Edge-aware nonlinear diffusion-driven regularization model

Our model integrates a high-order diffusion kernel, it includes a fractional regularization term sensitive to speckle noise in SAR images.

In this work, an Edge-aware nonlinear diffusion-driven regularization model is implemented to despeckle SAR images. The effective interplay between the proposed regularizer and the regularization term generates sharper and detailed SAR images with the minimum possible staircase and block artifacts - the common problems encountered by the classical Perona-Malik and Total variation models.