Edge detection based on fuzzy 2-partition entropy approach

This paper addresses the problem of detecting the edge pixels in the grey-level image which based on the maximum fuzzy 2-partition entropy principle. To divide the edge pixels from the image, we use the direction information measures of pixels as the edge characteristic index, and then an entropy function is used to justify if the information of an image is mostly retained after thresholding. In addition, in order to make sure that the threshold locates at the real valley, a fuzzy range is defined, here. The conducted experiment results demonstrate that the proposed approach can select an appropriate threshold automatically and detect the edge pixels effectively for each test image subject to the entropy within the fuzzy range.