摘要
该文提出了一种用极化SAR数据协方差矩阵的相关性和回波功率差异来定义的目标间的差异度,并在这一差异度的基础上提出了一种新的迭代分类方法。该迭代方法与基于Wishart距离的迭代分类方法相比,不需要矩阵的求逆运算和矩阵的对数运算,降低了迭代过程的计算量,也不再需要目标的先验信息,扩展了其适用范围。该方法应用于NASA/JPL的SIR-C系统在香港地区的实测极化SAR数据,得到了很好的分类效果。
A concept of difference degree which is based on the coherence of the covariance matrices and the power difference is given in this paper. An iterative method of classification is proposed based on difference degree. This new method not only reduces the compute, but also needs not apriority information of the targets compared with Wishart classifier. This new classifier is applied to the polarimetric SAR image of Hong Kong from the NSAS/JPL SIR-C data and gets an excellent classification result.
出处
《电子与信息学报》
EI
CSCD
北大核心
2006年第11期2007-2010,共4页
Journal of Electronics & Information Technology
基金
国家部级基金资助课题
关键词
极化SAR
差异度
迭代分类
Polarimetric SAR, Difference degree, Iterative classification
