| Summary: | 碩士 === 輔仁大學 === 電子工程學系 === 95 === To develop a visual inspection system using inexpensive CMOS Sensor is the purpose of this study. It is used to replace the vision inspection process in the mass production of electronic product on the detection of feature point. The detection method used here is image subtraction. But we found “noise” would make a critical influence on this method. For reducing this effect, we propose a statistical method to solve this problem. We use 600 pieces of normal images acquired to build the “background image”, then calculate the mean and standard deviation of every pixel’s R, G, B’s values on it for qualification. Experimental results show that 92% detection rated can be achieved through appropriate selection of standard deviation.
|
|---|
