Research Of PCB Inspection System Based On The Computer Vision

Machine vision plays a major role in automatic manufacturing industry.The technique has characteristics of high speed, precise, reliable and untouched.Advancement of production technology,computer techniques, image processing technique and artificial intelligence accelerate the spread and application of computer vision inspection technique.Using machine vision to detect PCB defect is one of most important applications. Unfortunately, most pieces of vision equipment are abroad and expensive, which is the major motivation of this research.This study develops a PCB inspection system to recognize PCB defects. Major research as follows:(1) This thesis analyses the state of the PCB examination technology in the world and the working principle of PCB Vision Inspection System. Analyses all defects and brings forward the defects inspection method which uses the combination of Reference Based Inspection Method and Design Rule Based Inspection Method.This method exerts the virtue of above two methods sufficiently, and remedies the disadvantage of ench.(2) Analysed and validated the traditional image segmentation method,this paper adopts improved OTSU method to segment PCB image.To the registration of criterial PCB image and inspected PCB image,this paper adopts Hough transformation to extract the centers of four marks on PCB,and uses subpixel edge detection method to advance the inspection precision for the centers.Uses morphology method to solve the shadow problem in the Reference Based Method.(3) The thesis recognizes short circuit,open circuit and etc defects with scanning the state of defects boundary,and earmarks their positions on PCB,which contribute to the following repair work.The PCB Defects Inspection System has been designed based on Labview8.2,and its inspection ability has been validated through experiment,several images have been tested using this system.The results show that it can recognize the faults effectively mentioned in the paper, which proved the methods used in the system is succeed and thus have practical value.This software can detect short,open,spur,pin hole,spurious copper,breakout,wrong size hole,missing hole with Reference Based Method and enmarks their positions.It also can detect the conductor line width and the distance between conductor lines with Design Rule Based Inspection Method.