摘要
本文将SVM用于全极化SAR图像分类,并提出一种新的应用于SVM分类的特征选择算法。该算法以支持向量个数作为特征评估准则,利用顺序前进法加入特征。基于NASA/JPL实验室AIRSAR系统的L波段荷兰Flevoland全极化数据的与RELIEF-F算法的对比实验表明,在特征个数更少(或相当)的情况下,本文特征选择算法能在更广泛的SVM参数取值范围内获得更高的分类精度。
SVM is used for fully polarimetric SAR image classification,and a novel feature selection algorithm followed by SVM- based classification is proposed in this paper. In the new algorithm, number of support vectors is taken as estimation rule, and sequential forward selection is used. Using L-band fully polarimetric SAR data of Flevoland, Netherlands, acquired by the NASA/JPL AIRSAR sensor, the new feature selection algorithm is compared with RELIEF-F, higher classification accuracy with less or equivalent number of features in a wider range of the SVM parameters is observed from the experiment results.
出处
《信号处理》
CSCD
北大核心
2007年第6期877-881,共5页
Journal of Signal Processing
关键词
极化合成孔径雷达
特征选择
分类
支持向量机
Polarimetrie Synthetic Aperture Radar
Feature Selection
Classification
Support Vector Machine
