image according to the unified acquisition timestamp, and identifying image acquisition time difference values ​​at different angles; Step S22: removing asynchronous images according to the image acquisition time difference values ​​at different angles to obtain a time-synchronous monitoring image; Step S23: performing multi-angle camera tag calculation on the time-synchronized monitoring image to generate camera intrinsic parameters and extrinsic parameters for each angle; Step S24: performing image posture jitter analysis at each angle based on the camera intrinsic parameters and extrinsic parameters at each angle, and generating image displacement errors at each angle; Step S25: performing posture deviation correction processing on the time-synchronized monitoring image based on the image displacement error at each angle to construct a posture-corrected monitoring image.

[0012] The present invention can ensure the consistency of all image acquisition times by unifying the acquisition timestamp, and provide a basis for subsequent time synchronization processing. Inconsistent image acquisition times at different angles may cause incomplete matching of image content, especially when the scene changes. By setting a unified timestamp, this problem can be effectively avoided. By performing difference analysis on the image acquisition times at different angles, the time difference between images can be accurately identified, thereby providing data support for subsequent asynchronous image elimination and synchronous processing. This step can help eliminate errors caused by different acquisition times and ensure the time consistency of images. In the actual acquisition process, there may be a certain deviation in the image acquisition time at different angles, which will affect the quality and consistency of the image. By eliminating asynchronous images based on time difference values, those images that do not match in time can be effectively excluded, ensuring that the images used for subsequent analysis are time synchronized. The acquisition of time-synchronized images ensures the time consistency of all images and avoids the problem of inconsistent content caused by acquiring images at different times. This is crucial for subsequent image analysis, feature extraction and crimping point recognition. By calibrating the camera for each angle, the camera's intrinsic parameters (such as focal length, principal point coordinates, distortion coefficient, etc.) and extrinsic parameters (such as the camera's rotation matrix and translation vector) for each angle are calculated. These parameters are very important for subsequent image processing, 3D modeling, and defect detection, ensuring the geometric accuracy of the image. Multi-angle camera calibration ensures that images at each angle can be accurately aligned with the global coordinate system, thereby eliminating the deviation caused by different camera angles. This helps to improve the consistency and reliability of multi-angle image data and provide more stable input data for 3D reconstruction and crimp point detection. By calculating the displacement error of each angle image, the source of error in the image can be more accurately determined, thereby providing quantitative error data for image posture deviation correction. This can significantly improve the accuracy of image correction and reduce the impact of displacement errors. Posture jitter analysis can help further optimize the image registration process, ensuring that when images at multiple angles are stitched or processed, the alignment between images is more accurate, improving the accuracy of subsequent image processing steps. Based on the image displacement error at each angle, posture deviation correction can eliminate image deviations caused by shooting angles, camera shake, etc., to ensure that all images are aligned in the same coordinate system. This correction step can improve the geometric accuracy of the image and provide more accurate image data for subsequent inspection and analysis. After posture deviation correction, the geometric errors in the image are corrected, the image details are clearer, and the features of the crimping points are more prominent. This provides more accurate image input for crimping point detection, defect identification and quality assessment. After posture deviation correction, images from different angles can be more accurately fused or spliced, providing more consistent and accurate image data for subsequent 3D reconstruction, feature extraction and defect detection.

[0013] Preferably, step S3 specifically comprises the following steps: Step S31: performing cable crimping area feature point recognition on the posture correction monitoring image, and marking a plurality of cable crimping feature points; Step S32: performing triangulation processing on a plurality of cable crimping feature points to obtain the three-dimensional position coordinates of each feature point; Step S33: performing a cable crimping area structure analysis based on a plurality of cable crimping feature points to generate a topological structure feature of the cable crimping area; Step S34: performing three-dimensional point cloud modeling on the topological structure features of the cable crimping area based on the three-dimensional position coordinates of each feature point, and constructing a three-dimensional point cloud model of the cable crimping point; Step S35: performing global registration optimization on the three-dimensional point cloud model of the cable crimping point, thereby obtaining a three-dimensional registration optimization model of the crimping point. The present invention can automatically identify and mark the key feature points of the cable crimping area through an image processing algorithm, thereby avoiding manual labeling errors and improving the accuracy and efficiency of identification. Feature point recognition is an important step in image processing. By automatically marking multiple cable crimping feature points, the position of the crimping point in the image can be efficiently and accurately located. This provides a reliable data source for subsequent three-dimensional reconstruction and structural analysis. Through triangulation, multiple feature points are converted from a two-dimensional image to point coordinates in a three-dimensional space. This step combines image information from different angles and uses geometric principles to calculate the precise three-dimensional position of each feature point, providing basic data for subsequent three-dimensional modeling and analysis. Triangulation can obtain the three-dimensional coordinates of each feature point with high precision, ensuring that the position of each crimping point in space is accurately determined. This is crucial for subsequent cable crimping area structural analysis and defect detection. By performing structural analysis on multiple feature points, the topological structural features of the cable crimping area can be extracted, including the shape of the crimping area, connection relationship, etc. This helps to fully understand the distribution of crimping points and their mutual relationship, and provide structural data for subsequent analysis. Structural analysis not only reveals the location of the crimping points, but also helps identify the spatial relationship between them, providing a reference for detecting crimping quality and identifying potential defects. For example, parameters such as the distance and angle between crimping points can be evaluated through topological structure analysis. Based on the 3D coordinates of each feature point, a complete 3D model of the cable crimping point is constructed through point cloud modeling technology. This not only provides an intuitive 3D visualization image of the crimping area, but also provides important geometric data for further crimping quality analysis. 3D point cloud modeling can restore the spatial structure of the cable crimping area very accurately, so that the details of the crimping point are clearly presented. This helps to identify any geometric deviations or defects that may exist during the crimping process. Global registration optimization adjusts the 3D point cloud models at multiple angles to eliminate the errors caused by different camera positions and angles during the shooting process at each angle, ensuring the accurate fusion of the final point cloud model. This can effectively reduce the errors caused by different data sources. Registration optimization can ensure the consistency of each point cloud model in space and improve the accuracy of the 3D model of the crimping point. This is very critical for further crimping quality analysis, defect detection, and structural optimization. The three-dimensional point cloud model that has been optimized through registration has higher geometric accuracy and spatial consistency, making the shape of the crimping point clearer and being able to more accurately detect possible defects in the shape, position, etc. of the crimping point.

[0014] Preferably, the specific steps of step S4 are: Step S41: performing cable surface detail analysis on the crimping point three-dimensional registration optimization model to extract cable surface detail features; Step S42: Performing visual inspection of crimping point defects based on the cable surface detail features to obtain crimping point defect data; Step S43: performing defect morphology analysis on the crimping point defect data to generate crimping point defect morphology features; Step S44: accurately locating the defect area of ​​the crimping point defect data to obtain crimping point defect location data; Step S45: Based on the crimping point defect positioning data, the crimping point defect morphological features are visualized and rendered on the crimping point three-dimensional registration optimization model to construct a crimping point defect morphological rendering model.

[0015] The present invention can deeply identify the tiny geometric features of the crimping point area, such as surface bumps, cracks, wear, etc., by performing detailed analysis on the cable surface of the three-dimensional registration optimization model, which helps to capture small defects that are not easy to detect, thereby improving the accuracy of detection. The extraction of cable surface detail features can help identify key surface features of the crimping point area, such as surface smoothness, crimping surface morphology, local irregularities, etc. These features provide an accurate basis for subsequent defect detection. Through visual inspection technology, combined with the extracted cable surface detail features, various defects in the crimping point area, such as incomplete crimping, material defects, cracks, over-crushing, etc., can be automatically identified. Degree compression, etc. Automated detection not only improves efficiency, but also reduces human errors. The visual inspection method based on surface detail features can identify a variety of defect types and has a strong generalization ability. Whether it is an appearance defect or a tiny surface irregularity, it can be accurately detected. Defect morphology analysis can extract the systematic characteristics of the defect, such as the length of the crack, the angle of the fracture surface, the depth of the depression, etc. These details are crucial for judging the impact of the defect on the cable function. Through morphological feature analysis, different types of defects can be classified to help further optimize the crimping quality assessment process, so that the harm of each defect can be accurately evaluated. By accurately positioning the defect data, the exact location of the defect can be determined, which is of great significance for subsequent quality assessment, repair and production optimization. Accurate positioning helps engineers quickly identify and solve problems. Accurate defect positioning can significantly improve diagnostic efficiency and avoid wasting time and resources on unnecessary regional inspections, which makes maintenance and adjustment work more efficient and reduces downtime in production. By visualizing the defect morphological characteristics and the three-dimensional model of the crimping point, the specific location, morphology and impact of the defect on the crimping point can be intuitively displayed. This process allows inspectors to more clearly understand the nature of the defect and its impact on product quality. The visual rendering model can help technicians, engineers and decision makers understand complex defect data more easily, and then make more accurate repair decisions, such as which areas need to be re-crimped, which defects have a greater impact on electrical performance, etc. By visualizing the defect morphology in the three-dimensional model, the cable crimping quality can be more comprehensively evaluated to ensure that each crimping point meets the production standards, which helps to improve the quality control level in the production process. Visual rendering not only helps inspectors to make accurate assessments, but can also be directly used for real-time monitoring and adjustment on the production line to improve the degree of automation and real-time response capabilities of production.

