Comparative study of SAR interferometric phase filtering algorithms

Interferometric phase filtering is an essential step of the processing chains related to Synthetic Aperture Radar interferometry (InSAR), which are usually used for the generation of digital elevation models and Earth's surface displacement maps through differential InSAR techniques. Performance of the involved processing steps, and in particular of the phase unwrapping operations, are greatly influenced by the ability of the employed filtering algorithms to mitigate the artefacts that affect the interferometric phase. An important impediment with a direct impact on the unwrapping process is due to the presence of spatial inconsistencies in the interferometric phase, called residues. More specifically, these phase inconsistencies are made evident by calculating the curl of the phase differences over any spatial closed loop, which is different from zero in the case a residue is present in an interferogram. In this paper, a comparative study of four different phase filtering algorithms are addressed: coherent averaging, modified median filtering, Lee filter and Goldstein-Wiener filter. The algorithms' performances are mainly assessed by their capacity to reduce the number of the above-mentioned phase residues in the noise-filtered interferograms.