Assembled PCB visual inspection machine using image processor with DSP

To automate the visual inspection of assembled printed circuit boards (PCBs), a visual inspection machine was developed. The machine detects a variety of defects of assembled PCBs, such as missing parts, polarity errors, improper names of ICs, and improper insertion of leads. To achieve this variety of functions, a microprogrammable image signal processor, the DSP-i, has been introduced into the machine. A character recognition algorithm called the multistep similarity vector comparison method is used in the processor. This method enables the inspection machine to recognize IC names reliably without the learning of various patterns of characters from many manufacturers. The operation speed of the DSP-i is about 20 times faster than that of a general 16-b microcomputer