CN-122023367-A - Visual inspection method for appearance defects of electronic cable

CN122023367ACN 122023367 ACN122023367 ACN 122023367ACN-122023367-A

Abstract

The invention relates to the technical field of machine vision detection, in particular to a visual detection method for appearance defects of an electronic cable, which comprises the steps of collecting cable images and establishing a three-dimensional coordinate system; carrying out sequential frame decomposition and spectral response analysis to separate image component layers with different physical characteristics, overlapping the component layers in a three-dimensional space to form an overlapped feature body, extracting feature data through layering stripping, importing the data into a dynamic threshold growth model to iterate and generate a self-adaptive defect judging threshold value, identifying a defect voxel by utilizing threshold scanning, tracking a morphological evolution path of the defect voxel to carry out connection and merging, reconstructing a defect topological structure network, mapping the defect topological structure network back to a three-dimensional coordinate system, and outputting a defect category and a defect position. According to the method, through multispectral component separation and self-adaptive topology analysis, the defect characteristic identification capability is enhanced, and the detection accuracy and the environmental adaptability are improved.

Inventors

LU SHIZHEN
WANG CHENGWEI

Assignees

衡阳鸿丰电子科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260205

Claims (10)

1. The visual detection method for the appearance defects of the electronic cable is characterized by comprising the following steps of: acquiring an initial surface image of an electronic cable, and establishing a three-dimensional space coordinate system corresponding to the electronic cable; performing serialized frame decomposition on the initial surface image to generate a frame image set arranged according to time sequence; Spectral response analysis is carried out on each frame image in the frame image set respectively, and image component layers with different physical characteristics are separated; According to the coordinate information of the three-dimensional space coordinate system, spatially overlapping a plurality of image component layers corresponding to each frame of image to form an overlapped feature body of the electronic cable; performing layering stripping treatment on the overlapped feature body, and extracting stripping feature data reflecting the surface state layer by layer; importing the stripping characteristic data into a dynamic threshold growth model, and iteratively generating a defect discrimination threshold for the current electronic cable; Scanning the overlapped feature body by utilizing the defect judging threshold value, and marking a space voxel set which accords with the defect feature; Tracking a morphological evolution path of the spatial voxel set in the superposition feature, and communicating and merging the scattered voxel sets according to the morphological evolution path; Reconstructing a topological structure network of the surface defects of the electronic cable based on the voxel set after the connection and the merging; and mapping the topological structure network back to the three-dimensional space coordinate system, and outputting the category and position information of the surface defects of the electronic cable.
2. The visual inspection method of appearance defects of an electronic cable according to claim 1, wherein the capturing an initial surface image of the electronic cable and establishing a three-dimensional space coordinate system corresponding to the electronic cable specifically comprises: synchronously capturing original image streams of a plurality of sides of the electronic cable through optical sensor arrays arranged at both sides of the electronic cable transmission path; calculating the space geometrical relation between the original image streams according to the physical position and the angle parameters of each sensor in the optical sensor array; According to the space geometrical relationship, uniformly registering each original image stream into a virtual three-dimensional space taking the central axis of the electronic cable as a reference; Defining a coordinate origin, an axial direction and scales in the virtual three-dimensional space, and completing establishment of the three-dimensional space coordinate system; the initial surface image is the initial expression of the multi-side image data after spatial registration in the three-dimensional space coordinate system.
3. The visual inspection method of an electronic cable visual inspection system according to claim 2, wherein the performing a sequential frame decomposition on the initial surface image generates a set of frame images arranged in time sequence, specifically comprising: performing time slicing on the initial surface image according to a preset sampling interval along the transmission direction of the electronic cable; Encoding the image data obtained by each time slice according to the axial position of the image data in the three-dimensional space coordinate system; recombining the coded image data according to the time sequence and the axial position sequence to generate a frame image sequence with strict time-space sequence; For each frame image in the frame image sequence, the acquisition time stamp and the axial position information in the three-dimensional space coordinate system are additionally recorded; The frame image set is a data structure formed by the frame image sequence and the incidental information thereof.
4. A visual inspection method for appearance defects of an electronic cable according to claim 3, wherein the spectral response analysis is performed on each frame image in the set of frame images to separate image component layers with different physical characteristics, specifically comprising: For each frame of image, analyzing the spectrum intensity distribution curve of the pixel point of the image; identifying characteristic absorption peaks and reflection peaks in the spectrum intensity distribution curve according to a preset spectrum characteristic library; classifying the pixel points corresponding to the spectrum curves with similar characteristic peaks into one type to form a plurality of pixel point clusters; Extracting all pixel points belonging to the same pixel point cluster from an original frame image, and reserving the original space position of the pixel points to generate an image component layer; after each frame of image is processed, a plurality of image component layers which are mutually independent and jointly form the complete information of the original image are obtained.
