Research On Calibration Method For InSAR System

In this dissertation the key problems of interferometric calibration are systematically investigated in InSAR systems. Many useful calibration ideas are provided through theoretical analysis and simulation, and valuable discussions have been done by this paper.To investigate the problem of ill-conditioned matrix in the sensitivity model, differential sensitivity model and weighted sensitivity model are derived to reduce the impact of ill-conditioned matrix caused by the differences of parameters, and three calibration algorithms are provided to take advantage of the new sensitivity models, also the location principles of ground calibrators are investigated to optimize calibration algorithms. Simulation results show that the new calibration algorithms have good convergence and stability.To investigate the problem of multi-platform interferometric calibration in the distributed satellites SAR system, united interferometric calibration scheme is derived. During this scheme, the calibration equations of all interferometric platforms are untied by the redundant information of satellites configuration, and all system parameters can be fixed within one scaling procedure. It can not only avoid mutli-calibration procedures, but also reduce the realization requirement of ground control points. Simulation results show the effectiveness of this united interferometric calibration scheme under accurate inter-satellite distance measurementsTo investigate the problem of configuration accuracy in united interferometric calibration scheme, DEM calibration algorithm is derived. Both configuration errors and uncalibrated parameters are introduced in the new algorithm, and the relations between them are modified duiring the iterated procedure. Simulation results show that the algorithm can not only reduce the calibration requirement of configuration accuracy, but also generate accurate DEM.To investigate the problem of algorithm convergence and inherent defection of sensitivity model, a new calibration scheme based on searching model is derived. Combined with the genetic algorithm, parameter errors can be fixed in the high-dimensional error-space by the optimization calibration algorithm. Simulation results show that this calibration scheme can avoid the defection of sensitivity model, and also have good convergence and stability.After systematically comparison of the calibration schemes based on sensitivity model and searching model, a new calibration algorithm is derived to take the advantages of both models. The new algorithm use the results computed by the least-square algorithm based on sensitivity model to be the original values of the gentitic algorithm based on searching model, and gene range could be dynamically reduced. Simulation results show that the algorithm can not only avoid the ill-conditioned matrix of the sensitivity model, but also reduce the computation complexity of the searching model.