Adaptive Rank Order Filter for Image Noise Removal

Abstract

This paper presents an adaptive rank order filter (AROF) which is applied to image noise removal. By two observations on the adaptive median filter (AMF) in [4], the AROF is motivated and developed. First, the expansion of window size in the AMF is determine by the criterion if the median is a noisy pixel or not. This criterion is not appropriate when the noise density is moderate or high. Second, the pixels processed by the AMF are reused in the filtering of AMF. This doing degrades the visual quality of restored image. To avoid the two problems found in the AMF, we proposes the AROF. In the AROF, the criterion to expand window size is all pixels in the window are noisy. If it is not the case, the center pixel is replaced by either median or non-median whcih is not noisy. Moreover, the pixels processed by the AROF are not put into account in the following filtereing process. Simulation results show that the proposed AROF has better performance than the AMF when the noise density is moderate or high while remains similar PSNR to those for the AMF in the case of low noise density.