Automobile wire harness precision fastener surface flaw detection system based on visual identification

Abstract

The invention belongs to the technical field of automobile wire harness precision fastener detection, and particularly relates to an automobile wire harness precision fastener surface flaw detection system based on visual identification. Comprising a multispectral synchronous imaging acquisition module, a dynamic threshold adaptive enhancement module, a microstructure feature deconstruction module, a multidimensional flaw mapping classification module, an intelligent decision feedback execution module and an operation supervision terminal. According to the method, the original data adapting to the material optical characteristics of the automobile wire harness precise fastener are obtained, illumination unevenness and noise interference are eliminated through self-adaptive image enhancement based on local feature statistics, geometric and texture attributes of flaws are quantified through multi-scale feature extraction and morphology reconstruction, and the precision of the automobile wire harness precise fastener is improved. According to the method, the surface defect classification result and the confidence score of the precise fastener of the automobile wire harness are output, a corresponding execution mechanism is driven according to the surface defect detection result, a high-precision, high-efficiency and intelligent quality control solution is provided for production of the automobile fastener, and the labor cost and the defective rate are remarkably reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile wiring harness precision fastener detection, and in particular to a surface defect detection system for automobile wiring harness precision fasteners based on visual recognition. Background Art

[0002] Automotive wiring harness precision fasteners are key components used to secure, protect, and connect wiring harnesses in automotive electrical systems. With the development of new energy vehicles and autonomous driving technologies, the requirements for the accuracy, materials, and functions of automotive wiring harness precision fasteners will continue to increase, driving the industry towards intelligent and lightweight development. During the production process of automotive wiring harness precision fasteners, it is necessary to detect surface quality defects. The surface quality of automotive wiring harness fasteners directly affects the reliability and safety of wiring harness connections. Traditional manual inspection methods are not only inefficient, but also difficult to ensure the accuracy of surface defect detection results. Existing visual inspection systems, however, mostly rely on a single light source or fixed algorithm, making them difficult to handle complex background interference, minor defects, and differences in material reflective properties. Furthermore, they struggle to analyze and judge the accuracy and discrepancies in the precise measurement results of automotive wiring harness fasteners during the inspection process. This hinders the smooth operation, efficiency, and reliability of system inspections, resulting in low levels of intelligence and significant challenges in oversight. In view of the above technical defects, a solution is now proposed. Summary of the Invention

[0003] The purpose of the present invention is to provide a surface defect detection system for automobile wiring harness precision fasteners based on visual recognition, which solves the problems that the existing technology relies on a single light source or fixed algorithm, is difficult to deal with complex background interference, minor defects and differences in material reflective properties, and is difficult to reasonably analyze and judge the accuracy of the precision measurement results and anomalies of automobile wiring harness precision fasteners in the detection process, which is not conducive to ensuring the smooth operation and high efficiency and reliability of the system detection, and has a low degree of intelligence and is difficult to supervise.

[0004] To achieve the above object, the present invention provides the following technical solutions: The visual recognition-based surface defect detection system for automotive wiring harness precision fasteners includes a multi-spectral synchronous imaging acquisition module, a dynamic threshold adaptive enhancement module, a micro-morphological feature deconstruction module, a multi-dimensional defect mapping and classification module, an intelligent decision-making feedback execution module, and an operation monitoring terminal. The multi-spectral synchronous imaging acquisition module acquires raw data on the optical properties of the materials of automotive wiring harness precision fasteners through the coordinated control of multi-band light sources and the simultaneous acquisition of high-precision images. The dynamic threshold adaptive enhancement module receives multispectral images and adaptively enhances them based on local feature statistics to eliminate uneven illumination and noise interference. It then outputs the enhanced images to the micro-morphology feature deconstruction module. The micro-morphology feature deconstruction module quantifies the geometric and textural properties of defects through multi-scale feature extraction and morphology reconstruction, and sends the deconstructed feature maps to the multi-dimensional defect mapping and classification module. The multi-dimensional defect mapping classification module receives the deconstructed feature map and outputs the surface defect classification results and confidence scores of the automotive wiring harness precision fasteners to the intelligent decision-making feedback execution module and the operation supervision terminal; the intelligent decision-making feedback execution module drives the corresponding actuator according to the surface defect detection results and sends the execution information to the operation supervision terminal.

