An adaptive fuzzy filter system for smoothing noisy images

This paper presents an adaptive fuzzy filter system that combines the outputs of several filters at any time. It is observed that each objective function associated with a filter yields differing capability in the filtering operation. Hence, all the objective functions are divided into 2 types: multiplicative type (Pi) and summation type (Sigma). In this paper, we present several filter combinations and show that each filter combination called hybrid filter, which consists of filters from both types, is effective in reducing Gaussian type and impulse noises. The outputs of the filters are combined depending upon the value of pixel compatibility, which is computed from the gray-level differences between the central pixel and its neighbours.