A new phase unwrapping algorithm combining minimum surface-fitting and network flow for InSAR

Phase unwrapping is a key step of interference synthetic aperture radar (InSAR) signal processing. The unwrapped results can be strongly influenced due to the reason such as noise and under-sampling. The network flow algorithm is able to effectively suppress noise in high coherence regions, as well as overcomes the side effect of `island' phenomenon and confines the error transfer to a very small limit. However, spikes and glitches still exist in the results. Combining the concept of residues, the surface fitting method and the network flow algorithm, this paper proposes a new phase unwrapping algorithm. The new algorithm locates the noise points based on the correlation degree of neighborhood pixels, reconstructs the noise value using minimum surface fitting, and finally corrects the integral results along the integration path. Repeat the iteration until all the noise points are eliminated and many minimum surfaces are fitted repeatedly to form the final unwrapping result. Results obtained from the simulated data and the measured data show that, the proposed method not only achieves better solution with an acceptable time consumption, but also weakens the spikes and glitches as well as restrains the error transfer of the unwrapping process, increasing the precision of phase unwrapping.