High-Resolution SAR Tomography In Complex Urban Scenarios

Thanks to the high resolution property of modern high-resolution Synthetic Aperture Radar (SAR) satellites, the details of buildings and urban infrastructure can be observed. However, layover and foreshortening effects, induced by the intrinsic side-looking geometry of the SAR sensors, make it difficult to interpret the data. Especially in urban scenarios, many resolution cells contain the superposition of responses from multiple scatterers (layover). As an advanced technique, SAR Tomography makes it possible to overcome the layover problem. It is able to detect multiple scatterers superimposed inside one resolution cell, aiming at a real and unambiguous 3D SAR imaging. The deformation of each scatterer is also detected in the case of Differential SAR Tomography.Singular Value Decomposition (SVD) is an efficient tomographic inversion tool. In order to limit the noise propagation induced by small singular values, the usual solution is cutting off the small singular values with a hard threshold, which is called Truncated-SVD or TSVD. The decomposition result is therefore highly dependent on the threshold, which means choosing the right threshold is vital to the success of TSVD. In order to avoid choosing a hard threshold and also make the tomographic procedure adaptive to different data stacks, we propose a Butterworth filter based singular value decomposition method in our previous research, which we call BSVD. For the purpose of evaluating the performance of BSVD, both simulated results and experimental results on TerraSAR-X data are analyzed and compared with TSVD. The outcome of both experiments with simulated and real data shows that Butterworth-SVD is a robust and efficient tool for SAR Tomography.Traditional non-parametric spectral estimators, e.g. Truncated Singular Value Decomposition (TSVD), Butterworth-SVD (BSVD), are limited by their poor elevation resolution. On the other hand, the compressive sensing based approaches using the Basis Pursuit (BP) strategy to find an L1-Norm minimization solution for SAR tomography are extremely time consuming. Therefore, a fast and robust tomographic algorithm with super-resolution capability is needed. We propose a new approach for SAR tomography based on Two-step Iterative Shrinkage/Thresholding (TWIST). TWIST uses a two-step strategy to speed up the L1-Norm minimization procedure and can achieve an exceptionally fast convergence speed for SAR tomography. Experimental studies with simulated signals and TerraSAR-X datasets are carried out to demonstrate the merits of the proposed TWIST approach in terms of robustness, fast convergence speed, and super-resolution capability.In this thesis, the proposed BSVD and TWIST approaches are also extended into Differential SAR Tomography, to estimate the deformation of each scatterer as well as their 3D position in space. Experimental studies with both simulated data and TerraSAR-X datasets are conducted to demonstrate the potential of BSVD and TWIST in Differential SAR Tomography. The research work on TomoSAR and D-TomoSAR in this thesis is able to make full use of the high resolution SAR datasets, making the dynamic monitoring of 4D City significantly promising.