[0005] Furthermore, the multi-spectral synchronous imaging acquisition module adopts a ring-shaped LED array light source, integrates independent control units for three bands of visible light, near-infrared and ultraviolet, and an FPGA-based synchronous trigger controller, which enables different spectral images to be spatially aligned within microsecond timing, and suppresses the mirror reflection of the metal surface through a polarizer group to retain the diffuse reflection characteristic information.

[0006] Furthermore, the operation process of the dynamic threshold adaptive enhancement module includes: The image is segmented into superpixel units, and the grayscale histogram distribution of each unit is calculated. An improved Otsu algorithm is used to dynamically generate local thresholds and perform nonlinear stretching on low-contrast areas. Wavelet denoising and guided filtering are combined to retain defect edge details while suppressing high-frequency noise.

[0007] Furthermore, the operation process of the micro-morphological feature deconstruction module includes: Directional texture features are extracted through Gabor filter groups; a morphological gradient reconstruction algorithm is applied to generate a three-dimensional topological map of the surface morphology; and SIFT feature point matching is used to achieve spatial positioning of defects across images.

[0008] Furthermore, the multi-dimensional defect mapping and classification module builds a defect feature library containing several groups of samples, covering typical defects such as burrs, scratches, and oxidation. During operation, the multi-dimensional defect mapping and classification module uses an improved twin network for feature similarity matching and introduces a confidence decay mechanism to trigger a secondary verification process for low-probability results. The intelligent decision-making feedback execution module designs a graded rejection strategy, directly triggers the pneumatic rejection device for high-confidence defective parts, generates visual reports for edge cases, and prompts manual re-inspection.

[0009] Furthermore, the operation supervision terminal communicates with the result accuracy verification module, and the result accuracy verification module samples and inspects the precision fasteners of the automobile wiring harness that have been inspected, and analyzes the inspection and review results to generate a high-precision result signal or a low-precision result signal, and sends an alarm message to the operation supervision terminal when a low-precision result signal is generated.

[0010] Furthermore, the specific analysis process of the result accuracy verification module is as follows: The inspection and review results of all randomly inspected automotive wiring harness precision fasteners are obtained. If the similarity between the surface defect review results of the corresponding automotive wiring harness precision fasteners and the previous inspection results is lower than the preset similarity threshold, the corresponding automotive wiring harness precision fasteners are marked as detection objects; the number of detection objects is obtained and the ratio is calculated with the total number of randomly inspected automotive wiring harness precision fasteners to obtain a detection statistical value, and the detection statistical value is numerically compared with the preset detection statistical threshold. If the detection statistical value exceeds the preset detection statistical threshold, a low-precision result signal is generated; if the detection statistical value does not exceed the preset detection statistical threshold, a high-precision result signal is generated.

[0011] Furthermore, the result accuracy verification module is communicatively connected to the entry / exit monitoring and judgment module, which is communicatively connected to the input impact assessment module and the output execution assessment module. The result accuracy verification module sends a high-precision result signal to the entry / exit monitoring and judgment module. When the entry / exit monitoring and judgment module receives the high-precision result signal, it monitors and assesses the conveying condition of the conveyor belt for the automotive wiring harness precision fasteners through the input impact assessment module, generates an input qualified signal or an input alarm signal accordingly, and sends the input qualified signal or the input alarm signal to the entry / exit monitoring and judgment module. And through the output execution evaluation module, the intelligent decision feedback execution module monitors and evaluates the execution status of the automobile wiring harness precision fasteners, and generates an output qualified signal or an output alarm signal based on this, and sends the output qualified signal or the output alarm signal to the input and output monitoring judgment module; when the input and output monitoring judgment module receives the input alarm signal or the output alarm signal, it generates an alarm message and sends it to the operation supervision terminal.

