摘要
利用SOC710VP成像光谱仪在湖南省东洞庭湖区域湿地保护核心区获取了典型挺水植物芦苇、湿生植物苔草、泥蒿和栽培植物青菜的成像光谱数据,采用光谱微分技术的一阶导数分析方法和包络线去除方法分析了几种植被类型成像光谱曲线波段特性,提取成像光谱数据"双边"参数和吸收特征,并利用Fisher线性判别函数进行湿地植被类型识别,总分类精度达到87.39%,Kappa系数达到0.831 6。苔草分类后的精度最高,达到92.55%,青菜地的识别精度为92.31%,芦苇居中,识别精度达到86.11%,泥蒿的识别精度为80.65%。结果表明,地面采集的成像光谱数据进行分析得到的植被光谱特征变量具有较好的普适性和可靠性,可以为湿地植被类型的识别提供良好的科学依据。
In order to accurately recognize the vegetation types,spectral information of typical wetland vegetation spectral data was collected by SOC710 VP in the core region of the wetland reserve of east Dongting Lake,Hunan province.The first derivative analysis method and the envelope removal method were adopted to analyze the spectral characteristics of the vegetations in the region.By using Fisher linear discriminant function to identify wetland vegetation types,the total classification accuracy reached 88.59%,Kappa coefficient was 0.8247.The recognition accuracy for Carex was the highest（93.55%）,followed by Brassica chinensis（92.31%）,weeds（86.11%）,and the recognition accuracy of Artemisia annua was low（80.65%）.The results indicated that the spectral characteristics of the vegetation spectra obtained from the imaging spectral data had good universality and reliability,and it could provide a scientific basis for the identification of wetland vegetation.
作者
凌成星
刘华
鞠洪波
张怀清
孙华
由佳
李伟娜
LING Cheng-xing;LIU Hua;JU Hong-bo;ZHANG Huai-qing;SUN Hua;YOU Jia;LI Wei-na(Institute of Forest Resource Information Techniques ,CAF,Beijing 100091 ,China;Research Center of Forestry Remote Sensing ＆ Information Engineering, Central South University Forestry ＆ Technology, Changsha, Hunan 410004, China)
出处
《西北林学院学报》
CSCD
北大核心
2018年第3期208-213,共6页
Journal of Northwest Forestry University
基金
中央级公益性科研院所基本科研业务费专项资金项目(IFRIT201505)
关键词
遥感
成像高光谱数据
光谱分析
湿地植被
植被类型识别
remote sensing
imaging spectrometer data
dpectrum analysis
wetland vegetation
classification of wetland vegetation
