Research On PCB Surface Assembly Defect Detection Method Based On Machine Vision

With the rapid development of machine vision and artificial intelligence,automated optical inspection(AOI)technology based on machine vision is widely used in PCB surface defect detection by printed circuit board(PCB)manufacturers.However,the current AOI system has problems such as low ease of use and intelligence,and lack of detection of through-hole components.In response to the above problems,the theoretical methods and key technologies involved in PCB image acquisition and PCB surface defect detection are researched in this paper.The specific research contents are as follows:(1)Design the framework of PCB surface assembly defect detection system.Through the analysis of PCB surface assembly defects and actual production requirements,the system's inspection tasks and various performance indicators are clarified.On this basis,the overall framework of PCB surface assembly defect detection system based on machine vision is designed,and the image acquisition method and defect detection method of the system are taken as the research emphases,and they are analyzed and proposed solution strategies.(2)The method of full coverage image acquisition based on path planning is studied.Aiming at the fixed field of view and much smaller than the area of PCB in single image acquisition of industrial cameras,a clustering algorithm of image acquisition windows based on coverage area is designed to obtain the minimum set of image acquisition windows that can cover all electronic components of PCB completely.A variable neighborhood ant colony algorithm is designed for the floating position of image acquisition windows,which utilizes the proposed variable neighborhood path search and variable neighborhood search.The strategy of window position adjustment improves the ant colony algorithm to obtain the shortest image acquisition path.The PCB image capture model is constructed by using the least window position data and the shortest image acquisition path data.The system can quickly complete the image acquisition task according to this model.(3)The method for detecting PCB surface assembly defects based on appearance inspection model is studied.Aiming at the poor usability and low intelligence of traditional defect detection methods and the multi-feature and multi-category characteristics of PCB surface assembly defect detection.An assembly category defect detection algorithm based on improved MobileNet-SSD is designed,which can automatically extract the appearance features of electronic components by deep learning methods,and the pose characteristics of the electronic components are extracted by combining the assembly pose defect detection algorithm based on image processing.The PCB appearance detection model is constructed based on the features extracted by the two algorithms.The detection of surface assembly defect is realized by comparing the qualified appearance detection model with the inspected appearance detection model.(4)Development of PCB surface assembly defect detection system based on machine vision.On the basis of the above research results and on the premise of ensuring the economy of the system,an efficient PCB surface assembly defect detection system is designed and developed,which can be used to detect hybrid-installed PCB.Taking a specific PCB as an example,the practical application of the system in the defect detection process is introduced,and the feasibility of the above scheme is verified.