Image Edge Detection Based on FCM and Improved Canny Operator in NSST Domain

Abstract

In order to extract edges more completely and accurately, one image edge detection method based on Fuzzy C-means (FCM) and improved Canny operator in Non-subsampled Shearlet Transform (NSST) domain is proposed in this paper. Firstly, the image is decomposed into high frequency component with more edge details and low-frequency component via NSST. Then, the improved Canny operator is adopted to extract few edge in low-frequency sub-bands. While, the modulus maximum detection is performed for each sub-band of high-frequency component, and then we use the FCM method to clustering analysis on the result of the modulus maximum detection to get the high frequency edge. The complete edge image is obtained through the simple weighted fusion of two different frequency edges and edge thinning processing. The experiment results show that the proposed method has better edge detection effect, and the edge location is more accurate, complete, distinct and richer in details.