Printed circuit board (PCB) is the crucial component of electronic product, which provides electrical connection, the excellent quality of PCB is definitely prerequisite for stability of electronic product, therefore, PCB defect inspection is necessary to manufacture, but conventional method cannot meet the demands such as high speed, accurate estimation and low undetectability, hence automatic optical inspection (AOI) has been mainstream, which has higher speed, accuracy and less cost.The PCB automatic optical inspection has been discussed in this paper. Based on theory of image processing and machine vision, the system scheme of defect inspection is established, and algorithms required in scheme are investigated, and software structure is designed and implemented in Visual C++.Defect inspection system is composed of illumination system, image acquisition device and image processing system which includes image process and defect inspection. Defect detail of PCB will clearly appear by using a sets of algorithms such as image preprocessing, registration, segmentation and comparison, then defects also can be accurately located. After analyzing the characteristic of multiple defects and three different design concept of inspection system, this paper estimates corresponding schemes against specific defects.The results of experiment show the schemes , which is contactless and meet the demand of accuracy conventional method cannot, can detect and locate typical defect, and dimensionally measure defect involved in size. |