[0012] Furthermore, the specific operation process of the input impact assessment module is as follows: During the conveying process of the conveyor belt, the real-time conveying speed of the conveyor belt is obtained, the variance of all real-time conveying speeds within a unit time is calculated to obtain a speed fluctuation value, the speed fluctuation value is numerically compared with a preset speed fluctuation threshold, and if the speed fluctuation value exceeds the preset speed fluctuation threshold, an input alarm signal is generated; If the transmission speed fluctuation value does not exceed the preset transmission speed fluctuation threshold, the difference between the real-time transmission speed and the set standard transmission speed is calculated and the absolute value is taken to obtain the speed difference detection value, the number of times the speed difference detection value exceeds the preset speed difference detection threshold in a unit time is marked as the speed difference abnormal value, and all speed difference detection values ​​in a unit time are averaged to obtain the speed difference characteristic value, the speed difference abnormal value and the speed difference characteristic value are numerically compared with the preset speed difference abnormal threshold and the preset speed difference characteristic threshold respectively, and if the speed difference abnormal value or the speed difference characteristic value exceeds the corresponding preset threshold, an input alarm signal is generated; If the speed difference anomaly and the speed difference characteristic value exceed the corresponding preset thresholds, then when the conveyor belt is in a stopped state to cooperate with the acquisition operation of the multi-spectral synchronous imaging acquisition module, the jitter amplitude and jitter frequency of the conveyor belt are obtained, and the jitter amplitude and jitter frequency are numerically compared with the preset jitter amplitude threshold and the preset jitter frequency threshold respectively. If the jitter amplitude or the jitter frequency exceeds the corresponding preset threshold, it is determined that the conveyor belt is in a stopped and unstable state; The duration of the conveyor belt in the stop unstable state per unit time is obtained and the ratio thereof is calculated with the total duration of the conveyor belt in the stop state per unit time to obtain the stop unstable detection time value, and the duration ratio of the conveyor belt in the stop unstable state in a single stop state is obtained. The duration ratio value is numerically compared with the corresponding duration ratio threshold value. If the duration ratio value exceeds the preset duration ratio threshold value, the stop non-optimal symbol TX-1 is assigned to the corresponding stop state; The number of times the stop non-optimal symbol TX-1 is assigned per unit time is obtained and the ratio thereof is calculated with the total number of times the conveyor belt is in a stopped state to obtain a stop non-optimal detection value, and the average value of the vibration amplitude and the average value of the vibration frequency when the conveyor belt is in a stopped state per unit time are marked as the amplitude characteristic value and the frequency characteristic value respectively; the stop stability evaluation value is obtained by weighted summing up the stop non-optimal detection time value, the stop non-optimal detection value, the amplitude characteristic value and the frequency characteristic value, and the stop stability evaluation value is numerically compared with the preset stop stability evaluation threshold. If the stop stability evaluation value exceeds the preset stop stability evaluation threshold, an input alarm signal is generated; if the stop stability evaluation value does not exceed the preset stop stability evaluation threshold, an input qualified signal is generated.

[0013] Furthermore, the specific analysis process of the output execution evaluation module is as follows: Obtain all execution devices involved in the intelligent decision feedback execution module, collect the operation delay time of the corresponding execution device when the intelligent decision feedback execution module issues a driving instruction, compare the operation delay time with the corresponding preset operation delay time threshold, and if the operation delay time exceeds the corresponding preset operation delay time threshold, mark the corresponding operation delay time as an operation delay abnormal value; The number of abnormal operation delay values ​​corresponding to the corresponding execution device in unit time is obtained and the ratio thereof is calculated with the number of execution times of the corresponding execution device in unit time to obtain the abnormal operation delay value, and the average value of all operation delay durations corresponding to the corresponding execution device in unit time is marked as the delay characteristic value; the abnormal operation delay value and the delay characteristic value are numerically compared with the corresponding preset abnormal operation delay threshold value and the preset delay characteristic threshold value respectively; if the abnormal operation delay value or the delay characteristic value exceeds the corresponding preset threshold value, the corresponding execution device is marked as a slow-action device; if there is a slow-action device in unit time, an output alarm signal is generated; if there is no slow-action device in unit time, an output qualified signal is generated.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention obtains raw data on the optical properties of materials suitable for automotive wiring harness precision fasteners, performs adaptive image enhancement based on local feature statistics to eliminate uneven illumination and noise interference, quantifies the geometric and texture properties of defects through multi-scale feature extraction and morphology reconstruction, outputs surface defect classification results and confidence scores for automotive wiring harness precision fasteners, and drives corresponding actuators based on the surface defect detection results. This provides a high-precision, high-efficiency, and intelligent quality control solution for automotive fastener production, significantly reducing labor costs and defective product rates. 2. In the present invention, the inspection of the automobile wiring harness precision fasteners is sampled and inspected through the result accuracy verification module to judge the accuracy of the results. When a low-precision result signal is generated, a cause investigation and analysis is conducted and corresponding optimization and improvement measures are taken to ensure the accuracy of the subsequent defect detection results for automobile wiring harness precision fasteners. When a high-precision result signal is generated, the input monitoring and evaluation analysis results and the output monitoring and evaluation analysis results are used to comprehensively judge the input and output anomalies. When the corresponding alarm information is generated, a cause investigation and analysis is conducted and reasonable improvement measures are taken to ensure the stability and efficiency of the system detection process and improve the system's performance in detecting surface defects of automobile wiring harness precision fasteners. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings; Figure 1 This is a system block diagram of Embodiment 1 of the present invention; Figure 2 This is a system block diagram of Example 2 of the present invention. DETAILED DESCRIPTION

