An Adaptive Fuzzy Entropy Algorithm in Image Edge Detection

Abstract

To improve the effectiveness of image edge detection, this paper proposed a weighted image edge detection algorithm based on fuzzy entropy. The scheme firstly defines three image metrics to represent image edge: orderliness measure, directivity measure and structural measure. Then it extracts these metrics from images based on fuzzy entropy, and weights three metrics to calculate a confidence degree for edge decision, in which the weighted factors may fully consider the metric impact to detect edges. Finally the algorithm adopts non-maximal image suppression with dynamic threshold to inhibit image pseudo-edge for further optimization. Experiment results illustrate better performance in edge detection with maintaining more image details.