多源数据的土地覆被样本自动提取

发布时间：2018-03-18 07:19

本文选题：自动化　切入点：样本提取　出处：《遥感学报》2017年05期 　论文类型：期刊论文

【摘要】：随着遥感数据获取能力的不断增强,自动化程度已经成为大尺度遥感土地覆被分类面临的关键问题。然而,现有训练样本的人工选取方法成为制约土地覆被分类自动化的瓶颈。本文以河南、贵州两省为研究区,提出一种基于多源数据的土地覆被样本自动提取方法,以构建适用于大尺度的土地覆被自动分类。首先,以2010年1∶10万土地利用数据CHINALC和30 m分辨率全球土地覆被数据Globle Land30为样本数据源;然后,利用空间一致性分析及异质性分析确定样本初选区域;最后,通过样本提纯去除无效样本。结果表明:(1)应用多源数据的土地覆被样本自动提取方法获得的分类产品总体分类精度高于人工样本提取方法制作的全球土地覆被产品MCD12Q1。(2)与单源样本自动提取方法相比,应用多源数据的土地覆被样本自动提取方法,可获得更好的分类稳定性。综上,多源数据的土地覆被样本自动提取方法可在保证精度的同时,提升土地覆被分类的自动化程度。
[Abstract]:With the increasing ability of remote sensing data acquisition, the degree of automation has become a key issue in large-scale remote sensing land cover classification. The manual selection of training samples has become a bottleneck restricting the automation of land cover classification. In this paper, a method of automatic extraction of land cover samples based on multi-source data is proposed in Henan and Guizhou provinces. In order to construct an automatic classification for large-scale land cover. First, take 1:10 land use data (CHINALC) and 30m resolution global land cover data (Globle Land30) as sample data sources; then, Spatial consistency analysis and heterogeneity analysis are used to determine the primary area of the sample. The result shows that the classification precision of the classification products obtained by the automatic extraction method of land cover samples based on multi-source data is higher than that of the global land cover production made by artificial sample extraction method. Product MCD12Q1.2) compared with single source sample automatic extraction method, The method of automatic extraction of land cover samples from multi-source data can obtain better classification stability. In summary, the automatic extraction method of land cover samples from multi-source data can improve the automation degree of land cover classification while ensuring the accuracy.
【作者单位】：河南大学环境与规划学院;河南大学计算机与信息工程学院;防灾科技学院;中国科学院地理科学与资源研究所;
【基金】：国家重点研发计划项目(编号:2017YFD0300400) 河南省高等学校重点科研项目(编号:16A520081) 中国科学院战略性先导科技专项(编号:XDA05050000)~~
【分类号】：P237;TP79

【相似文献】