Comparison of phase unwrapping algorithms for SAR interferograms

The interest in advanced methods of remote sensing has stimulated various investigations on interferometric synthetic aperture radar. The determination of the absolute phase from wrapped phase values has remained a critical step in SAR interferometry. The different approaches to the phase unwrapping problem can be broadly categorized into: integration methods, e.g. the branch-and-cut method, fringe tracking methods, and Least-Squares (LS) techniques. This paper provides a comparison of the branch-and-cut method and some LS techniques which were recently proposed [1,2]. In particular, we have studied the strategies of both approaches to deal with phase inconsistencies, i.e. phase residues, which are introduced by noise and/or undersampling in SAR interferograms. While the branch-and-cut method deals with phase residues directly by integrating along suitable paths that avoid the crossing of imaginary lines (cut-lines) connecting residues, it is not obvious how LS techniques are dealing with this problem. We illuminate this point and relate the LS approach to the one taken by integration methods. Finally, we draw some conclusions based on our findings.