In the production line of printed circuit boards (PCB), quality control requires constant counting and recording of the number of boards which, in some special cases, is done by workers in the factory. This form of counting consumes labor and time, not to mention the high error rate. In this paper, an automatic method to count PCBs is proposed. With the use of image processing techniques, the number of PCBs can be non-destructively calculated as they pass through the production line, which will effectively improve the production efficiency. The first step of the proposed system is to take a picture of a stack of PCBs using a digital camera. The captured image is transformed to a gray level picture and processed by a noise removal algorithm to improve the proposed mechanism's precision. The output of the counting algorithm includes PCB count and confidence level. In practical applications, the confidence level can help people to on-line diagnose erroneous counting and correct it when the visualization counting system fails to accurately differentiate the boards.
淡江理工學刊=Tamkang Journal of Science and Engineering 12(4)，頁471-479