5. The visual inspection method of appearance defects of an electronic cable according to claim 4, wherein the spatially overlapping the plurality of image component layers corresponding to each frame of image according to the coordinate information of the three-dimensional space coordinate system to form an overlapping feature of the electronic cable specifically comprises: reading axial position information attached to each frame of image in the three-dimensional space coordinate system; All the image component layers separated from the same frame of image are placed on the corresponding axial positions in the three-dimensional space coordinate system according to the original coordinates of the pixel points; Stacking all image component layers of all frame images in the three-dimensional space coordinate system according to the corresponding axial position sequence; Carrying out fusion calculation on the pixel point data occupying the same space position after stacking to generate a dense three-dimensional data array; The superposition feature of the electronic cable is the three-dimensional data array, and the superposition feature fully characterizes the comprehensive state of the surface of the electronic cable in three-dimensional space and spectrum dimension.
6. The visual inspection method of an electronic cable visual inspection system according to claim 5, wherein the step of performing a layered stripping process on the superimposed feature, and extracting the stripping feature data reflecting the surface state layer by layer, comprises: defining a series of virtual cut planes in the superimposed feature in a direction parallel to the surface of the electronic cable; carrying out data cutting on the overlapped feature body along each virtual cutting surface in sequence from outside to inside, and obtaining a surface layer slice by each cutting; analyzing the data distribution mode of each surface layer slice, and calculating the indexes of texture continuity, density uniformity and spectrum consistency of the surface layer slices; Binding the texture continuity, density uniformity and spectrum uniformity indexes obtained through calculation with depth information of the surface layer slice to form a characteristic data packet; the stripping characteristic data is a sequence consisting of characteristic data packets of all surface layer slices according to a stripping order.
7. The visual inspection method of appearance defects of an electronic cable according to claim 6, wherein the step of importing the stripping characteristic data into a dynamic threshold growth model to iteratively generate a defect discrimination threshold for a current electronic cable comprises: Taking the stripping characteristic data sequence as input, and sending the stripping characteristic data sequence into a preset initial threshold model; the initial threshold model generates an initial defect discrimination threshold based on the characteristic value of the first characteristic data packet in the stripping characteristic data sequence; Performing defect pre-judgment on the surface layer of the overlapped feature body by using the initial defect judgment threshold value to obtain a preliminarily judged defect area; According to the characteristic feedback of the defect area which is primarily judged, the internal parameters of the initial threshold model are adjusted to output an updated defect judging threshold; and repeatedly using the updated defect judging threshold value to conduct pre-judging and adjusting the model according to feedback until the defect judging threshold value output by the model tends to be stable, wherein the stable threshold value is the final defect judging threshold value aiming at the current electronic cable.
8. The visual inspection method of an electronic cable visual inspection system of claim 7, wherein the scanning the superimposed feature with the defect discrimination threshold identifies a set of spatial voxels in which defective features are met, comprising: Comparing the final defect discrimination threshold with the feature value of each voxel in the overlapped feature body; If the characteristic value of a certain voxel exceeds the final defect judging threshold value, marking the voxel as a candidate defect voxel; In the three-dimensional space of the superposition feature, carrying out space neighborhood analysis on all marked candidate defect voxels; aggregating spatially adjacent candidate defect voxels into a spatial voxel cluster; the set of spatial voxels is made up of all identified spatial voxel clusters, each cluster representing a potential defect region.
9. The visual inspection method of an electronic cable visual inspection system according to claim 8, wherein the tracking the morphological evolution path of the spatial voxel set in the superimposed feature, and the communicating and merging of the scattered voxel set according to the morphological evolution path, specifically comprises: Analyzing the morphological changes of each spatial voxel cluster in the spatial voxel set on different layers of the overlapped feature body along the depth direction; Recording a continuous change track of the shape center, the area and the contour of the space voxel cluster in the depth direction, wherein the track is a shape evolution path; Comparing morphological evolution paths of different spatial voxel clusters, and judging that the spatial voxel clusters are derived from the same physical defect if the paths are crossed, approach or have highly similar morphological characteristics in space; a plurality of space voxel clusters which are judged to originate from the same physical defect are subjected to data communication in a three-dimensional space and are integrated into a unified defect body; and performing analysis, comparison and merging operations on all the spatial voxel clusters to form a plurality of independent and complete defect volume data.
10. The visual inspection method of appearance defects of an electronic cable according to claim 9, wherein the reconstructing a topological structure network of the appearance defects of the electronic cable based on the connected and integrated voxel set specifically comprises: Taking each independent defect body as a topological node; Calculating three-dimensional geometric features of each defect body, including a volume, a surface area, a main shaft direction and a space external frame, and taking the three-dimensional geometric features as attributes of corresponding topological nodes; Analyzing the relevance between the relative position relation of any two topological nodes in the three-dimensional space and the geometric characteristics, and if the preset topological relevance rule is met, establishing a connecting edge between the two nodes; all topological nodes and connecting edges between the topological nodes together form a topological structure network of the surface defect of the electronic cable; and converting the space coordinates of each node in the topological structure network into actual coordinates under the three-dimensional space coordinate system, and outputting the category and position information of the surface defects of the electronic cable by combining the attributes of the actual coordinates.

