SAR interferometric phase filtering technique based on bivariate empirical mode decomposition

The empirical mode decomposition (EMD) has been widely applied in filtering synthetic aperture radar interferograms. A noisy interferogram can be adaptively decomposed into different frequency modes by EMD. Then, the noise can be eliminated based on the partial reconstruction of relevant modes. However, most fine detail and noise of an interferogram often locate in the same mode, which will lead to an inaccurate estimation of noise level in a local region. In this paper, we proposed an improved filtering method based on bivariate EMD. The idea of our method is to decompose both the phase image and pseudo-coherence map of an interferogram using EMD. The filter level of an interferogram is then controlled by the parameters calculated from the bivariate EMD components. The quantitative results from both simulated and real data show that the bivariate EMD filtering method outperforms the original univariate EMD-based methods. It could achieve a balance between suppressing noise and preserving fine detail of an interferogram.