Texture preserving despeckling of SAR images using GMRFs

Basic textures as they appear especially in high resolution SAR imagery are affected by multiplicative speckle noise and should be preserved by despeckling algorithms. Sharp edges between different regions also must be preserved. To despeckle images the authors use a maxirmum a posteriori (MAP) estimation of the cross section choosing between different prior models. The proposed approach uses a GMRF model for textured areas and allows an adaptive neighbourhood system for edge preservation between uniform areas. In order to obtain the best possible texture reconstruction an expectation maximization algorithm is used to estimate the texture parameters and that provide the highest evidence. The borders between homogeneous areas are determined using a stochastic region-growing algorithm. In this way, they perform a MAP estimate of the image reconstructing either texture of separated regions of uniform backscatter