Research And System Implementation Of PCB Surface Defect Inspection Based On Machine Vision

The manufacturing industry is the main part of the real economy and the key to national industrialization,while the printed circuit board is a hardware carrier of electronic and information technology.With the continuous advancement of innovation upgrading and industry upgrading in China,PCB automated quality inspection technology has become a vital part of economic production.At present,the method of PCB surface inspection is still based on manual visual inspection,with slow detection speed,poor stability,and high subjective factors.Automated Optical Inspection(AOI),as a technology widely used in various industrial environments,can complete inspection tasks in a non-contact manner.This technology takes machine vision technology as the core and can effectively deal with the problem of automatic inspection of PCB surface defects.In this thesis,the single-layer single-sided bare PCB is the main detection object.Its surface is copper-plated or nickel-plated,and its optical imaging characteristics are complex.Compared with the conventional PCB board,the size of the PCB board to be tested in this system is larger,the maximum is 550 mm ?600mm,and the amount of data processed by the system is large,so it is necessary to improve the real-time performance of the system under the premise of ensuring the detection effect.In consideration of accuracy,this thesis finally completed the selection and construction of hardware equipment,including LED light source,high-resolution industrial cameras,Mechanical motion device,etc.Then,for the four defects of short circuit,open circuit,bump,and depression,this thesis focuses on designing and implementing a software system,which takes image processing as the core.The core content of this thesis is to study the image preprocessing,defect detection and recognition methods in the software part.Before defect identification,it needs to do image preprocessing.First,for the problem of blurred contours on the edges of the densely distributed lines on the PCB,it is necessary to enhance the image.After comparing image enhancement methods such as histogram equalization,Sobel gradient sharpening,and Laplace sharpening,this thesis uses Laplace sharpening.The method effectively improves the imaging effect at the edge of the line.Then,in the overall framework of the reference method,the sample image and the standard image are registered.In consideration of accuracy,matching rate,structural similarity and other indicators,SURF is selected as the final registration method,and a fast calculation method based on the scale invariance for registration is proposed.After preprocessing,through the OTSU threshold segmentation method and the image XOR,the difference image between the sample image and the standard image is obtained,and then the small defects are removed by morphological filtering.Finally,the improved defect outer edge scanning method is used to realize defect identification,and based on the change point of the defect state,a quantitative evaluation method of protrusions and depressions is proposed.This thesis describes the key modules of the system one by one,including the overall structure of the system,the principle and process of the detection method,and the man-machine interface.Test Results show that under the premise of meeting the detection accuracy of 0.1mm,the system's missed detection rate is less than 3%,the false detection rate is less than 5%,and the detection time is less than 10 s,which can meet the demand of actual production.