SAR Image Despeckling Based on Lapped Transform Domain Dual Local Wiener Filtering Framework

In this paper, a Synthetic Aperture Radar (SAR) image despeckling technique, based on lapped orthogonal transform (LOT) domain dual local Wiener filtering framework, is proposed. A logarithmic transformation is employed to convert the speckle contribution into additive noise. It is demonstrated that the local distribution of dyadic rearranged LOT coefficients of logarithmically transformed SAR images are well approximated using Gaussian distribution. The proposed LT domain structure employs two local Wiener filtering procedures to despeckle the SAR images. The signal variance is estimated using elliptic directional windows for different oriented subbands. The motivation of using lapped transform (LT) is that they are robust to oversmoothing and preserve better oscillatory image components like textures. Experiments on real SAR images, show that the proposed method reduces speckle noise effectively while preserving textures and outperforms well known iterative probabilistic patch-based (PPB) filter and a recent directionlet based method, with much less computational complexity.