An adaptive edge enhancing order statistic filter

Abstract

A filter which combines L-type structure with a modified median filter (MMF) is proposed for suppressing nonimpulsive as well as impulsive noise. The MMF is designed to reduce edge jitter caused by the median filter. The MMF output is obtained by modifying the median filter output, based on results of hypothesis tests. The proposed filter, named the decision-based order statistic filter (DBOSF), generates the output by using asymmetrical alpha -trimming with respect to the MMF output. In addition to noise suppression, the DBOSF can enhance the gradient of blurred edges provided that the size of the window is greater than twice the transition with width of the blurred edges. Experimental results for one- and two-dimensional signals are presented to demonstrate performances of the proposed filter.<>