SAR图像的极化干涉非监督Wishart分类方法和实验研究

该文在合成孔径雷达图像的极化非监督Wishart分类的基础上，给出了一种利用极化干涉信息对合成孔径雷达图像进行非监督分类的方法。该方法主要利用一(66)的极化干涉相关矩阵，从而可以同时考虑单幅图像的全极化信息以及两幅像对之间的互相关信息。该文详细阐述了该方法的具体实现，并利用NASA/JPL的SIR-C/X-SAR系统在中国天山地区的L波段实测数据进行了实验研究。给出了利用该方法对实验数据进行分类的结果，并与极化非监督Wishart分类的结果进行了比较。结果表明，该方法能够很好地分辨不同类型的地物，保持地物的细节，并且比极化非监督Wishart分类结果有很大改善。

Study of Polarimetric Interferometric Unsupervised Wishart Classification of SAR Images

In this paper, an unsupervised Wishart classification scheme for polarimetric interferometric SAR data sets is introduced. A 6x6 polarimetric interferometric coherency matrix is used, in order to take simultaneously into account the full polarimetric information from one single image and the coherency information between the pair of these images. This classification scheme is applied to full polarimetric interferometric L-band SAR image pairs of Tien Shan, China, acquired by the NASA/JPL SIR-C/X-SAR sensor in 1994. The results show that this scheme can identify different terrain types and at the same time keep thr details of surface features, which improve greatly the results of the polarimetric unsupervised Wishart classification.