[0016] Preferably, the specific steps of step S5 are: Step S51: quantifying the degree of deformation of the defect area of ​​the crimping point defect morphology rendering model to generate a quantified value of the defect area deformation; Step S52: analyzing the stress distribution of the defective area based on the quantified value of the deformation of the defective area to obtain stress distribution characteristics of the defective area; Step S53: simulating stress field changes according to stress distribution characteristics of the defect area to generate stress field change simulation data of the defect area; Step S54: Perform a comprehensive crimping quality assessment on the stress field change simulation data of the defective area to generate a cable crimping quality assessment report.

[0017] The present invention can quantify the degree of deformation of the defective area by analyzing the rendering model of the crimping point defect morphology, and ensure that the deformation data of each crimping point is accurately recorded, which provides intuitive and reliable physical data for subsequent stress analysis. The quantified value of the deformation of the defective area can accurately reflect the size and degree of deformation, thereby providing a quantitative standard for evaluating the crimping quality. Through the quantified data, it is possible to better judge whether the structure of the crimping point will be affected and whether there are hidden dangers that may cause failures. Based on the quantified value of the deformation of the defective area, the stress distribution of the area can be analyzed, especially the identification of stress concentration areas, which is crucial to evaluating the structural safety of the crimping point and helps to find out the risk of crimping failure that may be caused by stress concentration. Through stress distribution analysis, the stress distribution characteristics of each defective area can be quantified, including indicators such as stress size and distribution uniformity, which provides a detailed mechanical basis for judging the crimping quality and avoids the inaccuracy of traditional empirical methods. Through the simulation of stress field changes, the stress changes in the defective area under different working conditions can be further understood in depth, and the simulation results can show the application of the stress field change simulation. The trend of force changes over time or external load helps predict the performance of the crimping area in actual use. Stress field simulation can provide designers with effective information on stress changes in the crimping area, optimize product design, reduce high stress areas, and ensure the stability and reliability of the crimping points in actual use. By comprehensively evaluating the simulation data of stress field changes in defective areas, it is possible to comprehensively consider factors such as deformation, stress distribution, and stress field changes to comprehensively evaluate the cable crimping quality. The evaluation report can cover all key quality indicators to ensure that the product meets quality standards in all dimensions. The comprehensive evaluation report based on simulation data can be used as a basis for decision-making to help companies achieve data-driven quality control. Through quantitative analysis, it avoids the subjectivity and inaccuracy brought about by traditional manual inspections and ensures the consistency of product quality. The crimping quality evaluation report can effectively prevent potential failures and reduce quality problems in production by deeply analyzing the potential risks of stress distribution and defective areas. This not only helps to improve production efficiency, but also reduces the risk of rework and recall caused by quality problems in the later stage.

[0018] Preferably, the specific steps of step S6 are: Step S61: performing crimping defect diagnosis based on the cable crimping quality assessment report to generate crimping defect diagnosis data; Step S62: making adaptive correction decisions on the crimping defect diagnosis data to construct a cable crimping defect correction strategy; Step S63: dynamically migrate and optimize the cable crimping defect correction strategy, and construct a cable crimping defect optimization model to perform cable crimping detection and diagnosis optimization operations.

[0019] The present invention can accurately identify various defect types existing in the crimping process, such as incomplete crimping, cracks, deviations, looseness, etc., through in-depth analysis of the cable crimping quality assessment report. The defect diagnosis data generated based on the assessment report provides a scientific basis for subsequent repair and optimization. Traditional defect diagnosis usually relies on manual experience and is easily affected by subjective factors. The diagnostic data generated by the assessment report is completely based on objective data analysis, which helps to improve the accuracy and reliability of the diagnosis and avoid misjudgment and missed judgment in manual detection. After receiving the defect diagnosis data, through adaptive correction decision-making, the system can formulate corresponding correction strategies according to the specific defect type and severity. This mechanism can automatically adjust the correction strategy according to different working conditions, thereby maximizing the repair effect and reducing human operation errors. The correction decision of this step is based on deep learning or adaptive control algorithm, which can flexibly adjust the correction strategy in different production cycles, environmental changes or product demand changes, which makes the correction process more intelligent and more adaptable. Through adaptive correction decision-making, it can be adjusted according to the specific characteristics of the defect (such as depth, position, shape). Correction method, for example, for serious crimping deviations, it may be necessary to replace the crimping tool or adjust the crimping pressure; while minor defects can be solved by local correction, which not only improves the repair efficiency, but also ensures the repair quality. With the changes in the production environment, process conditions or equipment status, the correction strategy may need to be constantly adjusted. Through dynamic migration optimization, the optimization model can obtain new data and adjust the strategy in real time to ensure the applicability and effectiveness of the correction measures in different environments. In this way, the optimization strategy can be dynamically adjusted according to changes in production to avoid the problem that the fixed mode cannot adapt to new situations. Through dynamic migration optimization, the system will continue to self-learn and improve, gradually accumulate experience data, and form a more complete and efficient defect correction model, which will help improve long-term production efficiency and product quality, and ensure that each correction strategy is more accurate and efficient than the previous one. The dynamic migration optimization model can not only improve the accuracy and adaptability of the correction strategy, but also improve the detection accuracy of crimping defects. Through the combined analysis of historical data and real-time data, the optimization model can identify new defect patterns or potential problems to help adjust the production process and repair plan in time.

[0020] In this specification, a cable crimping detection system based on image processing is provided, which is used to execute the cable crimping detection method based on image processing as described above, including: A background elimination module is used to obtain a multi-angle high-resolution monitoring image of the cable crimping point; and perform dynamic background elimination processing on the multi-angle high-resolution monitoring image to generate a background-eliminated cable crimping point monitoring image; The posture deviation correction module is used to perform posture jitter analysis of the background-eliminated cable crimping point monitoring image angle by angle, and perform posture deviation correction processing to construct a posture correction monitoring image; The 3D point cloud module is used to identify the characteristic points of the cable crimping area in the posture correction monitoring image and perform 3D point cloud modeling to obtain a 3D registration optimization model of the crimping point; The defect visual detection module is used to perform visual detection of crimping point defects on the crimping point three-dimensional registration optimization model, and perform visual rendering to build a crimping point defect morphology rendering model; The crimping quality assessment module is used to quantify the degree of deformation of the defect area of ​​the crimping point defect morphology rendering model and conduct a comprehensive assessment of the crimping quality, and generate a cable crimping quality assessment report; The correction decision module is used to make adaptive correction decisions based on the cable crimping quality assessment report, perform dynamic migration optimization, and build a cable crimping defect optimization model to perform cable crimping detection and diagnosis optimization operations.

[0021] The present invention can remove unnecessary interference factors such as background noise, lighting changes, etc. through background elimination, thereby improving the clarity and quality of the image and helping the subsequent processing module to more accurately identify the crimping points. Background elimination effectively reduces errors caused by environmental changes or equipment status fluctuations. By dynamically adjusting the elimination algorithm, the background elimination module can cope with changes in different environments to ensure the accuracy of the image content. Through posture deviation correction, it is ensured that images at different angles can be aligned in the same coordinate system, eliminating errors caused by unstable camera posture, thereby improving image consistency. Posture correction helps to accurately locate the crimping points and reduce errors caused by shooting angles. In the process of multi-angle image synthesis, the corrected images can be better integrated to ensure the data accuracy during 3D modeling, thereby improving the 3D modeling quality of the crimping point. Through accurate feature point recognition and 3D point cloud modeling, the 3D spatial information of the cable crimping point can be efficiently constructed, which provides detailed geometric data for subsequent defect detection and quality assessment. 3D registration optimization helps to solve the errors and mismatches in images at different angles, ensuring that the constructed 3D point cloud model has high accuracy. The 3D point cloud model can provide multiple functions for defect detection and stress analysis of crimping points. Dimensional data support facilitates comprehensive evaluation of crimping quality. Through advanced image processing technology, various defects at crimping points can be accurately identified, including poor crimping, surface cracks, deviations, etc. This provides an automated detection solution for the production line and reduces manual errors. Through rendering technology, the detected defects are visualized, allowing engineers to intuitively see the shape, location and severity of the defects, thereby making more accurate repair decisions. Through quantitative analysis of the degree of deformation in the defective area, the crimping quality assessment module can accurately assess the quality of the crimping points, avoiding the subjectivity of traditional quality assessment methods. This module comprehensively considers deformation, stress, defects, and other factors. The system can comprehensively evaluate the quality of crimping points from multiple dimensions based on factors such as shape and morphology, and provide more accurate quality data. Through adaptive correction decision-making, the system can automatically adjust the repair plan according to the crimping quality evaluation results without manual intervention, thereby improving the accuracy and efficiency of the repair. Dynamic migration optimization can continuously adjust the correction strategy according to changes in actual working conditions to ensure that the repair plan is always effective in different production environments, thereby continuously improving the detection and repair efficiency. The correction decision module can combine historical data and real-time data for optimization, accurately diagnose crimping defects and propose effective repair plans, reducing the occurrence of quality problems and improving the stability and reliability of the crimping process. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A schematic diagram of the steps of a cable crimping detection method based on image processing according to the present invention; Figure 2 Detailed implementation flow chart of step S1; Figure 3 Detailed implementation flow chart of step S2; Figure 4 Detailed implementation flow chart of step S3. DETAILED DESCRIPTION

[0023] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0024] The present application example provides a cable crimping detection method and system based on image processing. The execution subject of the cable crimping detection method and system based on image processing includes but is not limited to: mechanical equipment, data processing platform, cloud server node, network upload device, etc. equipped with the system can be regarded as the general computing node of the present application, and the data processing platform includes but is not limited to: at least one of an audio image management system, an information management system, and a cloud data management system.

