Double adaptive filtering of Gaussian noise degraded images

Abstract

Good estimate and simulation of the behavior of additive noise is central to the adaptive restoration of images corrupted with Gaussian noise. This paper presents a double adaptive filtering scheme in the sense that the filter is able to estimate the variance of additive noise in order to determine the filter gain for pixel updating, and also able to decide if the pixel should remain unfiltered. Experimental results obtained from the restoration of several images have shown the superiority of the proposed method to some benchmark image filters.