A Refined Bilateral Filter For InSAR Phase Denoising

Interferometric synthetic aperture radar (InSAR) is a powerful tool for terrain mapping, which can obtain the digital elevation model(DEM) by two or more SAR images. Phase filtering technology can improve the accuracy of interferometric phase, which determines the quality of the DEM. However, the traditional phase filtering methods struggle to balance noise suppression and terrain detail preservation. To solve this problem, this paper proposes a refined bilateral filter for InSAR phase. The main contributions are summarized into two aspects. Firstly, the bilateral filtering method is extended from real domain to complex domain. It consists of spatial and value kernels, which can balance noise suppression and detail preservation. Secondly, a refined bilateral filter based on nested processing is proposed. It embeds the spatial kernel in the filtering of value kernel, which can achieve high-accuracy phase estimation. Compared to the traditional filtering methods, the proposed method can simultaneously maintain the details and suppress phase noise. Computer simulations and real airborne InSAR data validated the proposed method.