A MODIFIED CORRELATION COEFFICIENT BASED PATTERN MATCHING METHOD FOR DEFECT DETECTION

摘要

In a machine vision inspection system, the most popular algorithm for pattern matching is the traditional 2D correlation coefficient method. Although the correlation coefficient method can easily detect defects in complex images, there are still some problems to overcome. In this paper, we proposed a novel modified correlation coefficient method. Instead of applying the method to the entire image, a v u×inspection window is used. The correlation coefficient is calculated through this window pixel by pixel. Next, in order to determine the threshold value, the most important pixel is chosen to compute the correlation coefficient. In a test using numerous sample images, the proposed method showed 100% accuracy in detecting the defects.