Abstract:

In order to improve the detection ability of fuzzy image edge and robust to noise,this paper proposed a novel edge detection method based on the fuzzy entropy and the characters of image edge structure.In this method,a nonlinear function was used to transform the feature space of image gray levels into the one of fuzzy entropy so as to enhance the contrast of the fuzzy edge region.Then,twelve valid edge patterns were defined in a 3×3 neighborhood of the pixel and were used to extract the map arrays of the structure-and-direction-information measures for each pixel.Finally,the non-maximum suppression was performed for the two arrays to determine the final edge pi-xels. Experimental results show that the proposed algorithm performs well in terms of ability to detect fuzzy edge and robust to noise.The final edge image is precisely localized with single pixel.

Key words: edge detection, fuzzy entropy, edge structure, non-maximum suppression