Defect Detection And Implmentation Of Printed Circuit Board Based On Digital Image Processing

In modern electronic products,the quality of printed circuit boards directly determines the performance of the final product,with the development of technology,the printed circuit board is more and more complex,the traditional artificial visual defect detection,because of slow speed,low efficiency,it is difficult to meet the production needs of large scale testing,so the study of a fast and efficient defect detection system is very necessary.The frame structure of automatic defect detection system is designed,including lighting,image acquisition,motion control and image processing software.Meanwhile,the process of image processing in the system is introduced.Histogram modification,gray transformation,adaptive median filter and iterative method are used to preprocess the PCB image to increase the contrast of the image.Secondly,an adaptive median filtering method is proposed to smooth the image.This method can change the size of the window and remove the impatient at the same time.The method can protect the edge details of the image.Finally,the method is used to select the appropriate threshold to transform the gray image into a binary image.This method reduces the image data and makes the features of the image more obvious.The improved SIFT algorithm is used to realize the location and registration of the measured image and the reference image.The classical SIFT algorithm is used to locate and match the measured image and the reference image.The improved SIFT algorithm is adopted and the other redundant information is deleted according to the key points of the SIFT algorithm.The description of the characters is changed from the rectangle feature to the circular feature,which improves the selection of characteristics,achieves the precise matching,simplifies the steps and speeds up the operation.This paper uses tree stratification to classify the characteristics,in order to detect the short circuit,circuit breakage,convex,hollow and hollow areas of the circuit board.The defects were identified by the feature information of the number of connected domains,the area of the wire and the gray level difference of the envelope of the defect.