Abstract:

Edge detection is a very important research aspect in image field. For the shortcomings the traditional Canny algorithm that thresholds requires human intervention, firstly, the paper replaced the original Gaussian filter with Edgepreserving filter, avoiding the phenomenon of oversmoothing for the original image and effectively suppressing the noise. Secondly, based on the smallest fuzzy divergence within a class, high and low thresholds were determined without requiring human intervention, so that the method is selfadaptive to some extent. Experiments show that the method has good performance in noise suppression and edge continuous, etc.

Key words: edge-preserving filter;the smallest fuzzy divergence within a class;edge detection