Research On Fringe Processing Of Interferometric Synthetic Aperture Radar

Synthetic aperture radar (InSAR) is a remote sensing technology, which integratethe theories of the synthetic aperture radar (SAR) imaging and the interferometricmeasurement and can be all-time and all-weather to obtain high-precision digitalelevation models and other value-added products over large areas, and is widely appliedin topographical surveying and mapping, environment monitor, geologic prospecting,hydrologic observation and military reconnaissance, etc. In the dissertation, the keyissues of the InSAR data processing mainly including image coregistration, phasefiltering and unwrapping, and differential interferometry are studied in depth, and thensome more efficient approaches are proposed, which can improve the performance ofInSAR data processing of our country. The major works and originalities in thisdissertation are as follows:(1) The issue of InSAR complex image coregistration is studied. The generalprocedures of the image coregistration are introduced, and the existent problem of leastsquare coregistration method is indicated, then a global optimized model of the InSARcomplex image coregistration is proposed and a stochastic parallel gradient descent(SPGD) algorithm is introduced to solve the proposed model. The proposed methodestimates gradient of coregistration performance evaluation function (PEF) usingstochastic parallel perturbation, and achieve the precise coregistration by making thePEF converge to its global extremum. The experimental results indicate that theproposed method can significantly reduce the influence of low coherent points to thefinal coregistration precision, and have advantages of high precision and easy operationcompared with the conventional coregistration method of LS fitting. To solve the issueof the image coregistration with big size, a block coregistration method based on SPGDis further proposed, which can further improve the precision of the coregistration.(2) The phase filtering method of InSAR based on fringe direction is studied. Thedefinitions of fringe direction and density are introduced, then combining with thefringe feature of InSAR, a method, cumulated gradient, with which the fringe directionand density can be calculated simultaneously is proposed. The proposed method utilizesthe anisotropic nature of the phase distribution, and can obtain the result of higherprecision and better robust when the InSAR phase image includes2π phase jumps.Finally, the directionally adaptive filter (DAF) whose foundation is the cumulatedgradient method filters the noise using the information of the fringe direction anddensity and its window size varies adaptively along with the fringe density. Theexperimental results from simulated and real data indicate that the DAF performs betteron phase filtering and can preserve more phase information simultaneously.(3) The method of InSAR phase unwrapping is studied. The conventional phase unwrapping methods are introduced and compared between the experimental results.Then the phase unwrapping method based on contoured window is proposed, whichunwraps the phase in the contoured window calculated from the fringe direction with afoundation of the conventional quality guiding method.(4) Another application of the InSAR, differential InSAR (D-InSAR), is studied.The principle and key procedures of data processing of D-InSAR are introduced. Tosolve the problem of serve decorrelation of the D-InSAR data, contoured correlationinterferometry (CCI) is developed and applied to data processing of D-InSAR. Theresult of the real D-InSAR data of Mani earthquake indicates that CCI method canimprove the phase quality of D-InSAR and can improve the final precision ofdeformation detection.