PCB Solder Defect Detection Based On Machine Vision

This paper studies the soldering defects on the PCB.Mainly from three aspects of image segmentation,contour extraction,target recognition to study the solder defects on the PCB board.Firstly,for the positioning of the image detection area.The contour shape attribute of the detection area is used,so the improved traditional Hough circle transform is used to detect the image circle for localization,and ROI detection area is generated based on the information of the detected circle Secondly,the object in the detection area is segmented.The threshold segmentation algorithm of Otsu statistical recursion relation is used,that is,the improved traditional multi-threshold segmentation algorithm for image segmentation,which can greatly reduce the time complexity of traditional multi-threshold Otsu algorithm,and thus improve the segmentation efficiency.Then the contour of the object is extracted,that is,the red and blue objects in this paper.Gradient algorithm,Laplace of Gaussian algorithm and Canny operator were used to extract the target.According to the extraction results and efficiency,the Canny operator is suitable for contour extraction in this paper.Secondly,the Contour Area operator of Opencv is used to calculate the area extracted into the contour respectively and obtain the data set of the area.Finally,based on the use of Support Vector Machine to identify the product,determine whether the solder on the PCB board is qualified.In the construction of the classifier model of different kernel functions,the optimal kernel function(The Gaussian kernel is chosen in this paper)and the parameters of the kernel function and penalty factors are selected by comparing the classification accuracy.Finally,the test samples were taken into the classification model and the classification accuracy reached 95.54%,so as to meet the production requirements of the actual factory.Therefore,the above methods can reduce the problems of missed judgment and misjudgment caused by manual detection and improve the production quality of PCB.