High Precision's Components Automatic Location Algorithm For SMT

The automatic positioning of PCB components is an important applied research task in the field of electronic design automation.The purpose of the automatic positioning of PCB components is to realize that each component can be accurately mounted in the process of the machine mounting components,so as to reduce the intermediate process of manual inspection of the patch position.Today,the development of artificial intelli-gence,have greatly promoted the development in the field of industrial automation,auto-matic SMT SMT technology already quite mature at this stage of development,the United States in 2018 ”digital erp”,emphasis on U.S.manufacturing transfer the application of artificial intelligence to electronic smart manufacturing industry,intelligent direction for industry in China also gradually improve,But the main core technology or rely on Europe and the United States and other technology developed countries,so the research of this paper is mainly about the application of computer vision in the direction of industrial au-tomation,improve the competitiveness of independent research and development,reduce manufacturing costs.The main research of this paper is the study of automatic positioning algorithm of components in PCB images.The algorithm mainly adopts two methods: automatic po-sitioning of components based on traditional digital image processing and automatic po-sitioning of components based on graph convolutional neural network algorithm.The automatic positioning algorithm of PCB components based on digital image processing includes image acquisition and imaging system,image preprocessing,matching of com-ponents and single reflection transformation to realize the automatic positioning of PCB components.Need of PCB image in the process of automatic positioning bias Angle cor-rection,because pixel match with physical coordinate points in the process of matching,because between components is dense,for the Angle between the point and point,length is strict,cases of error can cause pixel coordinates points matching with physical coordi-nates error phenomenon,Therefore,it is necessary to correct the bias Angle of PCB image to improve the initial positioning accuracy of PCB image components.The automatic location algorithm of components based on graph convolutional neu-ral network includes the steps of constructing convolutional neural network to extract features,constructing graph structure,node information aggregation and deconvolution output results to realize the automatic location of PCB components.Due to the large difference in the size of components in PCB images,the method of segmentation will af-fect the training effect of the network model in the process of image data segmentation.Therefore,the experimental results of different segmentation methods are compared in the experimental process to select the data set of the best segmentation method.The experiment of the localization algorithm also shortcomings,in the hope that in the subsequent research can further improve the accuracy and precision of the positioning,automatic positioning algorithm can push PCB components used in actual production,in-dependent research and development to realize automatic SMT placement machine,catch up with foreign components automatic positioning technology,improve the core compet-itiveness,the development of domestic industrial automation a hand.