An Improved Canny Edge Detection Algorithm

Tao Ke, Ruidong Zhang 

College of Mechanical & Vehicle Engineering, Chongqing Jiaotong University, Chongqing, 400074, China 

Abstract: Focused on that the vary of the illumination intensity and the disturb of noise can cause large dif-ferences in the edges and the weak edges may undetected based on the gradient magnitude double threshold method for image processing. A method for edge detection and connection based on gradient magnitude and edge direction is proposed. Make full use of the difference between the noise and edge points in the edge di-rection and the consistency of the edge points in the edge direction to protect the weak edge when suppressed the noise. Meanwhile, the sensitivity of the isolated noise is reduced through the way that calculated the gra-dient amplitude within 8 neighborhoods. The effectiveness of the improved Canny algorithm is verified by experimental results.

Keywords: image process; canny edge detection algorithm; gradient magnitude; edge direction angle; weak edge detection