Self-Adaptive Threshold Canny Operator in Color Image Edge Detection

Abstract

It needs to set the two thresholds of classical Canny Operator manually, which confined this algorithm. A lot of research on adaptive thresholds choice method has been done to conquer this shortcoming. Based on the gradient histogram, which is studied constantly in the past, a method of gradient histogram difference diagram with adaptive image classification techniques is proposed in this paper. It automatically sets the two thresholds, and avoids disconnected or false edges in detection. Experiments prove that the method is threshold-adaptive and the edge detection performs well in color image whose larger gradient amplitude pixels are mainly located in the edges between the target and background.