Research Of PCB Solder Joint Defect Detection Based On Machine Vision

With the rapid development of the electronics industry,electronic products have become very popular,and the quality of printed circuit board(PCB)solder joints directly affects the quality of electronic products.In order to improve the pass rate of electronic products,it is also very important to perform defect detection on the solder joints of PCB boards.In order to improve the detection accuracy and detection efficiency,the use of machine vision to achieve automatic detection has important theoretical research and practical application value.This paper is based on the research of solder joint defect detection in machine vision.The types of solder joints tested include multi-solder joints,less solder joints,suitable solder joints and leak solder joints.By constructing the machine vision system,the overall scheme of solder joint defect detection is designed.The image processing is used to study the multi-exposure image fusion method and the method of detecting and classifying PCB solder joint defects.The main work of the thesis is as follows:(1)The overall design of the machine vision system.Firstly,the evaluation criteria for the type of solder joint are determined,and the overall scheme of the solder joint defect detection system is designed.For the tablet circuit board used in the bank storage of the experimental object,the industrial camera and the optical lens are selected by analyzing the shape and size thereof.Then analyze the characteristics of the four lighting methods and their applicable occasions,choose the appropriate light source,and design a better lighting method.Finally,the motion controller,stepper motor and motor driver in the motion control system are selected to complete the overall design of the PCB solder joint inspection machine vision system.(2)Research on multi-exposure image fusion method.In the process of image acquisition,for the phenomenon of uneven exposure,a multi-exposure image fusion algorithm based on detail retention is proposed to fuse the solder joint image.Compared with the commonly used five image fusion algorithms,the subjective and objective aspects:information entropy and average structural similarity are used to analyze the quality of the image.The results show that the proposed method can better preserve the details of the original image.(3)Preprocessing and feature extraction of the merged image.Image preprocessing includes image denoising,image enhancement,image threshold segmentation,morphological processing,and edge contour extraction.The related pretreatment method is proposed and compared with many common methods,and the treatment effect is better.Then,the shape features,the texture features,and the Histogram of Oriented Gradient(HOG)feature parameters of the solder joint image are extracted for subsequent image detection and classification.(4)Research on the classification method of PCB solder joint defect detection.A multi-feature based support vector machine(SVM)multi-classification algorithm is proposed.Firstly,the extracted shape features and texture features are compared,and the effects of different kernel functions are compared.The optimal kernel function is used for SVM classification.The four types of multi-solder joints,less solder joints,suitable solder joints and leak solder joints are tested and classified.For the mis-inspected solder joints,the SVM multi-classification algorithm based on HOG feature is used to perform secondary detection and classification,and the final classification accuracy is obtained.Compared with the commonly used solder joint defect detection classification algorithm,the experimental results show that the proposed algorithm has higher detection accuracy than other classification methods,and the detection classification effect is better.(5)Design the PCB solder joint defect detection system software.Firstly,the development environment and implementation functions of the system software are explained.Then the overall design of the system software and the operation steps of the system software are described,and finally the block test is performed,mainly for the multi-exposure image fusion,image preprocessing,image segmentation,feature extraction and PCB solder joint defect detection.