[0025] See also Figures 1 to 4 The present invention provides a cable crimping detection method based on image processing, and the cable crimping detection method based on image processing comprises the following steps: Step S1: Acquire a multi-angle high-resolution monitoring image of a cable crimping point; perform dynamic background elimination processing on the multi-angle high-resolution monitoring image to generate a background-eliminated cable crimping point monitoring image; Step S2: performing posture jitter analysis of each angle of the background-eliminated cable crimping point monitoring image, and performing posture deviation correction processing to construct a posture correction monitoring image; Step S3: identifying feature points of the cable crimping area on the posture correction monitoring image, and performing three-dimensional point cloud modeling, thereby obtaining a three-dimensional registration optimization model of the crimping point; Step S4: performing visual inspection of crimping point defects on the crimping point three-dimensional registration optimization model, and performing visual rendering to construct a crimping point defect morphology rendering model; Step S5: quantifying the degree of deformation of the defect area of ​​the crimping point defect morphology rendering model and performing a comprehensive crimping quality assessment to generate a cable crimping quality assessment report; Step S6: Make adaptive correction decisions based on the cable crimping quality assessment report, perform dynamic migration optimization, and build a cable crimping defect optimization model to perform cable crimping detection and diagnosis optimization operations.

[0026] The present invention can obtain detailed images of cable crimping points from different viewing angles through multi-angle and high-resolution monitoring image acquisition, ensuring that all features of the crimping points are fully captured. This method can provide more comprehensive crimping point information and avoid missing crimping defects due to viewing angle problems. Background elimination technology effectively removes background noise and interference in the image to ensure that the image only contains real data of the cable crimping points. This processing can greatly improve the accuracy of subsequent image analysis, avoid the influence of the background environment on image recognition, and ensure the recognition accuracy of feature points and defects. Due to the image posture jitter caused by factors such as camera position deviation or object vibration during multi-angle image acquisition, these errors can be identified by dynamically analyzing the image posture. And enhance image stability and improve the accuracy of subsequent processing. After image correction, the shape of the crimping point is more accurate and stable, which can eliminate the error caused by angle deviation, ensure the standardization of image data, and provide accurate basis for subsequent feature point recognition and 3D modeling, which is crucial to improving the accuracy and reliability of defect detection. Through accurate feature point recognition, the key points of the cable crimping area can be located, providing important reference for subsequent 3D modeling. The extraction of feature points helps to improve the detail recognition ability of the crimping point, especially for the detection of tiny defects. 3D modeling can truly reproduce the 3D structure of the crimping point, providing more intuitive data support for defect detection. 3D point cloud data can accurately represent the shape of the crimping point in space. This allows for a better assessment of its geometry and deformation, providing a more sophisticated model for subsequent defect analysis and quality assessment. Defect detection is performed based on the 3D registration optimization model, so that every subtle defect at the crimping point can be accurately identified. The 3D model can help the algorithm better understand the geometric features of the crimping area, thereby improving the sensitivity of defect detection. Through visualization technology, the defects and morphology of the crimping points can be displayed, which can provide intuitive defect information to inspectors. This visualization method not only improves the accuracy of detection, but also facilitates subsequent quality analysis and decision support. Through quantitative analysis, the degree of deformation of the crimping point can be accurately assessed to determine whether it meets the product standard requirements. The quantification of the degree of deformation provides a quantitative basis for evaluating the crimping quality. The basis makes the test results more objective and scientific. The generated quality assessment report can provide operators and quality management personnel with a comprehensive crimping quality analysis. The report covers the status of the crimping points, defects and potential risks, provides data support for decision-making, and helps to implement effective quality control. Through the adaptive correction algorithm, the crimping process parameters can be dynamically adjusted according to the actual test results. If quality problems or potential defects are detected, the system can adjust the operating procedures in real time and automatically optimize the crimping parameters, thereby improving the stability of the production process and the quality control level. The dynamic optimization model can be adjusted according to real-time data to further optimize the detection algorithm and quality control strategy. This intelligent dynamic adjustment can cope with different production environments and product requirements.Significantly improve production efficiency and product consistency. Through optimized defect detection and diagnosis operations, crimping quality problems can be more effectively identified and repaired. The system can continuously learn and optimize, gradually improve detection accuracy, reduce human intervention, and improve the level of detection automation.

[0027] In the embodiment of the present invention, refer to Figure 1 , is a schematic flow chart of the steps of a cable crimping detection method based on image processing of the present invention. In this example, the steps of the cable crimping detection method based on image processing include: Step S1: Acquire a multi-angle high-resolution monitoring image of a cable crimping point; perform dynamic background elimination processing on the multi-angle high-resolution monitoring image to generate a background-eliminated cable crimping point monitoring image; In this embodiment, a suitable high-resolution camera (such as a DSLR or an industrial camera) is selected, and it is ensured that the camera supports multi-angle shooting. A camera stand and a stabilizer are prepared to ensure stability during the shooting process. The lighting of the shooting environment is ensured to be uniform and sufficient. Lighting equipment such as a ring light, a spotlight, or an LED light bar can be used to avoid the influence of shadows and reflections on the image quality. The exposure time, ISO value, and aperture of the camera are adjusted to obtain the best image quality. According to the structural characteristics of the cable crimping point, a reasonable shooting angle is designed. It is generally recommended to shoot from multiple directions (such as the front, side, and top) to fully obtain the details of the cable crimping point. The shooting position and distance of each angle are determined to ensure that each image can clearly present the key area. At each predetermined shooting angle, a camera is used to shoot multiple images to ensure that the best image is obtained. Multiple shots can be taken at different angles and distances, and the clearest and most detailed images are selected for subsequent processing. During the shooting process, the camera settings are ensured to be consistent so as to ensure uniformity in subsequent image processing. The images taken are classified and stored according to the shooting angle and time to ensure that the required images can be quickly found in subsequent processing. It is recommended to use efficient file naming and organization structure for subsequent recognition. Before background elimination, the acquired images are first The image is preprocessed, including denoising, sharpening and contrast adjustment. Filters (such as Gaussian filtering) in image processing software (such as OpenCV or MATLAB) can be used to remove image noise. Dynamic background modeling algorithms are used. Common methods include mixed Gaussian models (GMM) or mean background modeling. These algorithms can analyze the changes between multiple images, identify the background part and model it. During the processing, the system automatically calculates the background model of each pixel and updates the background information to adapt to lighting changes and scene dynamics. Use the previously established background model to extract the foreground of each image. By calculating the current The difference between the front image and the background model is used to identify the foreground part of the cable crimping point. A threshold can be set to filter out small changes to ensure that the extracted foreground part is accurate and clear. The extracted foreground image is post-processed, such as morphological operations (such as dilation and erosion) to eliminate small noise points and fill the holes in the foreground area. This step can further improve the clarity and connectivity of the cable crimping point. The processed background-eliminated cable crimping point monitoring image is saved as a new image file to ensure that it is distinguishable from the original image. A record can be generated for each processing result, including the processing time, the algorithm parameters used, etc., for subsequent tracking and analysis.

