Compressive Sensing framework for simultaneous compression and despeckling of SAR images

Recently Compressive Sensing (CS) has shown great potential in the field of Image processing applications. In this paper we propose a novel CS based technique for simultaneous compression and despeckling of Synthetic Aperture Radar (SAR) images. We incorporate Dual Tree Complex Wavelet Transform (DT-CWT) based denoising within a sparse regularization frame work and solve by a fast projected Landweber reconstrution algorithm. The proposed work contributes in three ways. First the processing is done on data acquired in spatial domain under sub Nyquist rate sampling. Then a two stage denoising in the recovery process eliminates speckles in an excellent way, preserving edges and details effectively. Finally simultaneous compression and denoising is achieved in a simple fast and efficient manner saving the computational cost substantially in both time and memory.