基于高光谱影像的舟曲曲瓦沟树种识别  

The Identification of Tree Types in Quwa of Zhouqu Based on Hyperspectral Data

作　　者：周利鹏[1] 马金辉[1] 

机构地区：[1]兰州大学资源环境学院,甘肃兰州730000

出　　处：《安徽农业科学》2014年第16期5298-5301,5323,共5页Journal of Anhui Agricultural Sciences

基　　金：科技部支撑计划项目(2011BAK12B00)

摘　　要：正确识别舟曲地区沟内森林树种是进行植被保护的基础,也为准确计算泥石流发生的降雨阈值提供参考。利用HYPERION高光谱影像,采用基于纯净像元指数(PPI)的端元提取方法,提取了6类端元,用波谱特征拟合方法(SFF)、波谱角分类(SAM)方法和二进制编码方法(BE)识别出核桃、矮灌木、橡树、刺柏、冷杉和灌木蒿。同时利用基于最小噪声分离(MNF)的最小能量约束法(CEM)分离出各自的分布范围。Identifying the types of trees in forest of Zhouqu correctly is the basis of vegetation protecting and an important reference for calculating the rainfall ihreshold of debris flows accurately.In this paper the HYPERION hyperspectral data was used.Six endmembers were extracted based on pixel purity index （PPI）.The endmembers were indentified as walnut,rabbitbrush,leather oak,juniper,firtree and sagebrush using spectral feature fitting （SFF）,spectral angle mapping （SAM） and binary encoding （BE） methods and their distribution was showed in a map based on constrained energy minimization （CEM） method.

关 键 词：高光谱 树种识别 HYPERION 波谱分析 混合像元分解 

分 类 号：S126[农业科学—农业基础科学]