A wire harness wire sequence and stripping quality combined detection method based on slot mapping

Abstract

The application discloses a kind of line bundle line sequence and stripping quality combined detection method based on slot mapping, belong to machine vision and line bundle processing quality detection technical field.This method reads slot formula after collecting line bundle end image, establishes slot coordinate system, and maps wire core area to corresponding slot;Wire core color is identified in slot and compared with standard line sequence template, while locating insulating layer end face, bare conductor area and residual skin area, calculating exposed core length, residual skin area, end face skew angle and loose silk state;Fusion line sequence result, stripping quality result and confidence, output slot level abnormal type, measurement, review mark and whole piece OK / NG result.This method can unify line sequence detection and stripping quality detection result, and is suitable for line bundle end processing quality control.

Description

Technical Field

[0001] This invention relates to the fields of machine vision, industrial inspection, and wire harness processing quality control, and in particular to a method for joint detection of wire harness sequence and stripping quality based on slot mapping. Background Technology

[0002] Wire harness products are widely used in electronic equipment, automotive electrical systems, communication equipment, and industrial control systems. During wire harness processing, the color sequence of the wire cores, stripping length, exposed core length, insulation residue, end face misalignment, and loose wire condition directly affect the reliability of subsequent crimping, plugging, and electrical connections. Traditional inspection usually relies on manual visual inspection or independent workstations, which suffers from low efficiency, inconsistent standards, and a high risk of missed inspections.

[0003] Existing wire sequence identification methods mostly focus on color sorting, typically determining the wire core color through fixed scanning areas, color thresholds, or color library matching. While these methods are effective for color sequence detection, they struggle to reliably locate the physical slot corresponding to the wire core when there are deviations in the wire bundle lead-out area, variations in wire core spacing, localized reflections, or end occlusions.

[0004] Wire stripping quality inspection typically focuses on stripped length, exposed core length, residual sheath, loose strands, and the condition of the insulation layer end face. Currently, wire stripping inspection and wire sequence inspection are often performed independently, lacking a unified slot numbering and mapping relationship between the two inspection results. This makes it difficult to directly pinpoint the specific core location from the inspection results, and also hinders the output of slot-level OK / NG judgments.

[0005] Therefore, there is a need for a detection method that can unify wire sequence identification and wire stripping quality inspection into the same slot coordinate system, so that wire core color, stripping size and abnormal status can be mapped to the corresponding slot, thereby improving the stability, interpretability and engineering adaptability of wire harness end quality inspection. Summary of the Invention

[0006] The purpose of this invention is to provide a method for joint detection of wire harness sequence and stripping quality based on slot mapping, which solves the problems in the prior art where wire sequence identification and stripping quality detection are separated, abnormal results are difficult to bind to specific slots, and detection results are not convenient for whole-piece judgment.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: after acquiring the image of the wire harness to be inspected, the product slot formula is read; the inspection area is calibrated, enhanced, and scaled; a slot coordinate system is established based on the slot formula, connector reference, or wire harness outer contour; the wire core area is identified and mapped to the corresponding slot; wire sequence identification and wire stripping quality measurement are performed within the slot; finally, the wire sequence results, wire stripping results, and confidence results are fused to output the slot-level quality status and the overall OK / NG result.

[0008] Compared with the prior art, the present invention has the following beneficial effects: (1) By binding the line sequence and stripping detection results to the same physical slot through the slot coordinate system, the consistency of anomaly localization is improved; (2) It supports multiple implementation methods such as traditional image processing, target detection, segmentation, key points and multi-task visual models, avoiding dependence on a single algorithm; (3) It can output anomaly types such as line sequence error, stripping too long, stripping too short, residual skin, end face deviation, and loose silk at the same time; (4) It improves the detection stability under complex lighting, reflection and occlusion conditions through the low confidence slot verification mechanism. Attached Figure Description

