Despeckling SAR image using nonsubsampled directionlets combining with GSM model

As speckle noise suppression is important for Synthetic aperture radar (SAR) images processing, this paper presents an approach for SAR image despeckling based on nonsubsampled directionlets. Firstly, images are partitioned into subbands using nonsubsampled directionlets. Then, the coefficients of subbands are modeled with Gaussian scale mixtures (GSM). Besides, for reducing the speckle noise, coefficients are estimated by Bayes least square estimation. Lastly, to evaluate the performance of this despeckling method, four performance evaluation parameters are used. Experimental results show that this method performs better in comparison with the spatial filter and standard wavelet transform not only in speckle suppression, but also in detail preservation.