[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0017] Example 1: Figure 1 As shown in the figure, the surface defect detection system for automotive wiring harness precision fasteners based on visual recognition proposed by the present invention includes a multi-spectral synchronous imaging acquisition module, a dynamic threshold adaptive enhancement module, a micro-morphological feature deconstruction module, a multi-dimensional defect mapping and classification module, an intelligent decision feedback execution module, a result accuracy verification module and an operation supervision terminal; The multi-spectral synchronous imaging acquisition module uses multi-band light source coordinated control and high-precision image synchronous acquisition to obtain raw data on the optical properties of materials suitable for automotive wiring harness precision fasteners. This overcomes the limitations of single-spectrum imaging and improves the detection rate of tiny cracks (≥0.02mm) and coating defects. It should be noted that the multi-spectral synchronous imaging acquisition module uses a ring-shaped LED array light source, integrates three independent control units for visible light (400-700nm), near-infrared (850nm) and ultraviolet (365nm), and an FPGA-based synchronous trigger controller, which enables different spectral images to be spatially aligned within microsecond timing, and suppresses the mirror reflection of the metal surface through a polarizer group to retain the diffuse reflection characteristic information.

[0018] The dynamic threshold adaptive enhancement module receives multispectral images and adaptively enhances them based on local feature statistics to eliminate uneven illumination and noise interference. It then outputs the enhanced images to the micro-morphological feature deconstruction module, providing high signal-to-noise ratio data for subsequent analysis. The operation process includes: The image is segmented into superpixel units, and the grayscale histogram distribution of each unit is calculated. An improved Otsu algorithm is used to dynamically generate local thresholds and perform nonlinear stretching on low-contrast areas. Wavelet denoising and guided filtering are combined to retain defect edge details while suppressing high-frequency noise.

[0019] The micro-morphological feature deconstruction module quantifies the geometric and textural properties of defects through multi-scale feature extraction and morphology reconstruction, and sends the deconstructed feature map to the multi-dimensional defect mapping and classification module, upgrading traditional two-dimensional detection to three-dimensional morphology analysis, improving the ability to recognize three-dimensional defects such as depressions and protrusions; the operation process can be summarized as follows: directional texture feature extraction through Gabor filter group; application of morphological gradient reconstruction algorithm to generate a three-dimensional topological map of the surface morphology; and cross-image spatial positioning of defects through SIFT feature point matching.

[0020] The multi-dimensional defect mapping and classification module receives the deconstructed feature map and outputs the surface defect classification results and confidence scores of automotive wiring harness precision fasteners to the intelligent decision-making feedback execution module and operation supervision terminal; the multi-dimensional defect mapping and classification module constructs a defect feature library containing several groups of samples (2000+ samples), covering typical defects such as burrs, scratches and oxidation. During operation, the multi-dimensional defect mapping and classification module uses an improved Siamese Network for feature similarity matching and introduces a confidence attenuation mechanism to trigger a secondary verification process for low-probability results.

[0021] The intelligent decision-making feedback execution module drives the corresponding actuator according to the surface defect detection results, and sends the execution information to the operation supervision terminal; the intelligent decision-making feedback execution module designs a graded rejection strategy, directly triggers the pneumatic rejection device for high-confidence defective parts, generates a visual report for edge cases and prompts manual re-inspection; preferably, the intelligent decision-making feedback execution module also outputs process optimization suggestions to the previous module (such as adjusting the light source angle or enhancing the threshold) as a terminal decision node, while receiving user interaction instructions and feedback execution status.

[0022] The technical solution of the present invention achieves high-precision identification and full coverage (geometric, surface and material defects) of tiny defects (0.02mm level) on the surface of fasteners by integrating multimodal sensors (such as visible light, infrared, 3D structured light, etc.) with deep learning algorithms. It maintains a detection rate of 99.9% and a false detection rate of less than 0.1% under complex working conditions (strong light, dark field, and highly reflective surfaces). At the same time, it improves the detection efficiency to 0.5 seconds per piece, supports full-process quality traceability and process optimization, significantly reduces labor costs (by more than 80%) and defective rates (by more than 30%), and provides a high-precision, high-efficiency, and intelligent quality control solution for automotive fastener production, promoting the industry's upgrade to intelligent manufacturing.

