基于光谱特征的湿地植物种类识别

发布时间：2018-01-15 10:38

本文关键词：基于光谱特征的湿地植物种类识别　出处：《国土资源遥感》2016年03期 　论文类型：期刊论文

【摘要】：光谱特征的选择对于湿地植被的识别精度和效率有直接的影响作用。以美国舍曼(Sherman)岛水域为研究区,基于Hy Map航空高光谱遥感影像数据,分析湿地植被的一阶微分光谱和光谱吸收特征,利用逐步判别分析法筛选识别精度较好的光谱特征参数参与C4.5决策树分类。结果表明:4种湿地植被的一阶导数光谱特征差异较小,吸收特征差异性相对较大;基于一阶微分光谱特征和光谱吸收特征利用C4.5决策树进行分类,可以实现湿地植被在物种水平上的识别,并达到较好的分类精度。
[Abstract]:The selection of spectral features has a direct impact on the accuracy and efficiency of wetland vegetation identification. Based on Hy Map aerial hyperspectral remote sensing data, the first order differential spectrum and spectral absorption characteristics of wetland vegetation were analyzed. Stepwise discriminant analysis was used to select the spectral characteristic parameters with better recognition accuracy to participate in the classification of C4.5 decision tree. The results showed that there was little difference in the first derivative spectral characteristics of the four types of wetland vegetation. The difference of absorption characteristics is relatively large; Based on the first order differential spectral feature and spectral absorption feature, the classification of wetland vegetation on species level can be realized by using C4.5 decision tree, and the classification accuracy is better.
【作者单位】：河海大学地球科学与工程学院;河南省南阳水文水资源勘测局;
【基金】：中国科学院战略性先导科技专项(编号:XDA05050106) 生态十年专项项目“复杂背景下地物信息提取规则集构建”(编号:STSN-01-05)共同资助
【分类号】：Q948;TP79
【正文快照】： 0引言近年来,高光谱遥感技术在湿地植被的精细识别和分类方面的应用逐渐增多。由于高光谱遥感数据具有波段多、信息量大、信息冗余等特点,如何对高光谱数据进行降维处理是国内外学者研究的重点问题之一[1]。Hestir等采用光谱角分类器对美国加州三角洲的入侵水生植物进行了物种

【相似文献】