A Fast Phase Unwrapping Method for Large-Scale Interferograms

Hanwen Yu,Mengdao Xing,Zheng Bao
DOI: https://doi.org/10.1109/tgrs.2012.2229284
IF: 8.2
2013-01-01
IEEE Transactions on Geoscience and Remote Sensing
Abstract:Two-dimensional phase unwrapping (PU) is a critical processing procedure of synthetic aperture radar interferometry. Thus far, many PU methods with high accuracy have been proposed. However, the limitation of computer's memory requirement is ignored in the design of most of these methods. To effectively solve this problem, a fast PU method for large-scale interferograms is proposed in this paper. With this method, a large-scale interferogram is first partitioned into small tiles according to a strategy based on the residue clustering characteristics, which is the extension and improvement of our previous work. The new tiling strategy has a significant advantage over our earlier work, since it can exactly ensure the consistency between local and global PU results of the L-1-norm criterion. In order to solve the dilemma that high execution speed and high accuracy cannot be satisfied at the same time, which is usually encountered in practice, each tile will be independently unwrapped by minimum-spanning-tree-based PU method either in parallel or in series after tiling processing. By comparing between two representative large-scale PU methods (the SNAPHU method proposed by Chen and Zebker and a large-scale minimum-cost flow method supplied by GAMMA software), it can be seen that the proposed approach is not only efficient in solving large-scale PU problems but also effective in avoiding the inconsistency between local and global PU results generated by image tiling.
What problem does this paper attempt to address?