基于不同抽样方法的遥感面积测量方法研究  

RESEARCHES ON REMOTE SENSING AREA MEASUREMENT BASED ON DIFFERENT SAMPLING METHODS

作　　者：胡潭高[1] 张锦水[1] 潘耀忠[1] 宋国宝[1] 董燕生[1] 贾斌[1] 

机构地区：[1]地表过程与资源生态国家重点实验室(北京师范大学),北京100875

出　　处：《国土资源遥感》2008年第3期37-41,I0001,共6页Remote Sensing for Land & Resources

基　　金：国家高技术研究发展计划项目(批准号:2006AA1201032006AA120101);教育部新世纪人才支持计划资助

摘　　要：众多研究结果表明,遥感和抽样技术相结合可以有效地进行地物面积的测量。目前,随机抽样、系统抽样和分层抽样方式在遥感抽样调查技术领域应用比较广泛。本文以遥感图像为基础,从不同角度对随机抽样、系统抽样及分层抽样(包括等样本量、等面积、等丰度抽样)进行了有益探讨,分析发现:对于同一地物,从平均误差百分比、标准差和极差3个角度分析,随机和系统抽样反推得到的总量精度都低于分层抽样精度;对于不同的地物类型,利用3种分层抽样方法反推的结果与地物所占百分比成正相关,地物所占的百分比越大,反推的结果越好;等样本量、等面积、等丰度分层抽样从平均误差百分比、标准差和极差3个角度分析各有优势,跟地物所占的百分比也有密切关系。It is held that the combination of remote sensing and sampling technology can be effectively used in area measurement. At present, the methods of random sampling, systematic sampling and stratified sampling are widely used in remote sensing sampling. Based on remote sensing images, this paper has discussed from different angles the random sampling, systematic sampling, and stratified sampling that includes equal -count, equal -area and e- qual - dist. It is found that ： （ 1 ） for the same features, from three indicators （the average error percentage, stand- ard deviation and difference）, the precision of stratified sampling is better than that of random sampling and systematic sampling, （2） for different features, the use of three stratified sampling method and the results obtained are proportional to the percentage of features, i. e. , the greater the percentage of features, the better the results, and （3） from three indicators （the average error percentage, standard deviation and difference）, the three kinds of stratified sampling methods （ equal - count, equal - area and equal - dist） have their respective own merits, and the precision is also closely related to the percentage of features.

关 键 词：抽样 遥感图像 精度 稳定性 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]