A new edge-preserving smoothing filter based on fuzzy control laws and local features

This paper proposes a new type of edge-preserving smoothing filter to be applied to an image corrupted by impulsive and white Gaussian noise. This filter is based on a weighted mean filter having two kinds of coefficient for impulsive and white Gaussian noise. These coefficients can be varied adaptively by local features in the window. Fuzzy control laws that can be used for the implementation of the proposed filter are described. To reduce the computational burden, the fuzzy inference model proposed by Ichihashi et al. (1990) is employed.