[0028] Step S2: performing posture jitter analysis of each angle of the background-eliminated cable crimping point monitoring image, and performing posture deviation correction processing to construct a posture correction monitoring image; In this embodiment, the cable crimping point monitoring image after background elimination is imported into the image processing software (such as OpenCV or MATLAB), ensuring that the images of each angle are loaded and arranged in order for subsequent processing, and basic preprocessing is performed on the image, including denoising and contrast enhancement. This process can help the subsequent feature point detection and posture analysis. A feature point detection algorithm (such as Shi-Tomasi corner point detection, ORB or SIFT) is applied to each image to identify key feature points. These feature points will be used for subsequent posture analysis and correction. For each image, the coordinates of the detected feature points are recorded for posture jitter analysis. By comparing the positions of feature points in images of the same scene at different angles, the degree of posture jitter is analyzed. The optical flow method or feature matching algorithm (such as FLANN or BFMatcher) can be used to calculate the position between feature points. Shift, calculate the posture difference between each image and the reference image (which may be the first image taken), determine whether the image has posture deviations such as rotation, translation or scaling during the shooting process, and use geometric transformations (such as affine transformations or perspective transformations) to correct each image based on the results of the posture jitter analysis. By setting the transformation matrix, the coordinate system of the image can be adjusted to a unified standard. During the correction process, an interpolation method (such as bilinear interpolation or cubic interpolation) is applied to smooth the corrected image to ensure that the image quality is not lost. After the correction is completed, check the corrected images one by one to verify whether their postures meet expectations. The overlap of feature points can be ensured by superimposing the original image and the corrected image. Save the corrected posture correction monitoring image as a new image file, and record the correction parameters of each image (such as transformation matrix, feature point positions before and after correction, etc.) for subsequent analysis and reference.

[0029] Step S3: identifying feature points of the cable crimping area on the posture correction monitoring image, and performing three-dimensional point cloud modeling, thereby obtaining a three-dimensional registration optimization model of the crimping point; In this embodiment, the posture-corrected monitoring images are imported into image processing software (such as OpenCV or MATLAB) to ensure that the images at each angle are loaded and clearly marked. The images are further preprocessed, including image enhancement and denoising, to improve the accuracy of subsequent feature point recognition. Feature point detection algorithms (such as SIFT, SURF, or ORB) are used to identify feature points of the cable crimping area in each image. These algorithms can extract significant and repeatable feature points, providing a basis for subsequent matching and modeling. Descriptors are generated for each feature point for feature point matching. These descriptors will be used for feature point matching between images of different viewing angles to ensure that subsequent three-dimensional reconstruction is effective. Feature matching algorithms (such as FLANN or BFMatcher) are used to match feature points in images of different angles. The best matching pair is found by calculating the distance between feature points, and a robustness test is performed. The RANSAC algorithm is used to eliminate erroneous matches to ensure the accuracy and robustness of the matching results. Reliability, according to the matched feature points, the three-dimensional coordinates of each feature point are calculated by triangulation. This process requires knowing the internal and external parameters of the camera (such as focal length, optical center position, etc.) in order to convert the two-dimensional image coordinates into three-dimensional space coordinates, integrate all the calculated three-dimensional points, and generate a preliminary three-dimensional point cloud model. Point cloud processing software (such as PCL or CloudCompare) can be used for data visualization and processing. After obtaining the preliminary three-dimensional point cloud, the registration algorithm (such as ICP algorithm) is applied to optimize the registration of the point cloud. Through an iterative method, the distance between the point clouds is minimized to make the final point cloud more accurate. In the registration process, ensure that all point clouds are in the same coordinate system, so as to achieve accurate three-dimensional reconstruction, and save the optimized three-dimensional point cloud model in a standard format (such as PLY or OBJ) for subsequent analysis and application, and record the relevant parameters and steps of the three-dimensional reconstruction, including the algorithms used, matching results, etc., for subsequent backtracking and optimization.

[0030] Step S4: performing visual inspection of crimping point defects on the crimping point three-dimensional registration optimization model, and performing visual rendering to construct a crimping point defect morphology rendering model; In this embodiment, the three-dimensional registration optimization model of the crimping point is imported into a visualization software (such as Blender, Unity or Maya) to prepare for defect detection and visual rendering. Ensure that the details and structure of the model are complete, and perform necessary cleaning and optimization to improve the rendering effect. Select a suitable visual inspection algorithm, usually using a deep learning model (such as a convolutional neural network CNN) or a traditional image processing method (such as edge detection, texture analysis, etc.). If a deep learning model is used, the trained network weights need to be loaded and the network needs to be configured as necessary to adapt to the task of crimping point defect detection. Slice the surface of the three-dimensional model to generate a two-dimensional slice image, or directly apply texture mapping to the three-dimensional model to map the image data to the model surface. Run the defect detection algorithm on each slice or processed model surface to identify potential defect areas. Combined with the training data set, the algorithm will identify different types of defects such as cracks, bubbles, and shedding. The defect area detected by the algorithm is calibrated, and information such as the location, type, and severity of the defect is recorded. Visualize the inspection results for subsequent analysis. Highlight the defect area on the model, and color coding or transparency changes can be used to highlight the nature of the defect, so that analysts can understand the specific situation of the defect. In the visualization software, set rendering parameters such as lighting, material, and camera perspective to ensure that the defect rendering effect is realistic and intuitive. By using a rendering engine (such as Cycles, Arnold, etc.), generate high-quality rendered images or animations to show the crimping points and their defect characteristics. Different perspectives and scene settings can be selected to show the performance of defects under different conditions. Save the rendering results as images or video files for subsequent display and sharing. At the same time, record the defect detection results, calibration data, and visualization parameters in the database for future reference and analysis. Write a comprehensive report that outlines the results and visualization effects of defect detection, including the type, location, rendering effect, etc. of the defect, to provide a basis for subsequent quality control and improvement.

[0031] Step S5: quantifying the degree of deformation of the defect area of ​​the crimping point defect morphology rendering model and performing a comprehensive crimping quality assessment to generate a cable crimping quality assessment report; In this embodiment, the applicable deformation quantification method is determined, and the commonly used ones are geometric deformation analysis, modal analysis, and difference comparison based on point cloud data. Select the corresponding algorithm according to the needs. Prepare the necessary tools and software, such as MATLAB, Python (combined with NumPy and SciPy) or dedicated analysis software for subsequent quantitative analysis. Extract the three-dimensional data of the defect area from the crimping point defect morphology rendering model. The area with obvious deformation can be identified by threshold segmentation or regional growing algorithm. Analyze the extracted defect area and extract key features such as volume, surface area, main axis direction, etc., which will provide basic data for quantifying the degree of deformation. Geometric deformation measurement methods such as Hausdorff distance and root mean square error (RMSE) are used to calculate the difference between the defect area and the ideal geometric shape. Compare with historical data to evaluate whether the degree of deformation exceeds the set threshold, so as to quantify the change in crimping quality. This step can be automated through programming. Determine the comprehensive evaluation indicators of crimping quality, including deformation degree, stress distribution, defect type and its influence degree. Set weights and scoring criteria for each indicator for comprehensive evaluation. The selection of these indicators should be combined with industry standards and actual production experience to ensure the scientificity and effectiveness of the evaluation. According to the previous quantitative results and the set evaluation indicators, each crimping point is comprehensively scored. This can be done by weighted average method, hierarchical analysis method (AHP) or fuzzy comprehensive evaluation method. Record the score and weight of each evaluation indicator to ensure the transparency and traceability of the comprehensive evaluation.

[0032] Step S6: Make adaptive correction decisions based on the cable crimping quality assessment report, perform dynamic migration optimization, and build a cable crimping defect optimization model to perform cable crimping detection and diagnosis optimization operations.

[0033] In this embodiment, the cable crimping quality assessment report is deeply analyzed to identify key defect types, occurrence frequencies, influencing factors and their correlation with crimping quality, with a focus on indicators mentioned in the assessment, such as deformation degree, defect mode and historical data, so as to provide a scientific basis for subsequent correction decisions. Based on the analysis results, an adaptive correction decision model is constructed. A rule-based system, decision tree model or machine learning algorithm (such as random forest, support vector machine, etc.) can be used to perform defect detection and correction decisions. The model should be able to automatically adjust the correction strategy according to real-time data and historical data. For example, when a new defect type is discovered, the corresponding correction measures are automatically updated. A dynamic migration optimization strategy is introduced, and a transfer learning algorithm (such as domain adaptation or cross-domain learning) is used to migrate historically successful correction strategies to a new production environment. By analyzing data in different production environments, applicable correction strategies are identified, and adjusted and optimized in the new environment. This process can improve the adaptability of the model. and accuracy. According to the strategy generated by the adaptive correction decision model, specific corrective measures are implemented, which may include optimizing the crimping process, selecting new materials, adjusting process parameters, etc. An implementation plan is set up, including responsible persons, implementation steps and timetables, to ensure that the corrective measures can be effectively implemented and to track progress. A real-time monitoring system is established to collect key data in the crimping process, such as temperature, pressure, time and other parameters. These data will be used to evaluate the effectiveness of the correction strategy. A feedback mechanism is established to regularly evaluate the implementation effect and further adjust the correction decision model based on the feedback information. This process can be summarized in the form of regular meetings or reports, regularly evaluate the performance of the optimization model, analyze the success rate of the corrective measures and their impact on the crimping quality, use statistical analysis methods (such as control charts, process capability analysis) to quantify the effect of the model, and update the model parameters and structure based on the evaluation results to ensure that it always adapts to the current production environment and quality requirements. This may include retraining the model or adjusting the decision rules.

