Interferometric phase denoising using Total Generalized Variation (TGV)

The quality of Interferometric phase significantly affects the result of phase unwrapping and even the entire interferometric SAR application. Since traditional space borne interferometric products always contain much noise due to the influence of geometric decorrelation and temporal decorrelation phenomenon, it is necessary to remove the noise from the real interferometric phase. Because of the feature interferometric phase is modulo 2p, the wrapping phase will leads many steep edges in the phase image. Therefore the filter methods must satisfy that they should not break the edge information. This paper proposes an approach for interferometic phase denoising with Total Generalized Variation (TGV) regulation. TGV is developed recently and its mathematical framework is very suitable for interferometic phase noise model. It has favorable abilities of edge preservation and noise removal. In order to test the validity of our algorithm, both simulated and real SAR interferogrms data experiments are performed. The results demonstrate the effectiveness of the TGV denoising algorithm.