Research On Key Technologies Of AOI System For PCB Inspection

In the modern electronics industry, PCB(Printed Circuit Board) is an important part of kinds of electronic products. The quality of PCB is an important factor of determining product quality. Along with the development of manufacturing industry, PCB is developing in the direction of ultra-thin, small components, high-density, fine pitch, which is a great challenge to test the quality of PCB. Therefore, studying the AOI(Automatic Optic Inspection) system has realistic meaning and practical value in testing the quality of PCB.In this thesis, by taking use of photoelectric technology, automatic control technology and computer technology, an AOI system of testing the quality of PCB is designed. This system consists of lighting system, image collection system, motion control system and image processing system. And this thesis has a further study with the image processing system, in which PCB image pre-processing, PCB image mosiac, PCB image aligning and the identification of PCB defects are studied mainly, on this basis a software system is developed for testing the quality of PCB.Firstly, during the image pre-processing color image gray processing, image denoising and image segmentation are studied. In order to acquire good results of image pre-processing, an algorithm of PCB image pre-processing is proposed, which is taking use of the weighted average algorithm to change color images into gray images, selecting the median filtering algorithm as the algorithm of image denoising, and completing the image segmentation by using the Otsu algorithm.Secondly, in order to realize the detection of large PCB, a mosiac algorithm based on features is proposed to achieve PCB image mosiac. On the ground of the same mosiac precision, the mosiac speed of the mosiac algorithm based on SIFT features is slow, so in this thesis the mosiac algorithm based on SURF features is selected to mosiacing, which has quick mosiac speed. In addition, the SURF match based on hardware is improved to accelerate the mosiac speed.Then, in order to align accurately with the test PCB image and the standard PCB image, the located circular hole is selected as the located mark. For the circle detection of the located circular hole, three algorithms of circle detection are realized, and experimental analysis with the different algorithms of circle detection are did. Lastly the algorithm of circle detection based on improved Hough transform is selected to improve the speed and precision of the circle detection.Finally, according to image contrast technology and mathematical morphology operations, image which contain defects is acquired, and the detection of the connected domain based on segment is applied to realize the connected domain detection of defects. On this basis, in order to improve the speed of defects identification, a recognition algorithm of tree-step hierarchical judgement is used to identify short circuit, open circuit, raised circuit, sunken circuit and hollow circuit according to the different features of different defects.The results of experiments show that AOI system is reasonable and available in this thesis, and the defecting results of PCB quality are valid and credible, so it provides some references to the research on AOI system in the future.