PCB Bare Board Defect Image Detection Considering Random Characteristics of Defects
A new method considering random characteristics of defects is proposed to improve PCB bare board defect detection accuracy.The method preprocesses PCB bare board images,including grayscale conversion,denoising,boundary and contour division.It separates defect and non-defect areas through fixed-point processing,extracting random defect features.Support Vector Machine is used for defect classification,while Random Forest regression model locates defects.Tests on randomly selected PCB component positions show that the method effectively detects PCB bare board surface defects,achieving high detection accuracy and low miss rate across multiple randomly selected PCB component positions.With high average detection precision at various stages,it can play a significant practical role in PCB production quality control.
PCB bare boardMedian filteringCanny edge detectionSVMRandom Forest regression
万方数据
维普
