Abstract: The automatic PCB optical defect detection algorithm based on mathematical morphology was promoted.The edge is extracted from referential image after erode operation.If the edge is used as the ROI for the distance transform of test image,each point on the edge will have the corresponding distance to the border of the track.Threshold of the distance image can rapidly detect defects of PCBs.Combined with the comparison of contour feature,the algorithm can accurately identify the types of defect.Experiments show that the algorithm can quickly detect all kinds of defects in PCB images,and can make accurate automatic classification and recognition.

Key words: Printed circuit board inspection,Morphology image processing,Distance transform,Defect detection