基于特征向量分析的极化SAR图像分类

发布时间：2018-05-20 05:42

本文选题：极化合成孔径雷达 + 图像分类　；参考：《西安电子科技大学》2014年硕士论文

【摘要】：极化合成孔径雷达(Polarimetric Synthetic Aperture Radar,POLSAR)是一种多通道、多参数的雷达成像系统,它是通过测量地面上每一个分辨单元内的全极化散射回波来获得目标极化信息的。与传统的成像雷达相比,极化合成孔径雷达能够在多种电磁波极化收发组合下对目标散射特性进行探测,所以它能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数据应用上的一个重要问题,本文主要研究基于极化相干矩阵特征向量分析的极化SAR图像分类算法。选择特征向量作为分类特征,结合支持矢量机(SVM)分类算法和谱聚类,提出了两种极化SAR图像分类算法,主要内容如下:1.对极化SAR数据相干矩阵进行特征值分解,统计分析了三个特征向量的数值特征,利用特征向量较好的可分性,选择最大特征值对应特征向量的实部和虚部作为特征量,结合SVM分类器实现极化SAR图像分类,该方法简单易实现,分类效果良好。2.进一步对相干矩阵特征向量的数学意义进行分析,结合极化SAR雷达成像的物理机制,提出了一种适用于相干矩阵特征向量的相似性度量方式。该方法中两个像素的相似度计算是在酉空间上进行的,实验验证了该相似性度量的正确性和有效性,在此基础上,提出了一种基于特征向量度量谱聚类的极化SAR图像分类算法。该方法计算复杂度低且分割效果好。本工作得到了国家自然基金(Nos.61003198,61472306)和中央高校基本科研业务费专项资金(JDYB140508)的资助。
[Abstract]:Polarimetric Synthetic Aperture Radarsard (Polarimetric Synthetic Aperture POLSAR) is a multi-channel, multi-parameter radar imaging system, which obtains the polarimetric information of the target by measuring the full polarization scattering echo in each resolution unit on the ground. Compared with the traditional imaging radar, polarimetric synthetic aperture radar can detect the scattering characteristics of the target under a variety of electromagnetic wave polarization transceiver combinations, so it can provide more and more comprehensive information for target interpretation. Polarimetric SAR image classification is an important problem in the application of polarimetric SAR data. In this paper, a polarimetric SAR image classification algorithm based on eigenvector analysis of polarimetric coherence matrix is studied. Two polarimetric SAR image classification algorithms are proposed by selecting feature vectors as classification features and combining support vector machine (SVM) classification algorithm and spectral clustering algorithm. The main contents are as follows: 1. The eigenvalue decomposition of the coherent matrix of polarized SAR data is carried out, and the numerical characteristics of the three eigenvectors are statistically analyzed. The real and imaginary parts of the eigenvector corresponding to the maximum eigenvalue are selected as the eigenvalues by using the better separability of the eigenvectors. The polarimetric SAR image classification is realized with SVM classifier. The method is simple and easy to implement, and the classification effect is good. 2. Furthermore, the mathematical meaning of the eigenvector of coherent matrix is analyzed, and a similarity measurement method suitable for the eigenvector of coherent matrix is proposed in combination with the physical mechanism of polarimetric SAR radar imaging. In this method, the similarity calculation of two pixels is carried out in unitary space. Experiments show that the similarity measure is correct and effective. On the basis of this, a polarimetric SAR image classification algorithm based on eigenvector metric spectral clustering is proposed. This method has low computational complexity and good segmentation effect. This work is supported by the National Natural Fund "Nos.6100319891472306" and the special fund for basic scientific research business expenses of central colleges and universities (JDYB140508).
【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】