Research On Machine Vision-based Defect Detection Method For Flexible Circuit Boards

Flexible printed circuit boards(FPC)are widely used in smartphones and LCD TVs,with the mass use of modern society's electronic products,the demand for FPC circuit boards will greatly increase at the same time will also be more and more inclined to thin and light,then the degree of precision and manufacturing complexity of FPC will gradually increase,at this time for the quality control of FPC circuit boards will be more stringent.With the advent of the Internet of Things and the rapid development of machine vision,the manufacturing industry will also win a huge innovation,and the implementation of a machine vision-based FPC board defect detection system has great significance and value.This topic researches a machine vision based FPC defect detection method,the main work content is as follows.(1)Research on the pre-processing algorithm of FPC images.Image pre-processing is an extremely important part of the FPC inspection system,and if it is handled properly,the efficiency of the subsequent inspection will be greatly improved.By means of image enhancement,image filtering and image segmentation,noise interference is eliminated as far as possible and the contrast of the FPC image is improved.Based on the characteristics of the FPC,the filtering algorithm is improved on the basis of the references,and the Otsu algorithm is applied to image segmentation,and the experimental comparison analysis shows that the pre-processing method is effective and reliable.(2)For the actual inspection process,the placement of samples can not be completely uniform phenomenon,need to be inspected sample target positioning and image alignment,this topic adopts the template matching method for target positioning,after experimental analysis and comparison,the target positioning method is based on the feature point template matching method.After achieving the target location,image alignment can be carried out.This topic adopts the feature-based alignment method to carry out image alignment,and analyzes and compares two methods: SIFT algorithm and SURF algorithm.(3)The experimental platform is introduced and the image morphology is applied to eliminate the spurious points that may still exist after pre-processing.After that,for the FPC board appearance defects in the foreign colour,foreign matter and pressure/scratch/scratch defects,according to their defect characteristics respectively design detection algorithms,and finally achieve the automatic recognition of FPC board defects classification.Finally,on the basis of the above,the average over-inspection rate of 5.66% and the average miss-inspection rate of 8.74% were calculated for 10 groups of FPC boards containing defects,which are lower than the values of manual visual inspection in enterprises,making the inspection method feasible and applicable in practice.