SVM在高光谱图像处理中的应用综述

摘要

Hyperspectral remote sensing has become a foremost technology in remote sensing,and it plays an impor-tant role in military and national economy. Support vector machine (SVM) has unique advantages in solving the problems of small sample size, nonlinearity, and high-dimensional modes; therefore, it is widely used in hyper-spectral data processing. Because of its advantages,SVM model has been applied widely in fields of hyperspectral imaging,such as band selection,classification,endmember selection,spectral unmixing,sub-pixel mapping,and anomaly detection. In this paper,the features of hyperspectral images are analyzed,and the development of hyper-spectral imaging across various fields as well as its main processing methods are summarized. The applications and advantages of SVM method in those fields are also discussed.%高光谱遥感已经成为遥感领域的前沿技术,在军事以及国民经济中发挥着重要作用.支持向量机(support vector machine,SVM)在解决小样本、非线性和高维模式等问题中具有特有的优势,因而被广泛用于高光谱数据处理.在高光谱图像的波段选择、分类、端元选择、光谱解混及亚像元定位、异常检测等主要领域,SVM模型皆因其特性而表现出独特优势并已广泛应用.分析了高光谱图像特性,总结了当前各领域的发展现状及主要的处理方法,并对SVM方法在各领域中的应用及优势进行了阐述.