Similar neighborhood criterion for edge detection in noisy and noise-free images

摘要

A novel approach for edge detection in noise-free and noisy images is presented in this paper. The proposed method is based on the number of similar pixels that each pixel in the image may have amongst its neighboring in the filtering window and within a pre-defined intensity range. Simulation results show that the new detector performs well in noise-free images but superior in corrupted images by salt and pepper impulse noise. Moreover, it is time efficient.