[0034] In this embodiment, refer to Figure 2 , is a flowchart of detailed implementation steps of step S1. In this embodiment, the detailed implementation steps of step S1 include: Step S11: Acquire a multi-angle high-resolution monitoring image of the cable crimping point; Step S12: Calculate the light intensity at each angle on the multi-angle high-resolution monitoring image to extract the light intensity parameters at each angle; Step S13: performing multi-angle image brightness difference analysis on the multi-angle high-resolution monitoring image to generate image brightness difference values ​​between multiple angles; Step S14: performing dynamic light source deviation compensation calculation on the image brightness difference values ​​between multiple angles based on the illumination intensity parameter of each angle to generate a multi-angle dynamic light source deviation compensation value; Step S15: optimizing the light source error of the multi-angle high-resolution monitoring image based on the multi-angle dynamic light source deviation compensation value, thereby obtaining a light source error optimized monitoring image; Step S16: Perform dynamic background elimination processing on the light source error optimization monitoring image to generate a background eliminated cable crimping point monitoring image.

[0035] In this embodiment, a high-resolution camera device is selected to ensure that it has enough pixels to capture details and has a suitable lens to achieve multi-angle shooting. The camera is installed on a fixed bracket to ensure that the camera can remain stable during shooting and has the ability to adjust at multiple angles. According to the characteristics of the cable crimping point, multiple shooting angles (such as 0 degrees, 30 degrees, 60 degrees, 90 degrees, etc.) are set. Each angle should cover a different viewing angle of the crimping point to obtain comprehensive image information. A turntable or manual adjustment of the camera position is used to ensure that the shooting position of each angle is accurate. At each set angle, the camera's shooting function is used to capture images. Timed shooting or remote control shooting can be used to ensure that images are obtained at each angle. To obtain clear images, save the captured images in high-resolution formats (such as TIFF or PNG) to ensure that the image quality is not affected by compression. All captured images are stored in a computer or cloud storage by angle classification for subsequent processing and analysis. Each image file should contain angle information for subsequent reference. Preprocess the images at each angle, such as denoising and contrast enhancement, to ensure the accuracy of subsequent light intensity calculations. For each processed image, calculate its light intensity. The brightness value of the image (such as the average grayscale value) can be used as an indicator of light intensity. Associate the calculated light intensity parameters with the corresponding angles and store them in a file to form a light intensity parameter table, which will be used for subsequent light The source deviation compensation provides a basis, calculates the brightness difference of each pair of angle images, selects a base angle, and compares the images of other angles with it, uses indicators such as mean square error (MSE) or peak signal-to-noise ratio (PSNR) to quantify the brightness difference, and stores the calculated brightness difference value in a matrix. Each element of the matrix represents the brightness difference between different angles, which provides necessary information for subsequent dynamic light source deviation compensation. The brightness difference value matrix is ​​saved to a file to ensure the integrity and traceability of the data. It can be saved in CSV format for subsequent analysis. The light intensity parameters are normalized to eliminate the dimensional differences between different angles, and the light intensity parameters and brightness differences are used. The value matrix is ​​used to calculate the dynamic light source deviation compensation value for each angle. For each image, the brightness of the image is adjusted according to its corresponding dynamic light source deviation compensation value. For example, the light source can be balanced by adding and subtracting the compensation value. The linear transformation formula is applied: new image = original image + compensation value. After processing, the optimized image is saved to generate a light source error optimization monitoring image. This process aims to eliminate the image non-uniformity caused by light source changes, making the analysis results more reliable and accurate. By analyzing multiple frames of monitoring images and establishing a background model, a Gaussian mixture model can be used to model the static background. Each frame of the image is compared with the background model to extract the dynamic foreground part, that is, the monitoring image of the cable crimping point.

[0036] In this embodiment, the specific steps of step S15 are: Perform deep background semantic segmentation on the light source error optimization monitoring image to extract the image background area; Calculating the background texture gradient amplitude of the image background area to generate the background texture gradient amplitude; Analyze the texture direction distribution of the image background area to obtain the background texture direction distribution characteristics; Mining dynamic background changes based on background texture gradient amplitude and background texture direction distribution features to extract dynamic background texture change features; The background area pixels of the light source error optimization monitoring image are eliminated according to the dynamic background texture change characteristics to generate a background-eliminated cable crimping point monitoring image.

[0037] In this embodiment, a suitable deep learning model is selected for semantic segmentation, such as U-Net, DeepLab or MaskR-CNN, which can effectively segment the background and foreground in the image. Using a pre-trained model can speed up the training process and improve the segmentation accuracy. Prepare a training data set with annotations, including background and foreground (crimp point) annotations. The annotation data can be obtained by manual annotation or using an existing data set. Divide the data set into a training set and a validation set, and perform data enhancement (such as rotation, flipping, and scaling) to improve the generalization ability of the model. Train the model on the training set, monitor the loss function and accuracy during the training process, and ensure that the model can effectively learn the characteristics of the background and foreground. After the training is completed, evaluate the model performance on the validation set to ensure that the segmentation effect meets the requirements. Use the trained model to predict the light source error optimization monitoring image to obtain the segmentation results of the background and foreground. Extract the background area, generate a binary mask based on the prediction results, and mark the background area. Use the previously generated binary mask to extract the background area in the optimized monitoring image. Save the background area separately for subsequent processing. Use the Sobel operator or the Scharr operator to calculate the gradient of the background area. Gradient calculation can be divided into X direction and Y direction to extract texture information. Apply Sobel operator to the background image to obtain horizontal and vertical gradient maps. Calculate the gradient magnitude by combining these two gradient maps. Store the calculated background texture gradient magnitude as a new image file to ensure it is available for subsequent analysis. Analyze the texture direction of each pixel based on the previously calculated gradient information. At each pixel, calculate the angle: use the inverse tangent function (atan2) to calculate the direction of each pixel with the formula angle =atan2(gradient_y, gradient_x). Count the texture directions of all pixels and generate a direction histogram. The frequency of each direction represents the intensity of the texture in that direction. Set a suitable angle interval (such as 0-360 degrees) to map the texture direction distribution into a histogram. Compare the background texture gradient magnitude and direction distribution at different time points or under different conditions to identify the changing parts. Extract dynamic change features by calculating the difference between the gradient magnitude and direction histogram at two time points. Apply statistical analysis methods (such as variance analysis) to evaluate the changes in texture features under different background conditions. Identify texture features that change significantly, which will serve as a sign of dynamic background changes. Record the extracted dynamic background texture change features in a document for subsequent reference and analysis. In the light source error optimization monitoring image, identify the pixels that belong to the dynamic background area. This can be achieved by comparing with the dynamic background texture change features. Set a threshold to identify pixels that are similar to the background features. Mark the identified background area pixels and then fill them in through interpolation or other methods to eliminate interference.Image inpainting algorithms, such as the non-local means method (NLM) or the fast inpainting algorithm, can be applied to process the eliminated areas to ensure image coherence. Visually inspect the resulting background-removed cable crimp monitoring image to ensure that the crimp is clearly visible and background interference has been effectively removed. Save the processed image to the file system, ensuring that the file naming and storage path are convenient for subsequent analysis.

[0038] In this embodiment, refer to Figure 3 , is a flowchart of detailed implementation steps of step S2. In this embodiment, the detailed implementation steps of step S2 include: Step S21: defining a unified acquisition timestamp; performing time difference analysis on the background-eliminated cable crimping point monitoring image according to the unified acquisition timestamp, and identifying image acquisition time difference values ​​at different angles; Step S22: removing asynchronous images according to the image acquisition time difference values ​​at different angles to obtain a time-synchronous monitoring image; Step S23: performing multi-angle camera tag calculation on the time-synchronized monitoring image to generate camera intrinsic parameters and extrinsic parameters for each angle; Step S24: performing image posture jitter analysis at each angle based on the camera intrinsic parameters and extrinsic parameters at each angle, and generating image displacement errors at each angle; Step S25: performing posture deviation correction processing on the time-synchronized monitoring image based on the image displacement error at each angle to construct a posture-corrected monitoring image.