[0009] Figure 1 A flowchart for joint detection of thread sequence and stripping quality provided in an embodiment of the present invention;

[0010] Figure 2 This is a schematic diagram of slot mapping and coordinate transformation provided in an embodiment of the present invention;

[0011] Figure 3 This is a schematic diagram illustrating the matching of the wire core area to the slot according to an embodiment of the present invention;

[0012] Figure 4 This is a schematic diagram of line sequence recognition and template comparison provided in an embodiment of the present invention;

[0013] Figure 5 This is a schematic diagram of the stripping boundary and dimension measurement provided in an embodiment of the present invention;

[0014] Figure 6 This is a schematic diagram of slot-level joint determination provided in an embodiment of the present invention;

[0015] Figure 7 This is a schematic diagram of the low-confidence slot verification process provided in an embodiment of the present invention;

[0016] Figure 8 This is a schematic diagram of the system structure provided in an embodiment of the present invention; Detailed Implementation

[0017] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments described are only for explaining the present invention and are not intended to limit the scope of protection of the present invention.

[0018] like Figure 1 As shown, the present invention first acquires images of the wire harness ends. These images can come from industrial cameras, area scan cameras, line scan cameras, or other image acquisition devices. To accommodate different product models, the system reads the slot formula before starting the inspection. The slot formula includes at least the number of slots, the slot arrangement direction, the standard wire sequence, the standard value of the stripping length, the tolerance of the exposed core length, the residual sheath area threshold, and the verification threshold.

[0019] In one implementation, the detection area can be determined by a fixed ROI, connector frame positioning, end reference edge positioning, or manual teaching. Brightness equalization, noise reduction, white balance, perspective correction, and pixel-to-millimeter calibration are performed on the detection area to map images from different batches and shooting positions to a unified slot coordinate system.

[0020] like Figure 2 As shown, the slot coordinate system can be established based on the product standard template, or it can be established from the connector reference edge, calibration hole, wire harness outer contour, or manual teaching point. For the i-th slot, it can be represented by Si={xi,yi,wi,hi,ci,li,ti}, where xi and yi are the center coordinates of the slot, wi and hi are the width and height of the slot, ci is the standard wire core color, li is the standard value of the stripped wire length, and ti is the set of detection thresholds.

[0021] When the image to be detected undergoes translation, scaling, rotation, or slight perspective changes, the recipe slots can be mapped to the current image through affine transformation, homography transformation, or geometric mapping based on a reference point. The mapping relationship can be expressed as: Pi = T(Pi0), where Pi0 is the coordinate of the recipe slot, Pi is the coordinate of the slot in the current image, and T is the geometric transformation.

[0022] like Figure 3 As shown, after identifying the core regions in the image to be detected, the system maps each core region to a slot. The core regions can be obtained through color segmentation, edge detection, instance segmentation, object detection, or manual teaching. For the j-th core region Lj and the i-th slot Si, the matching cost can be calculated as: Cij = λ1·dij + λ2·(1-IoUij) + λ3·(1-qj).

[0023] Where, dij represents the normalized distance between the center of the core region and the center of the slot, IoUij represents the overlap ratio between the core region and the slot region, qj represents the core region detection confidence, and λ1, λ2, and λ3 are weighting coefficients. The system can use minimum cost matching or Hungarian matching to assign each valid core region to a unique slot.

[0024] like Figure 4 As shown, after slot mapping is completed, the core color features are extracted within each slot. Color features can include RGB, HSV, LAB color mean, color histogram, color moments, or category probabilities output by a color classification model. For similar colors, multi-color space features and effective pixel masks within the slot can be combined to improve discriminative power. Dk = μ1·dh + μ2·ds + μ3·dv + μ4·dlab.

[0025] Where Dk is the distance between the color of the slot to be detected and the k-th standard color, and dh, ds, dv, and dlab represent the differences in different color spaces. The system selects the color with the smallest distance or the highest category probability as the slot color, and compares it with the standard line sequence template slot by slot, outputting results such as color consistency, out-of-order, missing lines, reverse insertion, or low confidence.

