A rule-based inspection system for printed circuit boards

Techniques are presented for defect detection in printed circuit boards (PCBs) by analyzing their two-dimensional gray level image. A software package has been developed to detect short circuit of tracks, bridging of two or more adjacent solder joints, the presence of all components and the quality of solder joints of the PCB. The recognition of these defects is achieved by comparing the image data with the predefined data which is obtained during the teaching phase. Lighting, image processing techniques, and the great variation in the possible appearances of the PCBs complicated the development of the algorithms. The software has been used for inspection of the inhouse-manufactured PCB as well as the commercial ones. From the experimental results obtained, it is shown that both the accuracy and reliability of the proposed inspection system are satisfactory