Performance of Detail-Preserving Weighted Median Filters for Image Processing
Ruikang Yang, Lin Yin, Moncef Gabbouj, Jaakko T. Astola, Yrjo A. Neuvo · 2005
In this paper, we present a method to design optimal two-dimensional detail-preserving weighted median (WM) filters in the sense that these filters preserve some desired image details, such as lines, corners, etc., under which they achieve the maximum noise attenuation. The performance of these detail-preserving WM filters is compared to that of FIR-median (FMH) hybrid filters which have shown good detail-preserving properties for image processing. Responses under different signal-to-noise ratios are carried out to characterize the performance of these detail preserving WM filters.