面向对象方法的时间序列MODIS数据湿地信息提

发布时间：2018-03-03 03:04

本文选题：时间序列　切入点：MODIS　出处：《遥感学报》2017年03期 　论文类型：期刊论文

【摘要】：以洞庭湖流域为研究区,对大范围湿地信息遥感提取方法进行了研究。先基于时间序列MODIS EVI及物候特征参数,通过J-M(Jeffries-Matusita distance)距离分析,构建了MODIS(250 m)最佳时序组合分类数据;其次,通过Johnson指数确定了最佳分割尺度,采用面向对象的遥感分类方法(Random tree分类器)提取了洞庭湖流域的湿地信息,并验证该方法的适用性。研究结果表明,基于时序数据与面向对象的Random tree分类的总体精度和Kappa系数分别为78.84%和0.71,较之基于像元的相同算法的总体分类精度和Kappa系数分别提高了5.79%和0.04。同时,基于面向对象方法的湿地整体的用户精度与生产者精度较基于像元方法分别提高了4.56%和6.21%,可有效提高大区域湿地信息提取的精度。
[Abstract]:Taking the Dongting Lake Basin as the study area, this paper studies the remote sensing extraction method of large-scale wetland information. Firstly, based on the time series MODIS EVI and phenological characteristic parameters, the best time series classification data of MODIS(250 m are constructed by J-M-MjiJeffries-Matusita distance analysis. The optimal segmentation scale is determined by Johnson index, and the wetland information of Dongting Lake basin is extracted by using an object-oriented remote sensing classification method named Random tree classifier. The total accuracy and Kappa coefficient of Random tree classification based on temporal data and object oriented are 78.84% and 0.71 respectively, which are 5.79% and 0.04 higher than that of the same algorithm based on pixel, respectively. The user accuracy and producer precision of the wetland based on object-oriented method are improved by 4.56% and 6.21, respectively, which can effectively improve the precision of wetland information extraction in large area.
【作者单位】：中南大学地球科学与信息物理学院;中南大学空间信息技术与可持续发展研究中心;
【基金】：国家自然科学基金(编号:41171326,40771198) 中南大学中央高校基本科研业务费专项资金(编号:2016zzts087)~~
【分类号】：P237;TP751

【相似文献】