[0023] The result accuracy verification module samples and inspects the inspected automotive wiring harness precision fasteners, analyzes the inspection and review results to generate a high-precision result signal or a low-precision result signal, and sends an alarm message to the operation supervision terminal when a low-precision result signal is generated to remind the supervisor to conduct a cause investigation and analysis and make corresponding optimization and improvement measures, thereby ensuring the accuracy of the subsequent defect detection results for automotive wiring harness precision fasteners, which is conducive to ensuring the quality of the output automotive wiring harness precision fasteners and avoiding the waste of high-quality products, and improving the system's performance in surface defect detection of automotive wiring harness precision fasteners. The specific analysis process of the result accuracy verification module is as follows: The inspection and review results of all randomly inspected automotive wiring harness precision fasteners are obtained. If the similarity between the surface defect review result of the corresponding automotive wiring harness precision fastener and the previous inspection result is lower than a preset similarity threshold, it indicates that the system's inspection result for the corresponding automotive wiring harness precision fastener is inaccurate, and the corresponding automotive wiring harness precision fastener is marked as a detection object; the number of detection objects is obtained and the ratio is calculated with the total number of randomly inspected automotive wiring harness precision fasteners to obtain a detection statistical value; The detection statistical value is numerically compared with the preset detection statistical threshold. If the detection statistical value exceeds the preset detection statistical threshold, it indicates that the system's detection accuracy for surface defects of corresponding automotive wiring harness precision fasteners is poor, and a low-precision signal is generated; if the detection statistical value does not exceed the preset detection statistical threshold, it indicates that the system's detection accuracy for surface defects of corresponding automotive wiring harness precision fasteners is good, and a high-precision signal is generated.

[0024] Example 2: Figure 2 As shown, the difference between this embodiment and the first embodiment is that the result accuracy verification module is communicatively connected to the entry-exit monitoring and judgment module, and the entry-exit monitoring and judgment module is communicatively connected to the input impact assessment module and the output execution assessment module. The result accuracy verification module sends a high-precision result signal to the entry-exit monitoring and judgment module. When the entry-exit monitoring and judgment module receives the high-precision result signal, it monitors and evaluates the conveying condition of the conveyor belt for the automotive wiring harness precision fasteners through the input impact assessment module, generates an input qualified signal or an input alarm signal based on the result, and sends the input qualified signal or the input alarm signal to the entry-exit monitoring and judgment module. And through the output execution evaluation module, the intelligent decision feedback execution module monitors and evaluates the execution status of the automobile wiring harness precision fasteners, and generates an output qualified signal or an output alarm signal based on this, and sends the output qualified signal or the output alarm signal to the input and output monitoring judgment module; when the input and output monitoring judgment module receives the input alarm signal or the output alarm signal, it generates an alarm message and sends it to the operation supervision terminal. When the operation supervision terminal receives the alarm message, it displays it and issues an early warning to remind the supervisor to conduct a cause investigation and analysis and make reasonable improvement measures, thereby ensuring the stability and efficiency of the system detection process, and further improving the system's performance in detecting surface defects of automobile wiring harness precision fasteners.

