Study On Signal Processing For SAR Interferometry

Synthetic aperture radar interferometry (InSAR) is an increasingly expanding technique allowing for the estimation of three-dimensional terrain images with high spatial resolution and height accuracy, independent of weather conditions and daylight. Based on SAR systems, InSAR needs several observations over the same scene at different time or along different tracks, together with the knowledge of the interferometer geometry and phase differences between each pair of the received signals that can be converted into a digital elevation model (DEM) or topography-changing map. The wide and the potential applications of InSAR in many areas such as military, civil and scientific researches, make one of the most active fields in radar and remote sensing.It is well known that image coregistration, interferometric phase noise filtering, and phase unwrapping are three major processing procedures of InSAR, and the three procedures are cascaded in conventional InSAR processing flow. If the information is impaired at the preceding stage, it is very difficult for those methods to retrieve the lost information at the current processing stage. Due to registration error, interferometric phase noise and wrapped phase, DEM reconstruction is still a challenging problem. Aiming at these problems above, this dissertation discusses the signal processing of interferometric SAR deeply, with emphasis on the key techniques such as SAR image coregistration, interferometric phase estimation (or noise filtering) and phase unwrapping. The main works of the dissertation can be summarized as follows.A. SAR Image CoregistrationSince image coregistration task is one of key processing procedures of InSAR. Furthermore, the performance of image-parameter- based techniques is also easily affected by non-linear distortions and noise present in the image. However, the interferometric phases obtained are random signals with their variances being inversely proportional to the correlation coefficients between the corresponding pixel pairs of two coregistered SAR images. It is very important to coregister SAR images using a statistical method before terrain interferometric phase estimate. The auto-coregistration image for interferometric SAR phase images is presented. The auto-coregistration of the SAR images is characterized by using the construction the model of the optimal weighted joint data (OWJD) vector via the MMSE criterion. The model is verified with the proof of the optimal weight vector and the measures of coregistration consistency. Numerical results on the simulated data and real data from ERS1/ERS2 demonstrate the efficiency and precision of the proposed method.B. Interferometric Phase EstimateWe study the issue of the noise subspace estimate via eigenvalue in the presence of the image registration error. A method of estimate the interferometric phase based on computation the noise subspace via the projection vector composed of the InSAR interferometric phase and the joint eigenvectors is presented. Theoretical analysis and computer simulation results show that the method can provide accurate estimate of the InSAR interferometric phase even if the registration error be close to one pixel. Moreover, a method of interferometric phase estimation based on the robust beamforming is introduced. The proposed method can take advantage of the optimal weighted joint data vector and the robust adaptive beamformer to auto-coregister the SAR images and mitigate the effect of the spatial steering error on the interferometric phase estimation.C. Phase Unwrapping and DEM ReconstructionFour robust processing techniques based on interferometric phase unwrapping and DEM reconstruction are introduced. The methods include (1) phase unwrapping based on the minimum cost network flow in SAR interferometry, (2) multibaseline InSAR with image auto-coregistration for phase unwrapping, (3) optimum data vector approach to multibaseline SAR interferometry phase unwrapping, and (4) application of array processing techniques in multibaseline InSAR for high-resolution DEM reconstruction. These theoretical analysis and experimental results, supplemented with the Cramér-Rao lower bounds (CRLBs) for the estimated interferometric phase show that the methods can accurately provide the unwrapped interferometric phases in the presence of the large image registration errors.