[0026] like Figure 5 As shown, wire stripping quality inspection is completed within the same slot. The system locates the insulation layer end face, exposed conductor area, conductor end, and residual sheath area, which can be achieved through edge detection, segmentation, key point regression, or local measurement models. The exposed core length can be expressed as: Lc = k·||Pj - Pi|| + b.

[0027] Where Pi is the end face point of the insulation layer, Pj is the end point of the conductor, k is the pixel-to-millimeter conversion factor, and b is the calibration compensation amount. The system can also calculate the residual area A, end face skew angle θ, loose wire state F, and insulation retention length Li. When any of these parameters exceeds the tolerance range given by the slot recipe, the corresponding stripping anomaly is output.

[0028] like Figure 6 As shown, the line sequence results and stripping quality results are fused using the slot-level quality score. The quality score of the i-th slot can be expressed as: Qi = w1·Si + w2·Li + w3·Bi + w4·Mi.

[0029] Where Si is the line sequence consistency score, Li is the stripping dimension qualification score, Bi is the boundary positioning confidence score, Mi is the slot matching confidence score, and w1 to w4 are weighting coefficients. When Qi is below the threshold or any key quality item is abnormal, the corresponding slot is judged as NG.

[0030] like Figure 7 As shown, when the confidence scores for color recognition, boundary positioning, or slot matching fall below a threshold, the system triggers a review for the corresponding slot. The review can employ methods such as local cropping, image enhancement, scale magnification, secondary color recognition, boundary refitting, or manual confirmation prompts. The review results, along with the initial inspection results, contribute to the final judgment.

[0031] like Figure 8 As shown, this invention can be deployed as a wire harness end quality inspection system, including an image acquisition module, a slot recipe module, a slot mapping module, a wire sequence recognition module, a wire stripping quality inspection module, a joint judgment module, and a result output module. The inspection results can output the slot number, measured color, standard color, stripped length, exposed core length, anomaly type, confidence level, and an overall OK / NG conclusion.

[0032] In one exemplary embodiment, the wire harness to be tested includes 6 slots, with a standard wire sequence of red, yellow, blue, green, white, and black. The standard exposed core length is 5.0 mm, with an allowable tolerance of ±0.3 mm. After the system performs color recognition and wire stripping measurement on each slot, if the 3rd slot is identified as yellow and the standard color is blue, the output indicates an abnormal wire sequence; if the exposed core length of the 5th slot is 5.6 mm, the output indicates an abnormal wire stripping length.

[0033] In one data configuration embodiment, the slot recipe is stored in the form of a table or configuration file. Each slot includes at least a slot number, standard color, color tolerance, standard value for exposed core length, upper limit for exposed core length, lower limit for exposed core length, residual sheath area threshold, and verification threshold. During inspection, the system generates slot objects sequentially according to the slot number to avoid misalignment of subsequent slots due to missed or repeated core inspections.

[0034] In one image enhancement embodiment, the system performs brightness normalization, reflection suppression, and edge enhancement processing on local areas of the slot. For scenarios where the reflectivity of the metal conductor area is strong, brightness upper limit truncation, local contrast enhancement, or the use of polarized light sources can be employed to reduce the impact of bright areas on boundary positioning.

[0035] In one embodiment of stripping boundary fitting, the system first obtains the conductor region mask Mk, then extracts the mask's outer boundary and the skeleton centerline; then, with the slot centerline direction as the main axis, it performs linear fitting on the set of insulating end face points. The end face skew angle can be expressed as the angle between the end face fitting line and the slot's perpendicular direction. θ = arctan(|a|), where a is the slope of the end face fitting line.

