I. Introduction
Noise elimination is a fundamental problem in signal processing. Classical signal processing has been dominated by the assumption of Gaussian model. However, a large number of real-world processes such as underwater acoustic noise[1], seismic signal[2] and communication signal[3] etc., are impulsive in nature and are better described by heavy tailed non-Gaussian stable distributions. In recent years, there has been considerable interest in the development of robust techniques for signal processing, based on heavy tailed non-Gaussian stable distributions for the signal statistics [4]–[6].