Nonlinear cumulant based adaptive filter for simultaneous removal of Gaussian and impulsive noises in images

Abstract

This paper presents an edge-preserving higher order statistics (HOS) based filter called nonlinear cumulant based adaptive filter (NCBAF) for noise suppression in images. The NCBAF algorithm combines the linear (averaging) characteristics of the two-dimensional cumulant based adaptive enhancer (2DCBAE) and the nonlinear characteristics of the median filter. The proposed algorithm allows the simultaneous removal of impulsive and Gaussian noise types in images, processed in a single filtering pass. The performance of the proposed method is compared to the commonly used median filter and the 2DCBAE.