Residue Classification-Based Robust Phase Unwrapping in High-Noise Region

Abstract

Synthetic aperture radar interferometry can obtain information on the surface of the earth. In fact, the accuracy of phase unwrapping directly affects the accuracy of the digital elevation model, but the problem that has been puzzling people is the low accuracy of unwrapping in high-noise regions. To solve this problem, we propose a new approach based on an improved branch-cut method, which classifies the branch cut. First, connect the short branch cuts and then use the network-flow method to connect the long branch cuts. By categorized branch cuts, each type of branch cut processes part of the residues and divides the region with high residue density into two regions with low residue density for unwrapped, which solves the problem of low phase unwrapping accuracy in high-noise regions. The experimental results of simulation and measured data show that RCBC method shows high unwrapping accuracy in high-noise regions, very short total length of connected branch cuts, and strong robustness.