SAR image despeckling using vector-based hidden Markov model in wavelet domain

Abstract

Despecking is an essential part of any synthetic aperture radar (SAR) imagery systems. In this work, we propose a new despeckling method for SAR images in the wavelet domain. The performance of a method can be significantly improved by taking into account the statistical dependencies between the wavelet coefficients. It has been shown that the vector-based hidden Markov model (HMM) is capable of capturing the subband marginal distribution and the inter-scale and cross orientation dependencies of the wavelet coefficients. In view of this, in order to estimate a speckle-free SAR image, the Bayesian maximum a posteriori estimator using the vector-based HMM as a prior for the wavelet coefficients of images is developed by using the real and synthetically-speckled SAR images. The performance of the proposed despeckling method is evaluated and shown to be superior to some of the existing techniques in terms of providing better preservation of the details and yielding better visual quality.