Anisotropic Diffusion Despeckling for High Resolution SAR Images

Abstract : The main purpose of this work is to perform a new denoising method based on a nonlinear anisotropic diffusion for the reducing of the multiplicative speckle in high resolution Synthetic Aperture Radar (SAR) images. In order to be applicable to the sampled data and reduce computing complexity, an efficient discretization scheme, e.g. additive operator splitting (AOS) scheme, is chosen here for the proposed approach. Examples are given comparing the new filter to established speckle noise reduction methods on an ERS-2 SAR image. Experimental results show that this method not only can effectively reduce speckle, while preserving important discontinuities present in the input data, but also obtain a better performance than the traditional local statistical filter.