[0025] The specific operation process of the input impact assessment module is as follows: during the conveyor belt's conveying movement, the real-time conveying speed of the conveyor belt is obtained, the variance of all real-time conveying speeds within a unit time is calculated to obtain a speed fluctuation value, and the speed fluctuation value is numerically compared with a preset speed fluctuation threshold. If the speed fluctuation value exceeds the preset speed fluctuation threshold, indicating that the speed fluctuation of the conveyor belt during movement is large, which is not conducive to the smooth arrival of the automotive wiring harness precision fasteners, an input alarm signal is generated; If the transmission speed fluctuation value does not exceed the preset transmission speed fluctuation threshold, the difference between the real-time transmission speed and the set standard transmission speed is calculated and the absolute value is taken to obtain the speed difference detection value. The number of times the speed difference detection value exceeds the preset speed difference detection threshold in a unit time is marked as the speed difference abnormal value, and the average of all speed difference detection values ​​in a unit time is calculated to obtain the speed difference characteristic value. The speed difference abnormal value and the speed difference characteristic value are numerically compared with the preset speed difference abnormal threshold and the preset speed difference characteristic threshold respectively; If the speed difference abnormal value or speed difference characteristic value exceeds the corresponding preset threshold, it indicates that the speed execution accuracy of the conveyor belt movement process is poor, which is not conducive to ensuring the smooth arrival of the automotive wiring harness precision fasteners while ensuring the conveying efficiency, and an input alarm signal is generated; If the speed difference anomaly and the speed difference characteristic value exceed the corresponding preset thresholds, then when the conveyor belt is in a stopped state to cooperate with the acquisition operation of the multi-spectral synchronous imaging acquisition module, the jitter amplitude and jitter frequency of the conveyor belt are obtained, and the jitter amplitude and jitter frequency are numerically compared with the preset jitter amplitude threshold and the preset jitter frequency threshold respectively. If the jitter amplitude or the jitter frequency exceeds the corresponding preset threshold, it indicates that the current jitter of the conveyor belt is relatively obvious, and the conveyor belt is judged to be in a stopped and unstable state; The duration of the conveyor belt in the stop unstable state per unit time is obtained and the ratio thereof is calculated with the total duration of the conveyor belt in the stop state per unit time to obtain the stop unstable detection time value, and the duration ratio of the conveyor belt in the stop unstable state in a single stop state is obtained. The duration ratio value is numerically compared with the corresponding duration ratio threshold value. If the duration ratio value exceeds the preset duration ratio threshold value, the stop non-optimal symbol TX-1 is assigned to the corresponding stop state; The number of times the stop non-optimal symbol TX-1 is assigned per unit time is obtained and the ratio thereof is calculated with the total number of times the conveyor belt is in a stopped state to obtain a stop non-optimal detection value, and the average value of the vibration amplitude and the average value of the vibration frequency when the conveyor belt is in a stopped state per unit time are marked as the amplitude characteristic value and the frequency characteristic value, respectively; The stop stability evaluation value is obtained by performing weighted summation on the stop non-stable detection time value, the stop non-optimal detection value, the amplitude characteristic value, and the frequency characteristic value. That is, the stop non-stable detection time value, the stop non-optimal detection value, the amplitude characteristic value, and the frequency characteristic value are respectively assigned corresponding preset weight coefficients, and the stop non-stable detection time value, the stop non-optimal detection value, the amplitude characteristic value, and the frequency characteristic value are respectively multiplied by the corresponding preset weight coefficients, and the sum of the four groups of product results is marked as the stop stability evaluation value. It should be noted that the larger the value of the stop stability evaluation value, the more unstable the conveying stop process per unit time is, and the less conducive it is to ensuring the visual acquisition effect of the precision fasteners of the automotive wiring harness; The stop stability evaluation value is numerically compared with the preset stop stability evaluation threshold. If the stop stability evaluation value exceeds the preset stop stability evaluation threshold, it indicates that the conveying stop process per unit time is unstable, which is not conducive to ensuring the visual acquisition effect of the precision fasteners of the automotive wiring harness, and an input alarm signal is generated; if the stop stability evaluation value does not exceed the preset stop stability evaluation threshold, it indicates that the conveying performance of the precision fasteners of the automotive wiring harness per unit time is generally good, and an input qualified signal is generated.

[0026] Furthermore, the specific analysis process of the output execution evaluation module is as follows: all execution devices involved in the intelligent decision feedback execution module (such as pneumatic rejection devices, report display devices, and voice prompt devices) are obtained, and when the intelligent decision feedback execution module issues a driving instruction, the operation delay time of the corresponding execution device is collected, and the operation delay time is compared with the corresponding preset operation delay time threshold. If the operation delay time exceeds the corresponding preset operation delay time threshold, it indicates that the corresponding execution device has not executed the operation in a timely manner, and the corresponding operation delay time is marked as an operation delay anomaly; Obtain the number of abnormal operation delay values ​​corresponding to the corresponding execution device within a unit time and calculate the ratio thereof with the number of execution times of the corresponding execution device within a unit time to obtain an abnormal operation delay value, and mark the average value of all operation delay durations corresponding to the corresponding execution device within a unit time as the delay characteristic value, and numerically compare the abnormal operation delay value and the delay characteristic value with the corresponding preset abnormal operation delay threshold value and the preset delay characteristic threshold value respectively; If the operation delay anomaly value or the operation delay characteristic value exceeds the corresponding preset threshold, indicating that the execution timeliness performance of the corresponding execution device in unit time is poor, the corresponding execution device will be marked as a slow-action device; if there is a slow-action device in unit time, it indicates that the output side execution performance of the automobile wiring harness precision fastener detection process is poor, and an output alarm signal is generated; if there is no slow-action device in unit time, it indicates that the output side execution performance of the automobile wiring harness precision fastener detection process is good, and an output qualified signal is generated.