Description

Visual inspection method for appearance defects of electronic cable Technical Field The invention relates to the technical field of machine vision detection, in particular to a visual detection method for appearance defects of an electronic cable. Background The conventional visual detection method for the appearance defects of the electronic cable is mainly based on visible light imaging, and defect identification is carried out by collecting RGB images and then applying edge detection, threshold segmentation or template matching algorithm. The methods rely on single spectrum information, have fixed processing flow, are sensitive to illumination change, surface texture or color interference, and are easy to cause false detection or omission. Multispectral or hyperspectral imaging is introduced in the prior art to acquire more physical information, but the images are generally directly fused or fixed threshold processing is adopted, so that independent features of component layers with different physical characteristics cannot be effectively separated and utilized, and the distinguishing capability of defects and backgrounds is limited. In the aspect of three-dimensional space analysis, partial technology is used for reconstructing a three-dimensional model of the cable surface through laser scanning or structured light, and defect detection is carried out by combining image processing. However, the defect discrimination depends on a preset threshold value, and the defect discrimination lacks adaptive adjustment for specific cable characteristics, so that the morphological change of complex defects cannot be processed. The connectivity analysis of the defect area by the existing method is mostly based on simple morphological operation or area growth, and the evolution of defects in space-time dimension is difficult to accurately track, so that the dispersed defect areas are combined inaccurately, the topological structure is reconstructed coarsely, and the defect classification and positioning accuracy are affected. Aiming at the problems that the defect detection is limited by single spectrum information, the threshold value is fixed and the defect morphology analysis is insufficient in the prior art, an independent physical characteristic component layer needs to be separated from a multispectral image so as to enhance the defect characteristic expression. Meanwhile, a defect discrimination threshold is required to be adaptively generated based on cable characteristics, accurate topology reconstruction is realized by analyzing a defect morphology evolution path, and the accuracy and the robustness of detection are improved. Disclosure of Invention The invention aims to solve the defects in the prior art, and provides a visual detection method for the appearance defects of an electronic cable. In order to achieve the purpose, the invention adopts the following technical scheme that the visual detection method for the appearance defects of the electronic cable comprises the following steps: acquiring an initial surface image of an electronic cable, and establishing a three-dimensional space coordinate system corresponding to the electronic cable; performing serialized frame decomposition on the initial surface image to generate a frame image set arranged according to time sequence; Spectral response analysis is carried out on each frame image in the frame image set respectively, and image component layers with different physical characteristics are separated; According to the coordinate information of the three-dimensional space coordinate system, spatially overlapping a plurality of image component layers corresponding to each frame of image to form an overlapped feature body of the electronic cable; performing layering stripping treatment on the overlapped feature body, and extracting stripping feature data reflecting the surface state layer by layer; importing the stripping characteristic data into a dynamic threshold growth model, and iteratively generating a defect discrimination threshold for the current electronic cable; Scanning the overlapped feature body by utilizing the defect judging threshold value, and marking a space voxel set which accords with the defect feature; Tracking a morphological evolution path of the spatial voxel set in the superposition feature, and communicating and merging the scattered voxel sets according to the morphological evolution path; Reconstructing a topological structure network of the surface defects of the electronic cable based on the voxel set after the connection and the merging; and mapping the topological structure network back to the three-dimensional space coordinate system, and outputting the category and position information of the surface defects of the electronic cable. As a further aspect of the present invention, the capturing an initial surface image of an electronic cable, and establishing a three-dimensional space coordinate system corresponding to