SVM全极化SAR图像分类中的特征选择

A New Feature Selection Algorithm for SVM-Based Fully Polarimetric SAR Image Classification

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.