[0039] In this embodiment, a unified timestamp format (e.g., ISO 8601 format) is determined to ensure that the timestamps of all images are consistent. UTC time can be selected to avoid differences caused by time zones. The current unified timestamp is recorded each time an image is captured. A high-precision timer (such as a system clock) is used to ensure the accuracy of the timestamp and avoid errors caused by delays. Each image file is associated with its corresponding timestamp to form a record table containing the image path and timestamp. This will provide basic data for subsequent time difference analysis. Using the recorded timestamp information, prepare for time difference analysis of images at different angles. Ensure that each image can be accurately identified and processed. According to the unified acquisition timestamp, the time difference between images at different angles is calculated. Each pair of images is compared and the time difference is recorded. The threshold of the time difference is set according to the application requirements. For example, the maximum allowed time difference is set to ±100 milliseconds. Images outside this range will be regarded as asynchronous images. Traverse all images and eliminate those images that do not meet the synchronization conditions based on the time difference and the set threshold. Record the eliminated images for subsequent analysis or review. Save the processed time-synchronized monitoring images to a new folder to ensure that only synchronized image data is used for subsequent analysis. Use camera calibration techniques (such as Zhang Zhengyou calibration method) to obtain camera intrinsic parameters (focal length, principal point position, distortion coefficient, etc.) and extrinsic parameters (camera position and orientation) for each angle. Calculate the camera's intrinsic and extrinsic parameters by shooting known calibration objects (such as a chessboard). Collect calibration image data for each angle, perform feature extraction and matching, and then calculate camera parameters using the least squares method or other optimization algorithms. Save the camera's intrinsic and extrinsic parameters for each angle to a database or file for subsequent use and reference. Verify whether the calculated camera parameters are reasonable. The accuracy of the calibration results can be confirmed by reprojection error analysis. Use the camera's intrinsic and extrinsic parameters to match the feature points in the image to determine the displacement of the image during the shooting process. Apply feature point detection algorithms (such as SIFT, ORB, etc.) to extract key points in the image and calculate their changes at different angles. Calculate the displacement error of the image at each angle based on the displacement data of the feature points. The Euclidean distance formula can be used to calculate the distance that the feature point moves in the image. Record the displacement error results in a data table to form a summary of the displacement error at each angle for subsequent analysis and processing. Compare the displacement errors at different angles, analyze their rationality and consistency, and ensure that the calculation results can reflect the actual situation. Establish a correction model based on the displacement error at each angle. You can choose to use affine transformation or perspective transformation methods for correction. Apply the established correction model to transform each time-synchronized monitoring image. Adjust the image's translation, rotation and other parameters to eliminate posture deviations. Verify the corrected image to ensure that the geometry and position of the crimp point are accurately restored. This can be confirmed by visual inspection or comparison with a reference image.Save the corrected posture monitoring image to the file system, ensuring that the file naming and storage path are convenient for subsequent review and analysis.

[0040] In this embodiment, refer to Figure 4 , is a flowchart of detailed implementation steps of step S3. In this embodiment, the detailed implementation steps of step S3 include: Step S31: performing cable crimping area feature point recognition on the posture correction monitoring image, and marking a plurality of cable crimping feature points; Step S32: performing triangulation processing on a plurality of cable crimping feature points to obtain the three-dimensional position coordinates of each feature point; Step S33: performing a cable crimping area structure analysis based on a plurality of cable crimping feature points to generate a topological structure feature of the cable crimping area; Step S34: performing three-dimensional point cloud modeling on the topological structure features of the cable crimping area based on the three-dimensional position coordinates of each feature point, and constructing a three-dimensional point cloud model of the cable crimping point; Step S35: performing global registration optimization on the three-dimensional point cloud model of the cable crimping point, thereby obtaining a three-dimensional registration optimization model of the crimping point. In this embodiment, the posture correction monitoring image is preprocessed, including denoising, contrast enhancement and edge detection, to improve the recognizability of feature points. Image processing techniques such as Gaussian filtering and Canny edge detection can be used. Feature point detection algorithms (such as Harris corner detection, Shi-Tomasi corner detection or FAST feature detection) are used to identify the feature points of the cable crimping area in the image. These algorithms can effectively find points with obvious changes in the image. During the detection process, a threshold is set to control the number of feature points to ensure that the selected feature points have high repeatability and stability. The identified feature points are marked on the image, usually using circles or cross symbols to mark the feature points. This process can be achieved through an image drawing tool to convert the coordinates of the feature points into image coordinates for visualization. The marked image is saved, and the coordinate information (pixel coordinates) of each feature point is recorded to provide data support for subsequent triangulation and structural analysis. Collect two-dimensional coordinate data of cable crimping feature points from different angles. Each feature point should have coordinate data of at least two images for triangulation. Select a suitable triangulation algorithm (such as basic triangulation or least squares method) and use the camera's intrinsic and extrinsic parameters and the 2D coordinates of the feature points for 3D reconstruction. For each feature point, solve the 3D coordinates in the camera coordinate system. Use the camera's projection model to map the 2D coordinates of the feature point to a point in 3D space, for example, use the camera's intrinsic and extrinsic parameters to back-project the 2D coordinates into 3D space. Record the 3D coordinates of each feature point (e.g., X, Y, Z coordinates) in a data table to ensure that they are used for subsequent analysis and model building. Define the topological structural features of the cable crimping area, including the connection relationship between the feature points. The connection between the feature points can be represented by an adjacency matrix or an edge list. Use graph theory or computational geometry to analyze the relationship between the feature points. Construct the geometric features of the cable crimping area by calculating the distance, angle, etc. between the feature points. Extract key features in the topological structure, such as the distribution density, connectivity, and geometric shape of the feature points. These features can reflect the structural characteristics of the cable crimping area. Represent the generated topological structural features in a visual graph and record the relevant data. These results will provide basic information for subsequent 3D modeling. Prepare point cloud data based on the 3D coordinates of the feature points obtained previously. The 3D coordinates of each feature point will be used as a point in the point cloud. Use a point cloud processing library (such as the PCL library) or 3D modeling software to convert the 3D coordinates of the feature points into a point cloud data structure. Each point in the point cloud should contain its 3D coordinates, color, and other attributes. Visualize the constructed point cloud to check the completeness and accuracy of the point cloud. You can use 3D visualization tools (such as Meshlab or CloudCompare) to display the point cloud. Save the generated 3D point cloud model in a standard format (such as PLY, PCD, or OBJ) to provide basic data for subsequent registration optimization.Select an appropriate global registration algorithm (such as ICP algorithm, NDT registration, or global optimization method), which can effectively align multiple point clouds to the same coordinate system. Use the camera's internal and external parameters and feature point information for initial registration, and try to align point clouds at different angles. Run the global registration algorithm to optimize the registration results by minimizing the distance between point clouds. This process may require multiple iterations to obtain the best registration effect. Verify that the registered 3D point cloud model meets expectations and check the registration results through visualization tools. Finally, save the optimized 3D registration model for subsequent use and analysis.

[0041] In this embodiment, step S4 includes the following steps: Step S41: performing cable surface detail analysis on the crimping point three-dimensional registration optimization model to extract cable surface detail features; Step S42: Performing visual inspection of crimping point defects based on the cable surface detail features to obtain crimping point defect data; Step S43: performing defect morphology analysis on the crimping point defect data to generate crimping point defect morphology features; Step S44: accurately locating the defect area of ​​the crimping point defect data to obtain crimping point defect location data; Step S45: Based on the crimping point defect positioning data, the crimping point defect morphological features are visualized and rendered on the crimping point three-dimensional registration optimization model to construct a crimping point defect morphological rendering model.

[0042] In this embodiment, an optimized three-dimensional point cloud model of the crimping point is obtained, and point cloud processing software (such as PCL or CloudCompare) is used to import and preprocess the data. The preprocessing may include denoising, downsampling and surface reconstruction. A surface reconstruction algorithm (such as Poisson reconstruction or triangular mesh generation) is used to convert the point cloud data into a continuous surface model, which can better analyze the detailed features of the cable surface. A feature extraction algorithm (such as normal vector calculation, curvature analysis, etc.) is used to obtain the geometric features of the cable surface. Normal vector calculation can help understand the direction of the surface, and curvature analysis reveals the degree of concavity of the surface. The extracted surface detail feature data (such as normal vector, curvature value, etc.) is recorded in a data table to provide a basis for subsequent defect detection. A visual inspection system is designed and built, combined with a high-resolution camera and appropriate lighting to improve the accuracy and sensitivity of defect detection. A suitable defect detection algorithm (such as edge detection, texture analysis or machine learning algorithm in image processing) is selected to identify crimping points. Defects at the joints. Match the extracted cable surface detail features with the pre-defined defect feature library. By comparing different feature values, identify potential defect areas. Record the detected defect data (such as defect type, location, severity, etc.) in the database for subsequent analysis and processing. Collect and organize the defect data obtained from the visual inspection system, including the spatial location, shape and size information of the defects. Apply morphological analysis methods (such as opening, closing, corrosion and expansion, etc.) to extract the morphological features of the defects. These features include the area, perimeter, shape index, etc. of the defects. Statistical defect morphological features, analyze the distribution and eigenvalues ​​of various defects, which can help identify which defects are more common and their potential impacts. Record the generated defect morphological features in the data table for subsequent positioning and visualization rendering. Calibrate the detected defects on the 3D model of the crimping point to determine the specific location of the defects. You can use 3D visualization software (such as MeshLab or 3D Slicer) for calibration and visualization, convert the two-dimensional coordinate information of the defect area into coordinates in three-dimensional space to ensure the accurate positioning of the defect position, use the internal and external parameters of the camera and the information of the defect feature points to perform coordinate conversion, confirm the located defect area, ensure the accuracy of calibration, which can be done through manual inspection or algorithm verification, record the accurately located defect data (such as three-dimensional coordinates, defect type) in the database, provide necessary information for subsequent visualization rendering, select a suitable three-dimensional visualization tool (such as Blender, Unity or ParaView) for defect rendering, ensure that the software can support the import and rendering of three-dimensional models, import the three-dimensional registration optimization model of the crimping point into the selected visualization software, and make necessary settings, including lighting, material and viewing angle adjustment, render the defect area on the three-dimensional model according to the defect positioning data and morphological characteristics, and use color coding, transparency or texture mapping to highlight the characteristics of the defect.After rendering is completed, check the effect to ensure that the visualization results of the defects meet expectations. Finally, save the rendering results as images or video files for subsequent display and analysis.

