PolInSAR地表参数反演方法研究

【作者】 韩迪；

【导师】 张晓玲；

【作者基本信息】 电子科技大学 ， 信号与信息处理， 2009， 硕士

【摘要】 极化干涉合成孔径雷达(PolInSAR)集PolSAR和InSAR测量技术于一体,可以同时把目标的精细物理特征与空间分布特性结合起来。极化干涉测量不仅能够提高常规InSAR测量的精度,而且有助于更好的理解目标的散射机理和所发生的散射过程,进一步拓展了SAR在成像遥感领域的应用,在军事和民用领域具有无法估量的作用。PolInSAR可用于反演地表植被参数,特别是森林的高度参数。地表参数的提取为地表分类及进一步目标识别奠定了基础,因此用极化干涉雷达数据提取地表植被参数是当前极化干涉研究的热点问题。论文主要研究了既有的PolInSAR植被高度反演方法,然后在此基础上提出了一种融合的植被高度反演算法,最后对极化相干层析(PCT)算法进行了初步的探讨。主要内容包括:极化干涉SAR相干最优理论,相干散射模型,单基线植被高度反演方法,基于DEM差分法和相干幅度法的融合反演算法,PCT算法。本文首先对PolSAR的基本理论进行了阐述,包括目标极化散射特性的表征和基本极化散射机理。然后对PolInSAR技术中的复干涉相干系数和极化相干最优理论给出了详尽的推导和分析。最后对相干最优理论进行了仿真验证。接下来,论文讨论了相干散射模型,重点研究了RVoG相干散射模型和相应的参数反演模型,并分析了模型中各个参数对干涉相干性的影响。在对PolInSAR理论进行深入讨论后,论文研究了两种既有的经典的单基线反演方法(三阶段法和ESPRIT算法),在此基础上研究了ESPRIT算法的改进算法,同时提出了一种基于ESPRIT算法和相干幅度法的联合反演方法。利用仿真数据对这各种方法进行了验证和性能比较。针对三阶段法和ESPRIT算法存在的问题,论文提出了一种基于DEM差分法和相干幅度法的融合反演方法,对其原理进行了详细的阐述,并利用PolInSAR仿真数据对此方法进行了仿真验证。最后,对PCT算法进行了初步探讨,并利用仿真数据对其进行了验证。更多还原

【Abstract】 Polarimetric SAR interferometry (PolInSAR) is the technique that integrates radar polarimetry and radar interferometry. The interferometric analysis of a full coherent polarimetric data set enables the combination of final structure properties and spatial information of the targets. PolInSAR is a potential technique that not only improves the accuracy of the interferometric measurements but also allows a more sophisticated physical interpretation of scattering mechanism. With lots of advantages in many areas of military, civil and scientific researches, PolInSAR has extented the applications of SAR remote sensing, and has wide and latent prospective applications and valuable researching aspects.PolInSAR has been used to invert vegetation parameters of terrain surface, especially for height of forest. These parameters are the bases of terrain classification and target detection and recognition. Therefore, terrain surface vegetation parameters inversion using PolInSAR data has been hot research topics in polarimetric interferometry in recent years.The dissertation mainly analyses and studies the established terrain surface vegetation parameters inversion methods for PolInSAR, on basis of these methods, the dissertation proposes a fast and robust vegetation height inversion method. The main contents include Polarimetric Interferometric Phase Coherence Optimization(PIPCO); Random Volume over Ground (RVoG) coherence scattering model and corresponding inversion model; single baseline vegetation height inversion method; vegetation height inversion method based on the fusion of DEM differencing algorithm and Coherence Amplitude algorithm; Polarimetric Coherence Tomography algorithm.Firstly, this dissertation provides an overview of basic theory of PolSAR, such as representation of polarimetric scattering characters of targets and elementary scattering mechanisms. Then the calculation of complex coherence coefficient and the theory of PIPCO for PolInSAR are discussed comprehensively. Furthermore, the theory of PIPCO is validated by simulated data.Secondly, coherence scattering models are studied. The study is focused on the RVoG scattering model and corresponding inversion model. The influences upon coherence of all the parameters in the model are discussed.Thirdly, after in-depth discussion of theory of PolInSAR, two existed and classical single baseline inversion methods (the Three-Stage method and the ESPRIT algorithm) are studied, on the basis of it, an improved ESPRIT algorithm is studies, and a hybird vegetation height inversion method based on the ESPRIT algorithm and Coherence Amplitude method is proposed. these methods are validated by simulated data and the performance of them is compared.For the questions of exist in the Three-Stage method and the ESPRIT algorithm, the dissertation proposes a kind of inversion method based on the fusion of DEM differencing algorithm and Coherence Amplitude method, the principles of it are discussed comprehensively, and the simulated PolInSAR data are used to validate the proposed method.Finally, the dissertation analyses and studies PCT algorithm, and the simulated data are used to validate this algorithm.更多还原