Phase unwrapping based on weighted bipartite matching

Two-dimensional phase unwrapping (PU) is a key technique in many scientific fields. In this paper, the L₁-norm-based PU is implemented via a novel method, which is simpler than the classic minimum-cost network flow (MCF) method. It is proved that the network associated with the PU problem can be simplified to a weighted bipartite graph, and matching this graph can solve the PU problem with higher efficiency than the MCF method. Tests on the simulated data and the real InSAR data confirm that the proposed method has a much lower computing complexity than the traditional MCF method.