基于特征向量统计的极化SAR地物分类

发布时间：2018-01-31 23:37

本文关键词： 极化SAR 特征向量三维高斯联合分布散射角方位角　出处：《西安电子科技大学》2014年硕士论文　论文类型：学位论文

【摘要】：极化合成孔径雷达(Polarimetric Synthetic Aperture Radar,简写POLSAR)是一种多参数,多通道的雷达成像系统,通过测量地面上每一分辨单元内的全极化散射回波来获取目标点极化信息。相比单极化合成孔径雷达仅能对特定的电磁波极化收发组合下的目标散射特进行探测,多极化的合成孔径雷达可以为目标解译提供更多更全面的信息。极化SAR图像分类是极化SAR数据应用中的一个关键问题,也是一个难点问题。本文主要研究了在统计框架下的基于特征向量的极化SAR地物分类方法。极化相干矩阵经过特征分解后得到的特征向量包含目标地物的主要极化信息,而最大特征向量所包含的散射信息被认为是占主要地位的。因此,本文从统计分布的角度出发,研究了特征向量的分布特性,并针对特征向量的分布特性给出了相关的分类方法。本文的主要工作包括以下几点:(1)将最大特征向量作为分类特征,对其进行空间转换,简化最大特征向量的表现形式;然后,从统计学的角度,用三维高斯联合分布概率模型来拟合最大特征向量的分布情况,建立了一种最大特征向量的三维高斯分布模型,选用贝叶斯分类器来进行初始分类,并选用加邻域信息的局部Wishart分类器进行二次分类来弥补贝叶斯分类器的不足、提高分类精度。分类结果表明了特征向量模型的有效性及特征向量的可分性。(2)针对最大特征向量对异物同谱地物中的分类效果差的现象,从所有特征向量中提取了散射角作为辅助信息,在统计框架下,将最大特征向量和散射角一起作为分类特征指导地物分类。从所有特征向量中提取了方向角对异质问题的分类结果进行修正,实验结果表明该方法提高了极化SAR图像的分类结果。
[Abstract]:Polarimetric Synthetic Aperture Radar (POLSAR) is a multi-parameter. Multi-channel radar imaging system. The point polarization information of the target is obtained by measuring the full polarization scattering echo in each resolution unit on the ground. Compared with the single polarization synthetic aperture radar, it can only detect the target scattering characteristics under the specific electromagnetic wave polarization transceiver combination. . Multi-polarimetric synthetic Aperture Radar (SAR) can provide more and more comprehensive information for target interpretation. Polarimetric SAR image classification is a key problem in the application of polarimetric SAR data. It is also a difficult problem. In this paper, we mainly study the polarimetric SAR ground object classification method based on the eigenvector under the statistical framework. The eigenvector obtained by the polarimetric coherence matrix contains the main object of the target. Polarization message. The scattering information contained in the maximum eigenvector is considered to be dominant. Therefore, from the point of view of statistical distribution, the distribution characteristics of the eigenvector are studied in this paper. The main work of this paper includes the following points: 1) take the maximum feature vector as the classification feature, and transform it into space. Simplifying the representation of the maximum eigenvector; Then, from the point of view of statistics, the distribution of the maximum eigenvector is fitted by using the three-dimensional Gao Si combined distribution probability model, and a three-dimensional Gao Si distribution model with the largest eigenvector is established. Bayesian classifier is selected for initial classification and local Wishart classifier with neighborhood information is used to make up for the deficiency of Bayesian classifier. The classification results show the effectiveness of the eigenvector model and the separability of the eigenvector. The scattering angle is extracted from all the Eigenvectors as the auxiliary information under the statistical framework. The maximum feature vector and scattering angle are taken together as the classification features to guide the classification of ground objects. The direction angle is extracted from all the feature vectors to correct the classification results of heterogeneous problems. Experimental results show that the proposed method improves the classification results of polarimetric SAR images.
【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

【相似文献】