A high speed image preprocessing method for IC wafer inspection

A lot of vital defects appear in the process of fabrication of IC wafer. Material defect which mainly causes circuits to be connection failure is an important factor of reducing the yield of IC. So the inspection of material defect is one of the most important things to do for guaranteeing the quality of IC. In this paper, a machine visual test system is designed with AOI (automated optical inspection) technology. The surface defects and geometric sizes of the IC wafer can be detected automatically. It introduces the method of image convolution of FPGA, by using assembly pipelining mode to design a real-time median filtering, to apply to the IC chip testing system which has good effect. Defect feature extraction and selection techniques were studied regarding the characteristics of inspection images of IC wafer. This machine vision test system is realized by FPGA based on XILINX Spartan-3A which is a high-speed parallel device.