A novel wavelet domain statistical approach for denoising SAR images

Abstract

In this paper, we present a novel Bayesian-based speckle suppression method for Synthetic Aperture Radar (SAR) images within the framework of wavelet analysis. We introduce two-dimensional Generalized Autoregressive Conditional Heteroscedasticity Mixture (2D-GARCH-M) model as an extension of two-dimensional GARCH (2D-GARCH) model and use it for statistical modeling of SAR images subbands. Similar to 2D-GARCH model, this new model can capture heavy tailed marginal distribution and the intrascale dependencies of wavelet coefficients. Also, 2D-GARCH-M model introduces additional flexibility in the model formulation in comparison with 2D-GARCH model, which can result in better characterization of SAR images' subbands and improved restoration in noisy environments. Then, we design a Bayesian estimator for estimating the clean image wavelet coefficients. Finally, we compare our proposed method with various speckle suppression methods applied on actual and synthetic SAR images and verify the performance improvement in utilizing the new strategy.