极化层析SAR人造目标三维重构与特征提取研究

Multi-dimensional imaging is an emerging technique in the radar imaging community. As a new trend of the technique, polarimetric tomographic SAR obtains the ability of discriminating multiple scatterers in the height direction by jointly processing multi-polarization and multi-baseline measurements, and is very promising in obtaining the 3D reconstructions of manmade objects (e.g., vehicles, buildings, bridges etc), which can not be achieved by classic InSAR or PolInSAR. Aimed to achieve the 3D reconstruction and feature extraction of typical manmade targets, this research is to investigate the joint imaging and feature extraction techniques, taking the advantages brought by full polarimetric and multi-baseline measurements. Firstly, we will investigate the sparse aperture design for polarimetric tomographic SAR imaging, and present the corresponding performance analysis results; secondly, based on multiple measurements vector compressive sensing（MMV-CS） model, joint polarimetric tomographic imaging techniques for manmade targets will be proposed; Thirdly, with the advantages brought by polarimetric tomographic SAR, feature extraction techniques for manmade targets in the space-frequency-polarization domain will be studied, aiming to produce the estimates for parameters of targets, such as geometrical size, structure, orientation and so on. The achievements of the research are expected to greatly promote the application of polarimetric tomographic SAR in China.

多维度成像是当前雷达成像领域的前沿热点课题，极化层析SAR是多维度成像领域的最新发展方向。它通过对多极化多基线数据的联合处理，实现高度方向上的多目标分辨能力，能够解决干涉SAR、极化干涉SAR不能适用于城市区域和人造目标（如车辆、建筑物、桥梁等）的三维成像难题。本项目立足于极化层析SAR 这一新兴体制，以典型人造目标的三维重构为目标，挖掘多基线、全极化等多维度优势，研究高适应性的多极化联合的目标三维成像重构与特征提取方法。首先研究稀疏基线设计与性能分析；进而，针对全极化测量的矢量性特点和层析成像的稀疏性特点，研究基于多通道压缩感知（MMV-CS）的极化层析SAR人造目标三维重构方法；在此基础上，充分利用高分辨、全极化和三维成像的优势，研究人造目标的空-频-极化等多维度特征提取技术，实现对人造目标的几何尺寸、结构、姿态指向等参数的反演。项目研究成果将有力推进我国极化层析SAR发展应用。

课题紧密围绕“极化层析SAR人造目标三维重构与特征提取”这一主题，重点开展了基于压缩感知的极化层析SAR三维成像方法及极性性能分析研究，人造目标极化层析SAR三维成像实验，基于CP-GTD模型的全极化散射中心特征提取等三个方面的研究。 .首先，课题组将分布式压缩感知这一压缩感知领域研究的最新成果，引入到极化SAR层析成像中，建立多通道压缩感知联合成像模型，提出一种针对 混合范数正则化模型的快速迭代成像方法，针对信号泄露引起的成像质量降低问题，提出了一种基于滑动窗口迭代的信号泄露抑制方法，实验结果表明该信号泄露抑制算法可以明显提高成像质量。 .其次，开展了极化层析SAR三维成像实验，详细介绍了搭建的地基轨道极化层析SAR的构成、参数和信号处理流程，以重型卡车等为实验对象开展了人造目标的极化层析SAR三维成像实验，实验结果表明通过多极化联合稀疏成像可以获得良好的三维重构图像。.最后，研究了基于三维CP-GTD模型的全极化散射中心特征提取，首先建立了能完整描述雷达目标高分辨极化特性三维CP-GTD模型，在此基础上按照散射中心位置、类型因子和散射矩阵估计、极化协方差矩阵估计和散射中心数目估计的顺序提取目标的目标全极化散射中心，最后以电磁仿真数据和实测数据验证算法能很好地完成目标全极化散射中心特征提取。