[0036] In one embodiment of residual skin detection, the system extracts non-conductor color pixels or non-conductor segmentation masks within the expected area of ​​the exposed conductor and calculates the residual skin area Ar. When Ar exceeds the formula threshold Amax, a residual skin anomaly is output; when discrete small connected regions appear at the edge of the conductor region and their length exceeds the threshold, a filament anomaly is output.

[0037] In one verification embodiment, if the color confidence score of the i-th slot is below 0.75, the stripping boundary confidence score is below 0.70, or the quality score Qi is below 0.80, the system performs a local verification of that slot. During the verification, the slot area can be cropped and enlarged by a scale of 1.5 to 3 times, and color recognition, boundary positioning, and size measurement can be re-executed.

[0038] In one output embodiment, the slot results generated by the system include: slot number, standard color, measured color, color confidence level, exposed core length, stripped wire length, residual sheath area, loose wire status, end face skew angle, wire sequence result, stripping result, verification status, and final OK / NG status. These results can be displayed via an interface, exported as a file, or communicated with a production line traceability system for output.

[0039] In one exemplary test, images of the wire harness ends with different lighting conditions, backgrounds, and stripping lengths were used for verification. The system can output slot-level detection results under normal lighting, low light, localized reflections, and slight wire core misalignment scenarios. The above values ​​are only parameters for this embodiment; in actual applications, they can be adjusted according to camera resolution, product size, and manufacturing tolerances.

[0040] In one production line deployment embodiment, the system can pre-store the slot recipes for the same product model in a database. At the start of inspection, the product model is determined by barcode, QR code, work order number, or manual selection, and the system automatically loads the corresponding recipe. If the number of wire cores detected in the current image does not match the number of slots specified in the recipe, the system does not directly output a qualified result but marks it as an anomaly and initiates a review process.

[0041] In one multi-view detection embodiment, front and side images of the wire harness can be acquired separately. The front image is mainly used for core color, wire sequence, and slot mapping, while the side image is mainly used for stripping length, exposed core length, and end face skew measurement. Detection results from multiple views are correlated using the same slot number to improve detection reliability in complex occlusion scenarios.

[0042] In one color correction embodiment, the system sets up a color calibration block near the detection area or uses a standard background area in the image as a color reference. By estimating the white balance coefficient and luminance coefficient, the color features collected under different light source conditions can be normalized to a unified color space. The normalized color features are then used for color distance calculation and standard line sequence template comparison. Ic = Γ(Iraw, r, g, b), where Iraw is the original image, r, g, and b are color correction parameters, and Ic is the corrected image.

[0043] In one boundary positioning embodiment, the point set on the end face of the insulation layer can be composed of edge points, segmentation mask boundary points, or key point detection results. After outlier removal from the point set, the system performs straight-line fitting. Outlier removal can be achieved using distance thresholding, RANSAC, or median filtering. The fitted end-face line is used to calculate the stripping end-face skew angle and exposed core length.

[0044] In one embodiment of wire stripping length compensation, if the wire core is bent or tilted, the system does not directly use the horizontal distance of the image as the stripping length, but instead calculates the projected distance from the end face to the conductor end along the direction of the wire core centerline. The projected distance can reduce the influence of wire core tilt on the measurement results. Lc = k·|(Pj-Pi)·u|+b, where u is the unit direction vector of the wire core centerline.

[0045] In one embodiment of line sequence anomaly classification, the system categorizes line sequence anomalies into types such as misordered lines, missing lines, multiple lines, low-confidence color, and slot unmapped. For cases where adjacent slots have interchanged colors, the system outputs a misordered slot pair; for cases where a slot lacks a defined core area, the system outputs a missing line or empty slot anomaly.

[0046] In one embodiment of wire stripping anomaly classification, the system categorizes wire stripping anomalies into excessively long strips, excessively short strips, excessively long exposed cores, excessively short exposed cores, residual insulation, skewed end faces, loose strands, and conductor end anomalies. These anomalies can occur individually or in combination with wiring sequence anomalies to form a composite anomaly.