[0027] The working principle of the present invention is as follows: when in use, the original data of the material optical properties of the automotive wiring harness precision fasteners are obtained through the multi-spectral synchronous imaging acquisition module, the dynamic threshold adaptive enhancement module performs adaptive image enhancement based on local feature statistics to eliminate uneven illumination and noise interference, the micro-morphological feature deconstruction module quantifies the geometric and texture properties of the defects through multi-scale feature extraction and morphology reconstruction, the multi-dimensional defect mapping classification module outputs the surface defect classification results and confidence scores of the automotive wiring harness precision fasteners, the intelligent decision-making feedback execution module drives the corresponding execution mechanism according to the surface defect detection results, and provides a high-precision, high-efficiency and intelligent quality control solution for the production of automotive fasteners, and the result accuracy verification module samples and inspects the automotive wiring harness precision fasteners that have been inspected and judges the accuracy of the results, and conducts cause investigation and analysis when a low-precision result signal is generated and corresponding optimization and improvement measures are taken to ensure the accuracy of subsequent defect detection results for automotive wiring harness precision fasteners, which is conducive to ensuring the quality of the output automotive wiring harness precision fasteners and avoiding the waste of high-quality products.

[0028] The thresholds, preset values, preset ranges, etc. in the technical solution of the present invention are set for result comparison and analysis in order to determine whether they are good or bad. As for their size, they are set for entry and storage based on a combination of large-scale model analysis of sample data and manual experience, and can also be appropriately adjusted based on seasonal or common-sense influencing conditions. As for the settings of preset weight coefficients, influencing factors, etc., specific numerical values ​​are assigned based on the influence of each parameter on the result, ultimately reflecting the influence on the result. They are also set for entry and storage based on a combination of large-scale model analysis of sample data and manual experience, and can also be appropriately adjusted based on seasonal or common-sense influencing conditions.

[0029] The preferred embodiments of the present invention disclosed above are intended only to help illustrate the present invention. The preferred embodiments do not describe all details in detail, nor do they limit the present invention to specific embodiments. Obviously, many modifications and variations are possible based on the contents of this specification. These embodiments are selected and described in detail in this specification to better explain the principles and practical applications of the present invention and enable those skilled in the art to better understand and utilize the present invention. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. The surface defect detection system for automotive wiring harness precision fasteners based on visual recognition is characterized by: It includes a multi-spectral synchronous imaging acquisition module, a dynamic threshold adaptive enhancement module, a micro-morphological feature deconstruction module, a multi-dimensional defect mapping and classification module, an intelligent decision feedback execution module, and an operation supervision terminal. The multi-spectral synchronous imaging acquisition module acquires raw data on the optical properties of materials suitable for automotive wiring harness precision fasteners through the coordinated control of multi-band light sources and synchronous acquisition of high-precision images. The dynamic threshold adaptive enhancement module receives multispectral images and performs adaptive image enhancement based on local feature statistics to eliminate uneven illumination and noise interference; The micro-morphological feature deconstruction module quantifies the geometric and texture properties of defects through multi-scale feature extraction and morphology reconstruction; the multi-dimensional defect mapping and classification module receives the deconstructed feature map and outputs the surface defect classification results and confidence scores of automotive wiring harness precision fasteners; the intelligent decision-making feedback execution module drives the corresponding actuator according to the surface defect detection results and sends the execution information to the operation supervision terminal.

2. The automotive wiring harness precision fastener surface defect detection system based on visual recognition according to claim 1 is characterized in that: The multi-spectral synchronous imaging acquisition module uses a ring-shaped LED array light source, integrates independent control units for three bands of visible light, near-infrared light, and ultraviolet light, and an FPGA-based synchronous trigger controller, which enables different spectral images to be spatially aligned within microsecond timing, and suppresses the mirror reflection of the metal surface through a polarizer group, retaining the diffuse reflection characteristic information.

