夏玉米最佳时序谱段组合识别模式研究  

Identification Model of Optium Time-series Spectral Group of Summer Maize Based on MODIS-EVI Data

作　　者：刘珺[1] 田庆久[2] 

机构地区：[1]太原理工大学矿业工程学院,太原030024 [2]南京大学国际地球系统科学研究所,南京210000

出　　处：《遥感信息》2015年第2期105-110,共6页Remote Sensing Information

基　　金：国家重点基础研究发展计划(2010CB951503);国家科技重大专项(30-Y20A01-9003-12/13)

摘　　要：针对夏玉米难以精确识别和分类时数据冗余的问题,提出夏玉米最佳时序谱段组合识别模式。基于时间序列的MODIS EVI数据,利用马氏距离(Jeffries-Matusita Distance,J-M)和构造的加权平均分离距离(Weighted Average Separability,DWAS)得到夏玉米区别于其他作物的增强型植被指数(Enhanced Vegetation Index,EVI)时序图像,并将其进行组合,从而构建了识别夏玉米的最佳时序谱段组合;并利用其提取了2001年~2010年黄淮海地区的夏玉米。同时利用典型试验样区TM影像分类结果和野外实地采集样地对提取结果进行了验证:典型样区MODIS与TM分类误差最大为11.4%,空间匹配度达到91.29%,基于TM数据提取的典型样区夏玉米种植面积(STM)与基于MODIS数据得到的夏玉米种植面积(SM)的精度均大于70%;地面506个样地检验的混淆矩阵总体精度达到81%。The optium model of time-series spectral group was put forward to solve accurate identification and classification for summer maize.The MODIS EVI data were picked up from time-series MODIS EVI images using J-M distance and weighted average separability distance (DWAS )structured.On the basis of images separated,the model of identification for summer maize was constructed.Then summer maize in Huang-Huai-Hai from 2001 to 2010 were classified based on the classifying model.And the results were verified by that of TM in test regions and samples collected from fields.The maximal error of spatial distribution in test regions between MODIS and TM is 11.4%,the total consistency is 91.29%,the accuracy of area between SM and STM were higher than 70%,the overall accuracy of confusion matrix between ground samples and results derived MODIS is 81%.

关 键 词：夏玉米 MODIS-EVI 时序谱段组合 黄淮海 

分 类 号：S513[农业科学—作物学] TP751[自动化与计算机技术—检测技术与自动化装置]