极化SAR特征量分析与分类研究

发布时间：2018-10-11 08:01

【摘要】：极化合成孔径雷达(Polarimetric Synthetic Aperture Radar, PolSAR)是一种有效获取遥感信息的工具。与传统的单极化SAR相比,它通过测量目标在不同极化收发组合下的散射特性,能够得到更加丰富的目标后向散射信息,有助于更好地感知地物环境,在环境灾害监测、目标检测、资源勘探、城市规划、农作物估计和战场侦察等领域具有广泛的应用前景。因此,如何从这些丰富的极化SAR数据中提取有效的信息,并将之应用于对实际地物目标的分析解译中,是目前雷达信号处理领域的一个研究热点。作为极化信息处理的理论基础,本文首先介绍了电磁波的极化状态、极化的矢量表征方式、不同极化基下的转换关系以及极化在能量域的表示方法,为后续的极化特征量分析与极化分类提供理论支撑。对极化特征量进行分析是极化分类与信息处理的前提,本文从极化通道间的代数运算、基本散射机理的极化特征图以及极化分解理论三个方面对极化特征量进行分析,揭示了极化特征量与散射机理的紧密联系及相关物理意义,并利用实测数据验证了分析结果的有效性。其次,根据H/α极化分解得到的特征量进行初步分类研究,并利用基于统计特性的Wishart非监督方法进行二次分类,实测数据处理验证了该极化特征量分析和分类算法的有效性。通过极化分解理论可以获得多种多样的极化特征量,这些特征量有助于地物分析与解译。然而仅利用几种特征量进行分类,并不能完整地表征目标的特性,导致分类结果失去很多细节信息。本文在多特征量结合的基础上,提出一种基于张量分解降维的极化分类方法。该方法首先联合极化分解特征量构成一个三维极化特征张量,再利用张量分解进行极化特征量降维得到本征特征量集,然后,将降维后的本征特征量输入到分类器中实现极化SAR图像的分类。与传统的基于矩阵降维的极化分类方法相比,该方法可以有效地提高极化SAR地物目标分类精度。仿真数据和实测数据的处理结果验证了所提方法的有效性。
[Abstract]:Polarimetric synthetic Aperture Radar (Polarimetric Synthetic Aperture Radar, PolSAR) is an effective tool for obtaining remote sensing information. Compared with the traditional single-polarization SAR, it can get more information of target backscattering by measuring the scattering characteristics of the target under different polarimetric transceiver combinations, which is helpful to better perceive the ground object environment and monitor environmental disasters. Target detection, resource exploration, urban planning, crop estimation and battlefield reconnaissance have wide application prospects. Therefore, how to extract effective information from these abundant polarimetric SAR data and apply them to the analysis and interpretation of real ground objects is a hot topic in the field of radar signal processing. As the theoretical basis of polarization information processing, this paper first introduces the polarization state of electromagnetic wave, the vector representation of polarization, the transformation relation under different polarization bases, and the representation of polarization in energy domain. It provides theoretical support for the further polarization characteristic analysis and polarization classification. The analysis of polarization characteristic is the premise of polarization classification and information processing. In this paper, the polarization characteristic is analyzed from three aspects: algebraic operation between polarization channels, polarization characteristic diagram of basic scattering mechanism and polarization decomposition theory. The close relationship between polarization characteristic and scattering mechanism and its physical significance are revealed, and the validity of the analytical results is verified by the measured data. Secondly, a preliminary classification study is carried out according to the characteristic quantity obtained from the polarization decomposition of H / 伪, and the Wishart unsupervised method based on the statistical properties is used to carry out the quadratic classification. The validity of the algorithm is verified by the processing of the measured data. Through polarization decomposition theory, a variety of polarization characteristics can be obtained, which are helpful to the analysis and interpretation of ground objects. However, only a few feature quantities can not be used to characterize the target completely, which leads to the loss of many details in the classification results. In this paper, a polarization classification method based on Zhang Liang decomposition and dimensionality reduction is proposed based on the combination of multiple features. In this method, firstly, a three-dimensional polarization feature Zhang Liang is formed by combining the polarization decomposition characteristic quantity, then the eigenvalue set is obtained by reducing the dimension of the polarization characteristic quantity by the Zhang Liang decomposition, and then the eigenvalue set is obtained. The reduced eigenvalue is input into the classifier to realize the classification of polarimetric SAR images. Compared with the traditional polarization classification method based on matrix reduction, this method can effectively improve the accuracy of polarimetric SAR ground object classification. The validity of the proposed method is verified by the processing results of the simulation data and the measured data.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN957.52

【共引文献】