Fuzzy one-mean algorithm on edge detection

A class of edge detection filters referred to as the fuzzy one-mean derivative filters (FOM-DF) is introduced in this paper. This class of filters is obtained with a modification of the fuzzy one-mean (FOM) algorithm where polarities are assigned to every input sample. The assignment of polarities are according to prototype derivative filter masks. A particular feature of FOM-DFs is that the output is a convex combination of the input samples. This feature prevents the occurrence of overflow. Another feature is the robustness of edge detection in noisy environments where the images are corrupted by a mixture of white Gaussian noise and outliers.<<ETX>>