Phase Unwrapping In High-noise Regions Of InSAR Based On Network Flow And Branch Cut Method

Interferometric synthetic aperture radar(InSAR)technology developed on the basis of synthetic aperture radar(SAR)is an important research topic in the field of telemetry and remote sensing.InSAR technology is less affected by weather,time and other factors.It can obtain three-dimensional digital elevation model(DEM)of the target area.It has been used in topographic survey,the detection of surface deformations and other detection activities.It obtains DEM information through registration,interference,removal of flat earth and phase unwrapping,the accuracy of phase unwrapping directly affects the accuracy of DEM.How to obtain the high-precision unwrapped phase of interferograms under high noise has always been the focus of InSAR technology.This paper studies the practical phase unwrapping algorithm suitable for high-noise interferograms,mainly include branch cuts classification,and then improves the long branch cuts set method and combines it with the quality guided method;uses the network flow method to set long branch cuts,and then designs a more suitable filtering algorithm for improved algorithms,etc.The main work of this paper is as follows:(1)An improved branch cut(IBC)method based on branch cuts classification is proposed.Based on the distribution characteristics of residues,the branch cuts are classified.The first type of branch cut is set first,and the second type of branch cut is set between the remaining residues,then the pixels outside the branch cuts are unwrapped,and finally the quality guided method is partially combined with the IBC method.The IBC method greatly shortens the length of branch cuts.The length of branch cuts set by the IBC method is shortened by about 50% compared with that set by the Branch Cut method in simulated data experiment,and by about 60% in measured data experiment.The IBC method improves the accuracy of phase unwrapping,and is suitable for phase unwrapping in high noise areas.(2)Residue classification-based robust phase unwrapping(RCBC)method in high noise is proposed.In this paper,the branch cuts classification method of the IBC method is improved.The network flow method is used to set the long branch cuts,the cost function of the shortest path is established,and the optimization algorithm is used to solve the shortest branch cuts.An approximate substitution method is proposed for the pixels in the branch cuts.A post-filtering algorithm is designed to improve the phase unwrapping accuracy and to enhance the robustness of the algorithm.The length of branch cuts set by RCBC method is very close to the length of the shortest branch cuts.The length of branch cuts set by RCBC method is shortened by about 55% compared with that set by the Branch Cut method in simulated data experiment,and by about 65% in measured data experiment.The RCBC method reduces the probability of "island",improves the accuracy of phase unwrapping,and has certain practicability.