Research On The Algorithm Of Phase Unwrapping For Interferometric Synthetic Aperture Radar

Interferometric Synthetic Aperture Radar (InSAR), as one of the researchhotspots in microwave remote sensing field, has been generally acknowledged asa powerful tool that can calculate the altitude and the variation of the earth’ssurface even under the adverse circumstances. The concept of InSAR is proposedbased on the SAR theory and interference theory, and it usually observes thesame area on the ground several times then extracts the altitude and its variationaccording to the orbital parameters and the relative positions between theobservation spot and the spacecraft.Data processing is the major part of InSAR imaging, whose proceduresincludes image registration, removal of flatten effect, filter processing, phaseunwrapping, geo-coding. This paper focuses on the phase unwrapping technique.The phase unwrapping algorithms for InSAR, according to whether residuesare taken into consideration or not, can be divided into two categories: the pathfollowing algorithm and the least square method. The branch-cut method as therepresentative of path following algorithm has the short unwrapping time, but ifthere are a certain number of residues in the wrapped phase data, the unwrappingprecision will get worse. The network flow programming algorithm is a newlydeveloped algorithm in recent years. For the reason that it also takes the residuesinto consideration, the network flow programming algorithm still belongs to pathfollowing algorithm. The data got from the network flow programming algorithmhas satisfactory precision and good continuity, whereas its demerit is the problemof occupying large memory and long processing time. The least square method isa kind of global optimal algorithm. It neglects the existence of residues and isbased on the criterion of the least quadratic sum of gradient difference betweenwrapped phase and unwrapped phase. The processing time of the least squaremethod is short and the result has better continuity, but the precision is worsethan the network flow programming algorithm.In order to take full advantage of both the path following algorithm and theleast square method, a new comprehensive algorithm is proposed in this paper.The algorithm is based on the quality graph, defines the concept of area average correlative coefficient, and comprehensively integrates the network flowprogramming algorithm and least square method based on FFT theory. A newway to realize image segmentation and fusion has also been applied for phaseunwrapping.