[0047] In one human-machine collaborative implementation, when a slot, after review, still falls below the manually confirmed threshold, the system highlights the slot on the interface and displays the standard color, measured color, stripped wire measurement value, and the reason for the anomaly. After operator confirmation, the system writes the manual confirmation result into the traceability record, which can also serve as a data source for subsequent model retraining or threshold optimization.

[0048] In one communication output embodiment, the test results can be output to a host computer or production management system as fielded data. Fields include product model, image number, slot number, standard wire sequence, measured wire sequence, stripped wire length, exposed core length, anomaly type, verification mark, test time, and overall part judgment result.

[0049] In one exemplary test result, the standard color of slot 1 is red, the measured color is red, the exposed core length is 5.1mm, and the quality status is OK; the standard color of slot 2 is yellow, the measured color is blue, the exposed core length is 5.0mm, and the quality status is abnormal thread sequence; the standard color of slot 3 is blue, the measured color is yellow, the exposed core length is 5.7mm, and the quality status is abnormal thread sequence and excessive stripping.

[0050] In one process adaptation embodiment, if different products have different focuses on the stripping quality, the recipe switch can be used to select whether to enable residual area detection, end face skew detection, or loose thread detection. Unenabled quality items do not participate in the OK / NG determination for that slot, but can still be output as optional information.

[0051] In one stability implementation, the system can statistically analyze the mean and standard deviation of the stripping length for each slot in the same batch of products. When the measurement result for a certain slot deviates from the batch statistical distribution but does not exceed the individual piece tolerance, the system can output a process warning flag so that the production line can adjust the stripping equipment parameters in advance.

[0052] The color recognition model, boundary localization method, and verification model described in the above embodiments can all be replaced according to the actual product. The quality inspection model can adopt rule-based algorithms, traditional machine learning models, object detection models, segmentation models, key point models, or multi-task visual recognition models.

[0053] In one threshold maintenance embodiment, the upper limit of stripping length, the lower limit of stripping length, the tolerance of exposed core length, the residual sheath area threshold, and the verification threshold can all be maintained independently by the formulation. Different product models can reuse the same testing process simply by switching the formulation, without the need to rewrite the testing program.

[0054] In one model training embodiment, the line sequence recognition part can use slot images and color labels to train a color classification model, while the line stripping quality part can use local slot images and boundary points, length values, or anomaly category labels to train a segmentation model, key point model, or regression model. The two types of models can be trained separately or combined into a multi-task model.

[0055] In one fault-proofing embodiment, when the detection result shows that the wire sequence is abnormal but the wire stripping quality is qualified, the system outputs NG for the wire sequence; when the wire sequence is correct but the wire stripping length or residual sheath condition is abnormal, the system outputs NG for wire stripping; when both modules are abnormal, the system outputs NG for both abnormalities and records the reasons for both abnormalities at the same time.

[0056] The scope of protection of this invention is not limited to the above-described embodiments. Any equivalent substitutions and improvements made to the slot establishment method, line sequence identification method, stripping boundary positioning method, verification method, or quality fusion rules under the concept of this invention shall fall within the scope of protection of this invention.

Claims

1. A method for joint detection of wire harness sequence and stripping quality based on slot mapping, characterized in that, Includes the following steps: S1. Acquire the original image containing the end of the wire harness to be tested, the core arrangement area, or the connector lead-out area, and read the slot formula corresponding to the current product model; S2. Determine the detection area according to the slot formula, and perform image correction, brightness enhancement, scale normalization, or pixel-to-millimeter calibration on the detection area to obtain the image to be tested; S3. Establish a slot coordinate system based on the slot formula, product reference edge, connector reference point, wire harness outer contour, or manual teaching point, and generate the 1st to Nth core slots; S4. Identify the core area in the image to be tested, and determine the positional relationship, overlap relationship, or matching cost between the core area and the slot. S5. Map each core area to its corresponding slot; S6. Extract core color features or color category probabilities in each slot, and compare the measured colors with the standard wire sequence template slot by slot to obtain the wire sequence detection results; S7. Locate the insulation layer end face, exposed conductor area, conductor end, or residual sheath area in each slot, and calculate the exposed core length, insulation retention length, residual sheath area, end face skew angle, or loose wire state to obtain the wire stripping quality detection results; S8. Merge the wire sequence detection results and wire stripping quality detection results to output the slot-level quality status, abnormal slot number, abnormality type, measured value, confidence level, verification mark, and overall OK / NG result.

