Research On Defect Detection Algorithm Of PCB Surface Mount Based On Image Processing

Electronic components on the printed circuit board(PCB)are usually installed by the placement machine and plug-in machine.In the installation process,once there are wrong component,missing component,polarity reversal and other problems,the whole PCB will be scrapped,so it needs to be tested before it is put into use.Compared with traditional manual visual inspection,Automatic Optical Inspection(AOI)technology has high relia-bility,good stability and detection accuracy.With the development of electronic industry,SMT components on PCB are developing towards higher integration,smaller volume and more layers,and manual visual inspection has been unable to meet the actual produc-tion needs.Therefore,the study of AOI technology for PCB surface mount defects based on image processing has important academic significance and application value.In this paper,the algorithms of Mark location,component image matching,circular component detection and polarity detection in PCB defect detection are studied in detail.The main research contents of this paper are as follows:1)A two-stage locating algorithm based on gradient and random circle detection is proposed for locating the Mark points in PCB.The improved Canny edge detection was used to extract the edge contours,and then the edge contours were classified according to the gradient and convexity,and the arcs were screened and clustered by certain constraint rules.Then the improved random circle detection algorithm and the least square method were used to fit the circle.2)In order to meet the requirements of high-precision positioning in AOI system,pre-cise positioning is carried out by subpixel edge detection based on Zernike moments.The traditional subpixel edge detection based on Zernike moments uses the ideal edge model,which does not consider the existence of transition zone in the actual edge.Therefore,this paper designs an improved three-gray edge detection model on this basis.Finally,simulation experiments and public datasets verify that the proposed algorithm has higher fitting accuracy and efficiency than other algorithms,and has better robustness.3)The traditional image matching algorithm can not match the rotation and scaling at the same time,and the matching time is long and the calculation is expensive,so this paper propose an improved ring projection matching algorithm.First we proposed the derivation ring projection characteristics.The similarity calculation and the scaling factor were estimated by the dynamic time warp algorithm.We used pyramid search strategy to accelerate at the same time.After obtaining the coarse matching candidates,fine matching based on pseudo Zernike moments was carried out,and then we estimated the rotation angle by improved ring-shifted technology.Experiments on PCB datasets and other public datasets show that our algorithm can efficiently match and locate targets,and can be used in rotation,scaling,illumination changes,noise interference,complex environment,multi-object matching and so on.4)The specific applications of the above algorithm in PCB defect detection are stud-ied,and the experiments verify that the Mark location algorithm can also be used in the detection of circular components.On this basis,the polarity detection methods of elec-trolytic capacitor and chip are analyzed.Finally,the application of image matching tech-nology in PCB defect detection is analyzed.