| As a new space earth observation technique,interferometric synthetic aperture radar(SAR,InSAR)technique shows great development potential in the fields of terrain mapping and deformation monitoring with its advantages of all-day,all-weather,high resolution,wide coverage and weak effects of climate conditions.With the development of spaceborne SAR system and the continuous accumulation of SAR data,the developed time series InSAR(TSInSAR)technique has gradually been widely used.This technique further weakens the influence of terrain,atmosphere,decorrelation,and other factors through time series analysis,and improves the deformation monitoring accuracy to the millimeter level.However,due to the sparse distribution of permanent scatterer(PS)in non-artificial surface areas such as mountainous areas and wastelands without typical features,it is unable to effectively obtain surface deformation information.Given this,this dissertation carries out the research on the TSInSAR technique with distributed scatterer(DS),and carries out detailed research on the optimization strategy of the phase information processing method,to improve the richness and monitoring accuracy of the inversed deformation field information.The main work contents and achievements are as follows:(1)Aiming at the problem of improving the optimization effect and efficiency of phase optimization algorithm,an adaptive weighted phase optimization algorithm based on the sigmoid model is proposed.By deriving the general function model of phase optimization,the effects of coherence estimation,weight sets,and solution strategy on the effect and efficiency of phase optimization are revealed.On the basis,a coherence bias correction algorithm based on the second kind of statistical characteristics is introduced to reduce the bias of coherence estimation.And,an adaptive weighting strategy based on the sigmoid model is proposed,and it improves the rationality of the weight setting under the application of all interferometric phase information.Additionally,an efficient solution strategy based on eigenvalue decomposition is derived to solve the redundant solution problem for the nonlinear optimization model.Notable,the proposed adaptive weighted phase optimization algorithm based on the sigmoid model integrates the above three sub-techniques,which effectively improves the overall optimization effect and efficiency of the algorithm.The experimental results show that the phase quality optimized by this algorithm is more than 12%higher than that of the traditional optimization algorithm,and the optimization efficiency is about 20 times higher than that of the nonlinear optimization model solution.(2)Aiming at the problems of blurring or even the loss of phase information in the traditional optimization model at the large gradient phase change,a phase optimization algorithm of fusion the adaptive spatiotemporal filter is proposed,which combined with the proposed adaptive weighted phase optimization model based on sigmoid model.In this algorithm,the phase information obtained by the covariance matrix estimation is treated as the process of filtering in the spatial dimension,and the optimization model estimation is regarded as the process of filtering in the time dimension.Among of them,a spatial adaptive filtering method integrating principal phase component extraction,window adaptive estimation,and post projection technique is proposed.Based on the spatially filtered results and the refined coherence estimator,a construction framework of the complex coherence matrix is built.Furthermore,combined with the adaptive weighting strategy and efficient solution strategy,a phase optimization algorithm fusion of the adaptive spatiotemporal filter is proposed.It effectively takes into account the protection of phase information at the dense fringes in large gradient deformation areas and the improvement of phase signal-to-noise ratio in high noise areas.The experimental results show that this algorithm has better phase information recovery ability at dense fringe areas and optimization effect than the traditional optimization algorithm.The phase quality optimized by this algorithm is improved by more than 15%on average compared with the traditional optimization algorithms,and the number of the screened DS points is maximum increased by about 30%.(3)Aiming at the problems of low accuracy and poor effectiveness of time series phase unwrapping in deformation monitoring applications in complex environments such as high noise and large gradient,a time series phase unwrapping algorithm based on L~p-norm optimized compressive sensing is proposed based on the pseudo three-dimensional phase unwrapping framework.In this algorithm,the theoretical model of the compressive sensing technique representing the sparse characteristics of signals is firstly studied.And,in view of the problem of solving the minimization L~0-norm which can represent the compressive sensing,the transformation idea of minimization L~p-norm with weaker constraints and wider applicability is proposed.Moreover,the iteratively reweighted least-squares(IRLS)is introduced to solve the minimization L~p-norm.Based on the constraints of time-dimensional phase triangular closure,the time dimension phase unwrapping algorithm and unwrapping error correction algorithm based on IRLS solution minimizing L~p-norm are proposed.Combined with model parameters estimation and spatial dimension phase unwrapping based on integer linear programming,a time series phase unwrapping algorithm based on L~p-norm optimized compressive sensing is proposed,which effectively improves the effectiveness and reliability of the time series phase unwrapping.The experimental results show that this algorithm has better noise robustness and unwrapping stability than the traditional time series phase unwrapping algorithm,and the unwrapping efficiency is also significantly improved,up to about 67 times.(4)Aiming at the problems of insufficient density and accuracy of monitoring points in deformation monitoring by the TSInSAR technique in complex environments such as high noise and large gradient,an advanced DSInSAR(ADSInSAR)technique integrating PS and DS points is studied.This technique uses the phase optimization algorithm of fusion the adaptive spatiotemporal filter to preprocess DS,and combines the traditional PS preprocessing to effectively fuse PS points and DS points.Then,the reliable monitoring points are selected effectively by the multi-threshold screening strategy of loosening first and then tightening.According to the time series phase unwrapping algorithm based on L~p-norm optimized compressive sensing,the phase information of monitoring points is accurately interpreted.And,combined with the auxiliary atmospheric data and the spatiotemporal filtering method based on the spatiotemporal characteristics of phase components,the time series deformation field information with high precision and high density of monitoring points is effectively separated.The deformation monitoring results for the Quantai mine and Xinjulong mine show that the ADSInSAR technique proposed in this dissertation can effectively improve the richness and monitoring accuracy of deformation field information inversion in complex monitoring environment.The density of monitoring points obtained by the ADSInSAR is 7 times and 19 times higher than that of the Sta MPS technique respectively.Moreover,the deformation monitoring accuracy is also significantly improved,with the root mean square error of only 2.58 mm and 6.69 mm respectively.This dissertation includes 83 figures,15 tables,and 218 references. |
PDF Full Text Request