2. The method according to claim 1, characterized in that, The slot formula includes one or more of the following: detection area location, number of slots N, relative position of slots, slot width, center point of slot, standard line sequence, color judgment threshold, standard value of stripping length, tolerance of exposed core length, residual skin area threshold, end face skew threshold, verification threshold, and output field configuration.

3. The method according to claim 1, characterized in that, The slot coordinate system is established using at least one of the following: product formula, calibration point, connector reference edge, wire harness outer contour, wire core centerline, or manual teaching point; when the image to be tested undergoes translation, scaling, rotation, or slight perspective change, the formula slot is mapped to the current image coordinates according to geometric transformation.

4. The method according to claim 1, characterized in that, The core region mapping includes at least one of nearest neighbor matching, overlapping area matching, centerline projection matching, Hungarian matching, or minimum cost function matching; wherein, the cost function can be expressed as Cij=λ1·dij+λ2·(1-IoUij)+λ3·(1-qj), where dij is the normalized distance between the j-th core region and the i-th slot, IoUij is the overlap ratio, qj is the core region confidence level, and λ1, λ2, and λ3 are weight coefficients.

5. The method according to claim 1, characterized in that, The line sequence detection includes one or more of the following: color mean, color histogram, color moments, texture features, or color category probabilities output by a machine learning model in RGB, HSV, and LAB color spaces; the color recognition model can be implemented using rule thresholding, template matching, traditional classifiers, object detection models, segmentation models, key point models, or multi-task visual recognition models.

6. The method according to claim 1, characterized in that, The stripping quality inspection includes: locating the insulation layer end face Pi, conductor end Pj, or residual sheath area Rk, and calculating the exposed core length according to Lc=k·||Pj-Pi||+b, where k is the pixel-to-millimeter conversion coefficient and b is the calibration compensation amount; when the exposed core length, insulation retention length, residual sheath area, end face skew angle, or loose wire condition exceeds the corresponding tolerance, it is determined that there is a stripping abnormality in the slot.

7. The method according to claim 1, characterized in that, The fusion process involves weighting the color consistency score, stripping dimension qualification score, boundary positioning confidence, and slot matching confidence of each slot to obtain a slot quality score Qi = w1·Si + w2·Li + w3·Bi + w4·Mi. When Qi is below the threshold or any key quality item is abnormal, the corresponding slot is output as NG.

8. The method according to claim 1, characterized in that, When the confidence scores of slot color recognition, stripping boundary positioning, slot matching, or fusion quality score are lower than a preset threshold, a review process is performed on the corresponding slot. The review process includes at least one of the following: local cropping, image enhancement, secondary recognition, boundary refitting, manual confirmation prompts, or review model judgment.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method for joint detection of wire harness sequence and stripping quality based on slot mapping as described in any one of claims 1 to 8.

10. A combined detection system for wire harness sequence and stripping quality, characterized in that, include: The image acquisition module is used to acquire images of the wire harness to be inspected. The slot mapping module is used to establish a slot coordinate system and map the core area to the corresponding slot; The wire sequence identification module is used to identify the color of the wire core in each slot and compare it with the standard wire sequence template; the wire stripping quality inspection module is used to measure the relevant dimensions and abnormal conditions of the wire stripping. The joint judgment module is used to integrate the line sequence detection results and the stripping quality detection results and output the slot-level and overall component quality results.

Images