A New Anisotropic Diffusion Method for Noise Filtering of SAR Interferometric Phase Images

Abstract:
This paper proposes a new nonlinear partial differential equation approach to remove interferometric phase noise. A new expression of diffusion coefficient based on the real interferometric phase is presented and applied to the partial differential process of the complex interferometric phase data. So the discontinuity and inconsistency brought by direct filtering on the real interferometric phase are avoided. The analysis shows that INRAD yields isotropic diffusion in homogeneous regions and yields anisotropic diffusion in the regions near a fringe edge. Furthermore, INRAD is independent of the size of the filter window.