成像光谱技术超谱图像分类研究现状与分析

State-of-Arts and Analysis on Hyperspectral Image Classification in Imaging Spectral Technique

超谱遥感图像由于其高光谱分辨率的特点正受到国外国内的广泛关注。在解决了前期的辐射校正及定标等问题以后 ,超谱图像的推广应用是当务之急。超谱遥感图像的分类研究对农作物生长状况的监测、矿物的识别、海洋水色分析以及其他方面的许多应用都是很有价值的。文章概述了国内外在成像光谱技术方面的发展 ,并首次以像元的构成成分为依据 ,从纯像元和混合像元的角度对分类方法进行了论述。对超谱图像分类技术研究现状的归纳和分析 ,旨在寻找突破点 。

Hyperspectral images are widely concerned due to their high spectral resolution.Their application is a task of top priority after solving the problems of radiance correction and calibration.The study on the hyperspectral image classification is valuable to crop growth monitoring,mineral identification and seawater analysis as well as many other useful aspects.In view of this,the development of imaging spectral technique is summarized and for the first time the classification methods are divided into two kinds:pure pixel and mixed pixel,according to the constitutions of a pixel.The object of induction and ana- lysis on the state-of-arts of hyperspectral image classification technique is expected to find breakthrough points and to strengthen our study on them.