Research on Non-uniform Three-dimentional Configuration Distributed SAR Tomography
|School||Harbin Institute of Technology|
|Course||Information and Communication Engineering|
|Keywords||SAR Tomography High Resolution Image Registration HybridBaseline Non-uniform Baseline|
SAR tomography technology uses multiple SAR observations from differentheights observing on the same ground scene, forming a large aperture in the height.It has a high ability to distinguish different targets. In recent years SAR tomographyhas been a cutting-edge and a hot topic in the field of synthetic aperture radar.Because of the orbital dynamics and relative perturbation, and taking therequirements of multi-task into account, the actual distributed small satellites havenon-uniform three-dimensional configuration. Tomographic imaging research basedon non-uniform three-dimensional configuration distributed SAR has an importantpractical significance.After detailed analysis, the differences between non-uniform three-dimensionalconfiguration SAR and ordinary linear SAR are hybrid baseline problem and thenon-uniform baseline problem, so tomographic imaging based on non-uniformthree-dimensional configuration distributed SAR needs to study not only theclassical decorrelation and image registration problems, but also the hybrid baselineand non-uniform sampling height-dimensional focus problems. This article studiedthese issues respectively in-depth, and then used the the ALOS spaceborne SAR dataand simulation data to verify the effectiveness of the proposed algorithm. Finally,after analysis of the disadvantages of the classic SAR tomography, the concept ofhigh-resolution tomography is proposed, given the imaging algorithm of lostspectrum estimation based on least squares. The results of spaceborne SAR data andsimulation data verify its feasibility.The main contributions of this thesis are presented as follows(1) To the image registration problem in image compression and distortionconditions, the relationship of image deformation and interferometric phase wasstudied, and deformation compensation based on phase estimation was proposed.The method can match the image as a whole, greatly enhancing the imageregistration of the anti-jamming capability, and spaceborne SAR measured dataexperimental results verified its effectiveness;(2) To the hybrid baseline problem, the article has analyzed and deduced theformula to calculate the impact of hybrid baseline on interferometric phasedifferences between different scatters for the first time. According to their structures,the phase compensation and variable substitution method was given respectively,creating a similar structure of the interferometric phase to the ground after theamendments, which made phase compensation unified, improving the robustness of the system. Simulation results verify its effectiveness;(3) To the non-uniform baseline problem, this paper proposed a frequencydomain least squares APES algorithm based on the SNR weighted, which canachieve a good spectrum estimation with sparse distributed SAR baseline samplesand therefore suited to deal with such height direction focusing problem. By usingthe algorithm on the measured spaceborne SAR data, we obtained tomographicimages of a high-precision, verified the validity of the proposed algorithm;(4) To the difficulty to raise resolution in classical tomography, new conceptfrom the perspective of the full use of spectrum resources was discussed. First theecho spectrum relation of distributed SARs over complex scenes was analyzed, andthen we established a high-resolution tomography model, at last lack spectrumestimation based on least squares algorithm was proposed. Spaceborne SARmeasured data experiment verified the validity of the theories and methods of thisthesis.