Design of weighted order statistic filters using linearly separable stack-like architecture
C.E. Savin, M. Omair Ahmad, M.N.S. Swamy · 2002
This paper introduces a novel procedure for the design of weighted order statistic filters which are approximately optimal in the mean absolute error sense, for estimating a signal from a noise-corrupted observation of the same. In the proposed procedure, the design is carried out by using a new type of filter architecture designated as a linearly separable stack-like filter. An application for restoring images corrupted by impulsive noise is considered.