[0043] In this embodiment, the specific steps of step S5 are: Step S51: quantifying the degree of deformation of the defect area of ​​the crimping point defect morphology rendering model to generate a quantified value of the defect area deformation; Step S52: analyzing the stress distribution of the defective area based on the quantified value of the deformation of the defective area to obtain stress distribution characteristics of the defective area; Step S53: simulating stress field changes according to stress distribution characteristics of the defect area to generate stress field change simulation data of the defect area; Step S54: Perform a comprehensive crimping quality assessment on the stress field change simulation data of the defective area to generate a cable crimping quality assessment report.

[0044] In this embodiment, the crimping point defect morphology rendering model is loaded into 3D modeling software (such as Blender or MeshLab) to ensure that the model can be analyzed in detail and the degree of deformation metrics are defined. For example, by comparing the geometric shape of the defect area with that in the normal state, using parameterized surfaces or benchmark models for comparison, and using shape analysis algorithms such as Hausdorff distance or root mean square error (RMSE) to calculate the geometric difference between the defect area and the ideal state. These algorithms can quantify the degree of deformation of the model, record the calculated quantitative values ​​of the deformation of the defect area, and generate a detailed report outlining the deformation of the model and related parameters. These quantitative values ​​will be used for subsequent Provide basic data for stress analysis, build a stress analysis model in finite element analysis (FEA) software (such as ANSYS, Abaqus), import the crimping point defect morphology rendering model, set material properties and boundary conditions, calculate stress distribution based on the deformation quantification value of the defect area, use static or dynamic analysis methods to simulate the stress state under external loads, analyze the stress concentration area, stress gradient and other characteristics through the calculated stress distribution results, pay special attention to the stress state in the defect area, determine whether it exceeds the yield strength of the material, record the stress distribution characteristics in graphical and numerical forms, generate stress distribution diagrams and related data tables, and this information will provide a basis for subsequent stress field change simulations. According to the data, in the finite element analysis software, set the simulation conditions, including load type, application position and time series, etc. These conditions should be consistent with the actual working environment to ensure the accuracy of the simulation, run the stress field change simulation, and analyze the stress changes in the defective area under different load conditions. You can use time domain analysis or frequency domain analysis to observe the dynamic changes of the stress field, extract the stress data in the simulation process, analyze the change trend and influencing factors of the stress field, and focus on the stress response of the defective area to evaluate its impact on the overall crimping quality. The generated stress field change simulation data is recorded in the database, and a visual graph is generated to show the dynamic changes of the stress field. These simulation results will provide important information for crimping quality evaluation. Determine the indicators for comprehensive evaluation of crimping quality, including uniformity of stress distribution, stress concentration, degree of deformation, etc., set reasonable thresholds for each indicator for evaluation, compare and analyze the simulated stress field change data with the set quality standards, and use statistical analysis and visualization tools to determine whether the crimping quality meets the requirements. Based on the analysis results, write a cable crimping quality evaluation report, describing in detail the quality status of the crimping points, the impact of defects and recommended measures. The report should contain charts, data and conclusions for the convenience of relevant personnel to understand. Archive the evaluation report and share it with relevant teams to ensure that all stakeholders can obtain the evaluation results of the crimping quality, which will provide a reference for subsequent quality monitoring and improvement.

[0045] In this embodiment, the specific steps of step S6 are: Step S61: performing crimping defect diagnosis based on the cable crimping quality assessment report to generate crimping defect diagnosis data; Step S62: making adaptive correction decisions on the crimping defect diagnosis data to construct a cable crimping defect correction strategy; Step S63: dynamically migrate and optimize the cable crimping defect correction strategy, and construct a cable crimping defect optimization model to perform cable crimping detection and diagnosis optimization operations.

[0046] In this embodiment, based on the data in the evaluation report, statistical analysis or machine learning models (such as decision trees, support vector machines, etc.) are used to identify potential patterns and causes of crimping defects, compare historical data and existing defects, find similarities and patterns, and provide a basis for subsequent diagnosis. Based on the identified defect patterns, crimping defect diagnostic data are generated, which include information such as defect type, frequency of occurrence, possible causes and impact level. These data are structured and stored in a database for subsequent analysis and formulation of correction strategies. The generated crimping defect diagnostic data are sorted out and a diagnostic report is written. The report should include a detailed description of the defects, diagnostic results and recommended measures. This report will provide a basis for the formulation of subsequent correction strategies. Based on the diagnostic data, a set of correction decision models are constructed. A rule-based system or decision tree model can be used to map the diagnostic results to specific corrective measures. Adaptive algorithms (such as reinforcement learning or fuzzy logic control) are used to dynamically adjust correction decisions. These algorithms can continuously optimize correction strategies based on real-time data and feedback to ensure the effectiveness and timeliness of corrective measures. The policy model is used to formulate specific correction strategies. These strategies should include adjustments to the crimping process, improvements in material selection, optimization of process parameters, etc. to reduce the possibility of defects. The generated correction strategies are recorded in the system, and an implementation plan is formulated, including responsible persons, implementation steps and timetables. This will ensure that the correction measures can be effectively implemented and track progress. Based on historical data and real-time monitoring data, a dynamic optimization model is constructed. The model should be able to analyze data changes in the crimping process in real time, adjust correction strategies, apply transfer learning technology, migrate and optimize correction strategies under different situations, use historical successful correction examples to quickly adapt to new situations, and improve the efficiency and accuracy of correction strategies. The optimization model is verified, and the effectiveness of the model is evaluated through actual test data and correction effects. If the model is found to be inaccurate, the model parameters and structure should be adjusted in time to improve its performance. The optimization model is implemented in actual cable crimping detection and correction operations, and a feedback mechanism is established. By collecting the results after implementation, the model is further improved to provide a basis for future crimping defect diagnosis and correction.

[0047] In this embodiment, a cable crimping detection system based on image processing is provided, which is used to execute the cable crimping detection method based on image processing as described above, including: A background elimination module is used to obtain a multi-angle high-resolution monitoring image of the cable crimping point; and perform dynamic background elimination processing on the multi-angle high-resolution monitoring image to generate a background-eliminated cable crimping point monitoring image; The posture deviation correction module is used to perform posture jitter analysis of the background-eliminated cable crimping point monitoring image angle by angle, and perform posture deviation correction processing to construct a posture correction monitoring image; The 3D point cloud module is used to identify the characteristic points of the cable crimping area in the posture correction monitoring image and perform 3D point cloud modeling to obtain a 3D registration optimization model of the crimping point; The defect visual detection module is used to perform visual detection of crimping point defects on the crimping point three-dimensional registration optimization model, and perform visual rendering to build a crimping point defect morphology rendering model; The crimping quality assessment module is used to quantify the degree of deformation of the defect area of ​​the crimping point defect morphology rendering model and conduct a comprehensive assessment of the crimping quality, and generate a cable crimping quality assessment report; The correction decision module is used to make adaptive correction decisions based on the cable crimping quality assessment report, perform dynamic migration optimization, and build a cable crimping defect optimization model to perform cable crimping detection and diagnosis optimization operations.

