一种双频极化SAR图像分类方法  

Classification of 2-Frequency Polarimetric SAR Images

作　　者：王文光[1] 孙作为[1] 孙进平[1] 武鹏[1] 

机构地区：[1]北京航空航天大学电子信息工程学院,北京100191

出　　处：《信号处理》2011年第10期1552-1556,共5页Journal of Signal Processing

基　　金：国家自然科学基金(61001137);ATR重点实验室基金(9140C8004011008)

摘　　要：本文提出了一种利用两种不同频率下的极化SAR图像进行地物分类的新方法,该方法包括了两个过程,初始类划分过程和迭代修正过程。初始类划分过程是基于目标的散射特性随频率变化而改变的趋势和程度实现的。与传统的双频极化SAR图像分类中的初始类划分方法不同,新的划分方法基于不同频率下所提取的特征量,定义了特征变化量和特征变化平面,在特征平面上直接进行多类目标的初始类划分,而不需要迭代和类合并过程。这种新的初始类划分方法反映了目标散射特性随频率的变化关系,物理意义直观,实现方法简单易行。将这种初始类划分方法与Wishart分类器相结合,就可以实现对极化SAR图像的无监督迭代分类。实测的SIR-C数据和AIRSAR数据的分类结果表明,该方法是一种有效的极化SAR图像分类方法。A new classification method of 2-frequency pol-SAR images is proposed in this paper,which is composed of two processes,the initialization and iteration.This initialization classification method is based on that the scattering properties vary with the changing of imaging frequencies.The new method is different from the initialization process of traditional 2-frequency H-α- Wishart classifier in that it define new feature variation parameter and feature variation plane,which are based on feature extraction of single frequency pol-SAR images.The initialization can be obtained from the feature variation plane directly,not by iteration and classes clustering. This method is simple both to understand and to apply.The new unsupervised classifier is developed by combining this initialization with the Wishart classifier.The new classifier is applied to the measured NSAS/JPL SIR-C data and AIRSAR data,and satisfied results are obtained.As a conclusion,this new unsupervised classifier is effective to pol-SAR images.

关 键 词：极化合成孔径雷达 分类 散射特性 双频 

分 类 号：TN957[电子电信—信号与信息处理]