Defect Detection Of A Low-voltage Circuit Breaker Complex Texture Panel Based On Machine Vision

In terms of industrial measurement technology,machine vision inspection technology is the future of modern industrial production.However,its application in product appearance inspection is relatively backward,especially for the detection of complex texture panel defects.Manual detection is the traditional way,whose shortcomings include high detection cost,poor accuracy,and low efficiency.With these,it fails to contribute to the production and development of enterprises.Given this,this paper will detect and determine the defects of a complex textured panel of a low-voltage short circuit based on machine vision technology,so as to reduce production costs and improve enterprises’ production efficiency.The main work and research results of this paper are as follows:(1)Construction of image acquisition system and preprocessing.Firstly,the research object of this subject is analyzed.Based on the actual needs of this project,the laboratory image acquisition system is rebuilt,which mainly includes the selection design of industrial cameras and lenses,the selection design of lighting equipment,and the construction of the overall bench.According to project requirements,an image dataset of complex textured panels is collected and constructed.For the different positions in the process of panel image collection,an accurate correction processing algorithm with rectangular frame images is designed.Through experimental comparison,the accuracy of the algorithm is verified,which provides a reliable and effective guarantee for the subsequent algorithm processing and corresponding defect detection.(2)Texture parameter learning and self-defect detection algorithm design.Analyze the textures in the panel,design a texture parameter learning algorithm,and perform it on textures’ distance,length and width,grayscale range,and arrangement order.With the panel texture feature parameter learning algorithm,varieties of inherent defects of the panel texture including misprinting,missing printing,ghosting,skewing,and sticking,can be accurately classified and located.(3)Design of the detection algorithm for defects caused outside the panel.In terms of external defects such as oil stains,scratches,and film folds,different algorithms are designed for classification and identification: for the oil stains on the panel surface,this paper designs a threshold segmentation algorithm based on V-channel filtering;for the scratches on the panel surface,this paper designs a high degree of reduction detection for scratches and a highaccuracy measurement algorithm;for the film folds on the panel surface,this paper designs a threshold segmentation algorithm based on gray histogram.(4)System software design and development.According to the actual needs of enterprises,the system software is developed based on the accuracy of defect detection and the efficiency of vision algorithms.The system mainly includes data communication module,image acquisition module,image processing module,and data storage module.Online experimental tests proved that the accuracy rate of defect detection on the panel surface reached 99.24%,which met the project’s demand for defect detection accuracy.