南京大学学报（自然科学版）Issue(5)：1058-1067,10.DOI:10.13232/j.cnki.jnju.2015.05.017
基于混合像元分解的 HJ-1卫星时间序列影像城市植被信息提取
Vegetation information extraction based on HJ-1 satellite time-series images by pixel unmixing
摘要
Abstract
Aiming at the case that single remotely sensed image cannot gain phenological character and the problem of mixed pixels among the image during urban vegetation information extraction,this paper tried to obtain the distribution information of urban vegetation based on HJ-1 NDVI time-series data,which can be regarded as simulated hyperspectral data.This information can be got by unmixing the S-Gfiltered NDVI time-series data using unconstrained,semi-constrained and fully constrained linear spectral mixture pixels unmixing algorithm.Taking the downtown and suburbs of Nanjing City as study area,we come to the conclusion that unmixing algorithm of uncon-strained linear spectral mixture model can finely extract the five kinds of vegetation information,including shrub,grassland,coniferous forest,broadleaved deciduous forest,evergreen and deciduous broad-leaved mixed forest.The RMS error is minimum among the three methods,and it is considered to be the best.关键词
类高光谱/线性混合像元分解/城市植被/南京市Key words
Hyperspectral simulated data/linear spectral mixture pixels unmixing/urban vegetation/Nanjing city分类
天文与地球科学引用本文复制引用
黄银友,冯莉,朱榴骏,刘寒,李成蹊..基于混合像元分解的 HJ-1卫星时间序列影像城市植被信息提取[J].南京大学学报（自然科学版）,2015,(5):1058-1067,10.基金项目
国家自然科学基金(41301446) （41301446）