[0048] The present invention can remove unnecessary interference factors such as background noise, lighting changes, etc. through background elimination, thereby improving the clarity and quality of the image and helping the subsequent processing module to more accurately identify the crimping points. Background elimination effectively reduces errors caused by environmental changes or equipment status fluctuations. By dynamically adjusting the elimination algorithm, the background elimination module can cope with changes in different environments to ensure the accuracy of the image content. Through posture deviation correction, it is ensured that images at different angles can be aligned in the same coordinate system, eliminating errors caused by unstable camera posture, thereby improving image consistency. Posture correction helps to accurately locate the crimping points and reduce errors caused by shooting angles. In the process of multi-angle image synthesis, the corrected images can be better integrated to ensure the data accuracy during 3D modeling, thereby improving the 3D modeling quality of the crimping point. Through accurate feature point recognition and 3D point cloud modeling, the 3D spatial information of the cable crimping point can be efficiently constructed, which provides detailed geometric data for subsequent defect detection and quality assessment. 3D registration optimization helps to solve the errors and mismatches in images at different angles, ensuring that the constructed 3D point cloud model has high accuracy. The 3D point cloud model can provide multiple functions for defect detection and stress analysis of crimping points. Dimensional data support facilitates comprehensive evaluation of crimping quality. Through advanced image processing technology, various defects at crimping points can be accurately identified, including poor crimping, surface cracks, deviations, etc. This provides an automated detection solution for the production line and reduces manual errors. Through rendering technology, the detected defects are visualized, allowing engineers to intuitively see the shape, location and severity of the defects, thereby making more accurate repair decisions. Through quantitative analysis of the degree of deformation in the defective area, the crimping quality assessment module can accurately assess the quality of the crimping points, avoiding the subjectivity of traditional quality assessment methods. This module comprehensively considers deformation, stress, defects, and other factors. The system can comprehensively evaluate the quality of crimping points from multiple dimensions based on factors such as shape and morphology, and provide more accurate quality data. Through adaptive correction decision-making, the system can automatically adjust the repair plan according to the crimping quality evaluation results without manual intervention, thereby improving the accuracy and efficiency of the repair. Dynamic migration optimization can continuously adjust the correction strategy according to changes in actual working conditions to ensure that the repair plan is always effective in different production environments, thereby continuously improving the detection and repair efficiency. The correction decision module can combine historical data and real-time data for optimization, accurately diagnose crimping defects and propose effective repair plans, reducing the occurrence of quality problems and improving the stability and reliability of the crimping process.

[0049] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0050] The above is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A cable crimping detection method based on image processing, characterized in that: The following steps are involved: Step S1: Acquire a multi-angle high-resolution monitoring image of the cable crimping point; Performing dynamic background elimination processing on the multi-angle high-resolution monitoring image to generate a background-eliminated cable crimping point monitoring image; Step S2: performing posture jitter analysis of each angle of the background-eliminated cable crimping point monitoring image, and performing posture deviation correction processing to construct a posture correction monitoring image; Step S3: identifying feature points of the cable crimping area on the posture correction monitoring image, and performing three-dimensional point cloud modeling, thereby obtaining a three-dimensional registration optimization model of the crimping point; Step S4: performing visual inspection of crimping point defects on the crimping point three-dimensional registration optimization model, and performing visual rendering to construct a crimping point defect morphology rendering model; Step S5: quantifying the degree of deformation of the defect area of ​​the crimping point defect morphology rendering model and performing a comprehensive crimping quality assessment to generate a cable crimping quality assessment report; Step S6: Make adaptive correction decisions based on the cable crimping quality assessment report, perform dynamic migration optimization, and build a cable crimping defect optimization model to perform cable crimping detection and diagnosis optimization operations.

2. The cable crimping detection method based on image processing according to claim 1 is characterized in that: The specific steps of step S1 are: Step S11: Acquire a multi-angle high-resolution monitoring image of the cable crimping point; Step S12: Calculate the light intensity at each angle on the multi-angle high-resolution monitoring image to extract the light intensity parameters at each angle; Step S13: performing multi-angle image brightness difference analysis on the multi-angle high-resolution monitoring image to generate image brightness difference values ​​between multiple angles; Step S14: performing dynamic light source deviation compensation calculation on the image brightness difference values ​​between multiple angles based on the illumination intensity parameter of each angle to generate a multi-angle dynamic light source deviation compensation value; Step S15: optimizing the light source error of the multi-angle high-resolution monitoring image based on the multi-angle dynamic light source deviation compensation value, thereby obtaining a light source error optimized monitoring image; Step S16: Perform dynamic background elimination processing on the light source error optimization monitoring image to generate a background eliminated cable crimping point monitoring image.

3. The cable crimping detection method based on image processing according to claim 2 is characterized in that: The specific steps of step S15 are: Perform deep background semantic segmentation on the light source error optimization monitoring image to extract the image background area; Calculating the background texture gradient amplitude of the image background area to generate the background texture gradient amplitude; Analyze the texture direction distribution of the image background area to obtain the background texture direction distribution characteristics; Mining dynamic background changes based on background texture gradient amplitude and background texture direction distribution features to extract dynamic background texture change features; The background area pixels of the light source error optimization monitoring image are eliminated according to the dynamic background texture change characteristics to generate a background-eliminated cable crimping point monitoring image.

4. The cable crimping detection method based on image processing according to claim 1, characterized in that: The specific steps of step S2 are: Step S21: defining a unified acquisition timestamp; performing time difference analysis on the background-eliminated cable crimping point monitoring image according to the unified acquisition timestamp, and identifying image acquisition time difference values ​​at different angles; Step S22: removing asynchronous images according to the time difference values ​​of image acquisition at different angles to obtain a time-synchronous monitoring image; Step S23: performing multi-angle camera tag calculation on the time-synchronized monitoring image to generate camera intrinsic parameters and extrinsic parameters for each angle; Step S24: performing image posture jitter analysis at each angle based on the camera intrinsic parameters and extrinsic parameters at each angle, and generating image displacement errors at each angle; Step S25: performing posture deviation correction processing on the time-synchronized monitoring image based on the image displacement error at each angle to construct a posture-corrected monitoring image.

5. The cable crimping detection method based on image processing according to claim 1, characterized in that: The specific steps of step S3 are: Step S31: performing cable crimping area feature point recognition on the posture correction monitoring image, and marking a plurality of cable crimping feature points; Step S32: performing triangulation processing on a plurality of cable crimping feature points to obtain the three-dimensional position coordinates of each feature point; Step S33: performing a cable crimping area structure analysis based on a plurality of cable crimping feature points to generate a topological structure feature of the cable crimping area; Step S34: performing three-dimensional point cloud modeling on the topological structure features of the cable crimping area based on the three-dimensional position coordinates of each feature point, and constructing a three-dimensional point cloud model of the cable crimping point; Step S35: performing global registration optimization on the three-dimensional point cloud model of the cable crimping point, thereby obtaining a three-dimensional registration optimization model of the crimping point.

6. The cable crimping detection method based on image processing according to claim 1, characterized in that: The specific steps of step S4 are: Step S41: performing cable surface detail analysis on the crimping point three-dimensional registration optimization model to extract cable surface detail features; Step S42: Performing visual inspection of crimping point defects based on the cable surface detail features to obtain crimping point defect data; Step S43: performing defect morphology analysis on the crimping point defect data to generate crimping point defect morphology features; Step S44: accurately locating the defect area of ​​the crimping point defect data to obtain crimping point defect location data; Step S45: Based on the crimping point defect positioning data, the crimping point defect morphological features are visualized and rendered on the crimping point three-dimensional registration optimization model to construct a crimping point defect morphological rendering model.

7. The cable crimping detection method based on image processing according to claim 1, characterized in that: The specific steps of step S5 are: Step S51: quantifying the degree of deformation of the defect area of ​​the crimping point defect morphology rendering model to generate a quantified value of the defect area deformation; Step S52: analyzing the stress distribution of the defective area based on the quantified value of the deformation of the defective area to obtain stress distribution characteristics of the defective area; Step S53: simulating stress field changes according to stress distribution characteristics of the defect area to generate stress field change simulation data of the defect area; Step S54: Perform a comprehensive crimping quality assessment on the stress field change simulation data of the defective area to generate a cable crimping quality assessment report.

8. The cable crimping detection method based on image processing according to claim 1, characterized in that: The specific steps of step S6 are: Step S61: performing crimping defect diagnosis based on the cable crimping quality assessment report to generate crimping defect diagnosis data; Step S62: making adaptive correction decisions on the crimping defect diagnosis data to construct a cable crimping defect correction strategy; Step S63: dynamically migrate and optimize the cable crimping defect correction strategy, and construct a cable crimping defect optimization model to perform cable crimping detection and diagnosis optimization operations.

9. A cable crimping detection system based on image processing, characterized in that: The method for performing the cable crimping detection method based on image processing as claimed in claim 1 comprises: A background elimination module is used to obtain a multi-angle high-resolution monitoring image of the cable crimping point; and perform dynamic background elimination processing on the multi-angle high-resolution monitoring image to generate a background-eliminated cable crimping point monitoring image; The posture deviation correction module is used to perform posture jitter analysis of the background-eliminated cable crimping point monitoring image angle by angle, and perform posture deviation correction processing to construct a posture correction monitoring image; The 3D point cloud module is used to identify the characteristic points of the cable crimping area in the posture correction monitoring image and perform 3D point cloud modeling to obtain a 3D registration optimization model of the crimping point; The defect visual detection module is used to perform visual detection of crimping point defects on the crimping point three-dimensional registration optimization model, and perform visual rendering to build a crimping point defect morphology rendering model; The crimping quality assessment module is used to quantify the degree of deformation of the defect area of ​​the crimping point defect morphology rendering model and conduct a comprehensive assessment of the crimping quality, and generate a cable crimping quality assessment report; The correction decision module is used to make adaptive correction decisions based on the cable crimping quality assessment report, perform dynamic migration optimization, and build a cable crimping defect optimization model to perform cable crimping detection and diagnosis optimization operations.

Images