PCB Apparent Defect Detection Based On Machine Vision Technology

As the development of the Printed Circuit Board(PCB, Printed Circuit Board)manufacture industry, PCB defect detection changes to be more and more complex, the commonly used artificial detection method can neither match the modern PCB processing speed, nor meet the requirement of quality control in the process of PCB production. Therefore, Optical detection based on image recognition AOI(Automatic Optical Inspection) technology is widely applied in all kinds of PCB defect detection.Based on the actual project about the PCB batch producted by Ma'anshan Senge Electronics Co. Ltd., in this paper designed a PCB surface defect detection system,which can test the PCB defect accurately and efficiently.Firstly, this paper investigated the research and application of PCB defect detection system currently, based on this, selected the camera, camera lens and light source, and set up the hardware platform of the system; In view of the PCB image noise, it filtered out noise and retained the image information effectively, through the local median filtering algorithm, and get the target area effectively through the image enhancement and threshold segmentation for the image.Then, recognition and detection was carried out on the PCB, puts forward a improved algorithm that combined gradient Hough transform and contour matching together for the Mark point detection about PCB, it improved the accuracy of the Mark point detection; Through regional growth filling algorithm, extracted the defect of the PCB routing area effectively; Based on HALCON character training and recognition algorithm was designed, the algorithm identified and returned to the character information effectively; Through the combination of Hu moment and image pyramid,improved the accuracy of the template matching process.Finally, the software platform of defect detection system is developed, platform applied some programming ways comprehensively such as split view, multithreading,singleton pattern and so on, combined with the algorithms fine, and the software platform and the algorithm was tested, verified the reliability of the system, and laid the foundation of subsequent further extensions about the function of the algorithm and defect inspection of the scene environment.