Study On Key Algorithms For Signal Processing In Spaceborne Interferometric SAR

Up to now since L. C. Graham proposed the concept of interferometric synthetic aperture radar (InSAR) in 1974, numerous scientists, researchers and engineers have done a lot of works on InSAR theory and signal processing, and have made significant progresses with the advancement of InSAR hardware and increase of experimental data taken by both spaceborne and airborne systems. However many of them focuses on how to process the data and analyze the results qualitatively and quantitatively with correlation coefficients and residues, less of them focus on the accuracy issues, i.e. root-mean-square (rms) of height measurment involved in different processing steps because of lack of accurate (Digital Elevation Model) DEM data in experimental areas. The purpose of the dissertation is to build a spaceborne InSAR simulation model and analyze algorithms and parameters in key steps of InSAR signal processing in view of rms of height measurement based on our simulation model, some new algorithms and ideas have been proposed.The main work of dissertation is to firstly build a spaceborne SAR and InSAR simulation model based on orbital dynamics. According to the difference between airborne and spaceborne SAR imaging, the effective velocity must be estimated in the spaceborne SAR because the spaceborne Chirp Scaling algorithm need the parameter to take place of the satellite track velocity. After the spaceborne InSAR simulation model is correctly established and the right interferograms are obtained, analysis and discussions on height accuracy of algorithms and parameters involved in InSAR signal processing are conducted, such as InSAR image-pair registration, phase filtering and phase unwrapping. Detailed works and conclusions are summarized as follows,(1) In the issue of complex images registration, regarding to the problems of conventional InSAR images registration methods, such as inaccuracy and inability of calculation of small rotation angles between images pair, we put forward modified correlation coefficient algorithm and a new registration algorithm based on Fourier-Mellin transformation and correlation coefficient algorithm to solve the problems. We also analyze the influence of the multiple interpolations and the size of matching window on final height accuracy under different signal noise ratio (SNR).(2) In issue of noise filtering of interferogram, we firstly compare the effects of six common filtering algorithms with different window size of filter and then obtain the best filtering algorithm and the best window size of filter under different SNR. Moreover, we study the performance of multi-stage filtering method and the effect of combination of different filtering algorithms. The results show that two-stage filtering always has better effect than single-stage filtering and higher-stage filtering (three stages or more) is unnecessary.(3) In issue of phase unwrapping, we discuss the typical phase unwrapping algorithms in detail, and make a comprehensive comparison among them under different SNR and window size of filter, and draw conclusion on the most appropriate phase unwrapping algorithm under different SNR.(4) Finally, we investigate the interferometric baseline combination issue of longer length with shorter length formed by distributed satellites InSAR system. In view of the problems of maximum likehood multi-baseline data fusion method and iterative multi-baseline data fusion method, we propose a new multi-baseline data fusion algorithm by combining the advantages of maximum likehood and iterative methods. Our simulation show that the new algorithm performs much well than the above two methods do.