3. The automotive wiring harness precision fastener surface defect detection system based on visual recognition according to claim 1 is characterized in that: The operation process of the dynamic threshold adaptive enhancement module includes: The image is segmented into superpixel units, and the grayscale histogram distribution of each unit is calculated. An improved Otsu algorithm is used to dynamically generate local thresholds and perform nonlinear stretching on low-contrast areas. Wavelet denoising and guided filtering are combined to retain defect edge details while suppressing high-frequency noise.

4. The automotive wiring harness precision fastener surface defect detection system based on visual recognition according to claim 1 is characterized in that: The operation process of the micro-morphological feature deconstruction module includes: Directional texture features are extracted through Gabor filter groups; a morphological gradient reconstruction algorithm is applied to generate a three-dimensional topological map of the surface morphology; and SIFT feature point matching is used to achieve spatial positioning of defects across images.

5. The automotive wiring harness precision fastener surface defect detection system based on visual recognition according to claim 1 is characterized in that: The multi-dimensional defect mapping and classification module builds a defect feature library containing several groups of samples. During operation, the module uses an improved twin network for feature similarity matching and introduces a confidence decay mechanism to trigger a secondary verification process for low-probability results. The intelligent decision-making feedback execution module designs a graded rejection strategy, directly triggers the pneumatic rejection device for high-confidence defective parts, generates visual reports for edge cases, and prompts manual re-inspection.

6. The automotive wiring harness precision fastener surface defect detection system based on visual recognition according to claim 1 is characterized in that: The operation supervision terminal communicates with the result accuracy verification module, which samples and inspects the precision fasteners of the automotive wiring harness that have been inspected, and analyzes the inspection and review results to generate a high-precision result signal or a low-precision result signal. When a low-precision result signal is generated, an alarm message is sent to the operation supervision terminal.

7. The automotive wiring harness precision fastener surface defect detection system based on visual recognition according to claim 6 is characterized in that: The specific analysis process of the result accuracy verification module is as follows: The inspection and review results of all sampled automotive wiring harness precision fasteners are obtained. If the detection statistical value exceeds the preset detection statistical threshold, a low-precision result signal is generated; otherwise, a high-precision result signal is generated.

8. The automotive wiring harness precision fastener surface defect detection system based on visual recognition according to claim 6 is characterized in that: The result accuracy verification module is communicatively connected to the entry-out monitoring and judgment module, and the entry-out monitoring and judgment module is communicatively connected to the input impact assessment module and the output execution assessment module. When the entry-out monitoring and judgment module receives the result high-accuracy signal, it monitors and evaluates the conveying condition of the conveyor belt for the automobile wiring harness precision fasteners through the input impact assessment module, and sends the input qualified signal or input alarm signal to the entry-out monitoring and judgment module; and monitors and evaluates the execution condition of the automobile wiring harness precision fasteners through the output execution assessment module, and sends the output qualified signal or output alarm signal to the entry-out monitoring and judgment module; when the entry-out monitoring and judgment module receives the input alarm signal or the output alarm signal, it generates an alarm message and sends it to the operation supervision terminal.

9. The automotive wiring harness precision fastener surface defect detection system based on visual recognition according to claim 8 is characterized in that: The specific operation process of the input impact assessment module is as follows: If the speed fluctuation value exceeds the preset speed fluctuation threshold, an input alarm signal is generated; if the speed fluctuation value does not exceed the preset speed fluctuation threshold, the speed difference abnormal value and the speed difference characteristic value are numerically compared with the preset speed difference abnormal threshold and the preset speed difference characteristic threshold respectively. If the speed difference abnormal value or the speed difference characteristic value exceeds the corresponding preset threshold, an input alarm signal is generated; If the speed difference abnormal value and the speed difference characteristic value exceed the corresponding preset threshold value, the stop stability evaluation value is calculated by weighted summing up the stop non-stable detection time value, the stop non-optimal detection value, the amplitude characteristic value and the frequency characteristic value. If the stop stability evaluation value exceeds the preset stop stability evaluation threshold, an input alarm signal is generated; if the stop stability evaluation value does not exceed the preset stop stability evaluation threshold, an input qualified signal is generated.

10. The automotive wiring harness precision fastener surface defect detection system based on visual recognition according to claim 8, characterized in that: The specific analysis process of the output execution evaluation module is as follows: All execution devices involved in the intelligent decision-making feedback execution module are obtained. If the operation delay abnormality value or the operation delay characteristic value exceeds the corresponding preset threshold, the corresponding execution device will be marked as a slow-action device; if there is a slow-action device within the unit time, an output alarm signal will be generated; otherwise, an output qualified signal will be generated.

Images