基于福鼎高分五号高光谱遥感数据森林优势树种识别的应用研究  

Application Research on the Identification of Forest Dominant Tree Species Based on Fuding Gaofen-5 Hyperspectral Remote Sensing Datas

作　　者：赵小星 Zhao Xiaoxing(Fujian Institute of Geology Survey and Research,Fuzhou,350013)

机构地区：[1]福建省地质调查研究院,福州350013

出　　处：《福建地质》2023年第2期126-132,共7页Geology of Fujian

基　　金：福建省地质矿产勘查开发局地勘费项目“高光谱遥感数据在自然资源调查中的应用研究”。

摘　　要：通过分析高光谱遥感数据在自然资源调查中的研究现状及目前高光谱遥感数据地物光谱识别的主要技术方法及其精度情况,选取福鼎市一带作为试验区,基于GF5高光谱数据预处理后成果分别通过对数变换后一阶微分法、归一化变换后一阶微分法、主成分分析法、波谱角填图法等处理及分类方法,对竹林、马尾松、杉木、桉树、茶树、草地进行了试验提取。采用林业数据及第三次国土调查数据,结合外业调查进行精度评价,结果表明波谱角填图方法准确率最高。This paper analyzes the research status of hyperspectral remote sensing datas in natural resources investigation and the main techniques and accuracy of surface object spectral recognition in hyperspectral remote sensing data.Fuding city as a test area,based on the pre-processin results of GF5 hyperspectral datas,the first order differential method after logarithmic transformation,the first order differential method after normalized transformation,the principal component analysis method,and the spectral angle mapping method were respectively processed and classified,extracted experimentally the Bamboo forest,masson's pine,Chinese fir,eucalyptus,tea tree,grassland,and then used the forestry datas and the third land survey datas,combined with field survey precision evaluation,the results show that the spectral Angle mapping method is the highest accuracy.

关 键 词：GF5号高光谱 数据变换 SVM 波谱角填图 树种识别 福鼎 

分 类 号：S771.8[农业科学—森林工程]