摘要
在考虑刻画监测区域样地的遥感和G IS因子间存在多重相关性对主分量分析造成危害的基础上,通过变量选择,确定刻画监测区域样地的主要遥感和G IS因子。根据这些因子进行主分量分析,研究监测区域样地的分布及异常样地的探测。比较考虑多重相关性前后所得结果与监测区域样地的实际分布,系统研究了多重相关性对主分量分析的危害及如何利用主分量分析了解监测区域样地的分布状况。
In this paper, on the basis of thinking of the harm to principal component analysis from the multi correlation between the factors of remote sensing and GIS depicting the sample plot in the monitoring area, through the selection of variable screening, the main factors of remote sensing and GIS depicting sample plot is determined in the monitoring area. The distribution of sample plots and the detection of abnormal sample plots in monitoring area are studied by means of principal component analysis according to these factors. Comparing the results of thinking of the multi-correlation before and latter and the practical distribution of sample plots in the monitoring area, the harm from multi-correlation to principal component analysis and how to using principal component analysis to comprehend the distribution of sample plot in monitoring area are studied systematically.
出处
《遥感技术与应用》
CSCD
2006年第1期1-5,共5页
Remote Sensing Technology and Application
基金
国家自然科学基金项目(30371159)
关键词
主分量分析
特征值
多重相关性
异常样地
Principal component analysis, Eigenvalue, Multi-correlation, Abnormal sample plot