Improved Canny edge detection algorithm based on deep learning

Shenghui Pan,Dong Lv

Abstract:In the edge detection algorithm, effectively reducing the noise in the image is the key to edge extraction. In order to solve the problem that the traditional Canny operator needs to set the variance and high and low threshold artificially, the particle swarm optimization algorithm in deep learning is proposed to adaptively optimize the Gaussian filter. The advantages of the proposed method are verified by comparing with the traditional Canny algorithm. The experimental results show that the improved algorithm has better adaptability, can suppress the noise well, reduce the false detection rate, and the detected image contour is clearer.

Computer Science