地面、图像光谱数据的预处理

Pretreatment of Flied and Image Spectra Data

在高光谱遥感研究中,需要地面光谱和图像光谱的结合分析处理。地面光谱的正确采集,两种光谱数据的预处理,由于与分析过程的直接联系不大,往往被忽视。其实地面和遥感所采集的原始数据并不能直接用于分析,对其的预处理涉及到格式转换、数据消噪等问题,这些都是高光谱科学分析研究的前提,关系到结果的正确性。虽然目前介绍光谱分析研究的文章和书籍很多,但是完整介绍光谱采集和预处理,并可用于实践的甚少。对于刚刚涉入高光谱领域的同学和老师,对此类问题经常感到迷茫。作者在近些年内参加了多次地面光谱采集和遥感飞行试验,对大量的地面光谱和图像光谱进行预处理。主要讲述作者通过实践过程中摸索比较,总结出的简单易行、能获得较好效果的光谱预处理方法;同时介绍光谱采集的正确方法。地面光谱数据为ASD-FR2500采集数据(该种野外光谱仪在国内外比较普遍),图像数据为OMIS图像。其它光谱仪或者遥感成像仪器的数据可以类似处理。

In the research of hyper-spectral remote sensing, we need combine field spectra and image spectra to analyze. The correct collection of field spectra and pre-process of two kinds of spectra are often ignored by us, for there is little relationship between them and information analysis. In fact, before analysis directly, the raw data gotten from field or from remote sensing need format-change and de-noise etc. Though there are many books about spectra analysis now, none of them introduce how to collect spectra or spectra pre-process. Many persons who just begin to deal with hyper-spectral always feel confused. The author of this article has taken part in many projects about field spectra collection and airborne remote sensing, pre-processed a lot of field and image spectra in these years. This article introduces the author＇s experience and methods summarized from experiments from three aspects： （1）how to collect field spectra correctly; （2）how to pre-process field spectra; （3）how to pre-process remote sensing images with field spectra. In this paper, the field spectra are collected by ASD_FR2500 and images are OMIS images. Data from other spectral apparatus and remote sensors can be dealt with analogously.