针对高光谱图像的目标分类方法现状与展望

Status and prospects of target classification methods based on hyperspectral images

进入21世纪,遥感技术成为一项非常重要的空间成像技术。高光谱图像分类是遥感技术应用中非常重要的一项研究内容,在民用和军用上都实现了应用。高光谱图像分类是通过给每个像元添加分类标签,最终达到区分地物并且识别目标的目的。本文简要阐述了高光谱图像的分类过程及其面临的主要问题;在总结前人研究的基础上归纳了4类主要的高光谱图像分类策略,简要分析了其优缺点及适用范围;分析了近年来出现的新型分类器及其优化方法。最后,对于高光谱图像分类研究存在的主要困难进行了总结,并对未来发展的方向进行了展望。

Hyperspectral remote sensing is one of the most important research directions of remote sensing technology,and hyperspectral image classification is a very important research content in its technical application.This paper briefly expounds the classification process of hyperspectral image and its main problems.Four main hyperspectral image classification strategies are summarized;the new classifiers and their optimization methods in recent years are analyzed.Finally,the main difficulties in the research of hyperspectral image classification and the focus of research and development are summarized.