| Abstract: | An image, in-process inspection and accuracy positioning system has been developed as automated measurement machine to deal with ball grid array type printed circuit boards (PCB). A new technique is proposed to automate the boundary scan and circularity estimation. The attributes of a pixel at a pad or hole are calculated from the RGB color on PCB image. Here we introduce the idea of the least-square circle to find the center of the holes and pads. The value of alignment shift can be obtained according to the four images in a special image capturing procedures. If the circle of the pad and hole can be determined correctly, the vision system can easily inspect the local deviation when positioning table locates the PCB on the image capturing point. The results of positioning and inspection are presented and evaluated including through hole locations, alignment calculation, system calibration, and statistical functions. |
