关键词: 机器视觉, 边缘掩膜, Canny边缘检测算法, 形态学去噪

Abstract: To deal with the problem of decreased quality of the captured image in machine vision based printed label defect detection system, which is caused by non-synchronization of image acquisition and by transmitting platform shaking, a simple and feasible label printing detection system based on machine vision is proposed. Through the comparison of several edge detection algorithms, a relative high reliability of the edge detection algorithm is used to make edge mask, and then the edge mask is covered on the template image. After image subtracting, morphological filtering is performed, and the shadow of contour and tiny defects, which is difficult to identify, is dispelled. This method has been applied to the label printing detection system, and can effectively lower the false detection rate and meet the human visual characteristics.

Key words: machine vision, edge mask, Canny edge detection, morphology filtering