Skip to Main Content
In this paper, a new type of rank ordered filter, the "Wilcoxon filter" is introduced. We show that this filter performs efficiently in estimating noisy signals in nonstationary symmetric noise environments where the noise model deviates from an assumed one. It is shown that this filter is the locally most powerful filter for logistic distributed noise. The statistical and deterministic properties of this filter are obtained and compared to those of the median filters; finally, the filter computational complexity is analyzed.