摘要
论文对研究区ALI数据进行了纯净像元提取,在此基础上,对纯净像元进行N-D散点图分析,选择出不同端元并进行归类分析,作为后期分类识别的样本。这里采用决策树方法对研究区岩矿进行识别,研究发现:样本区在MNF变换后图像上的波谱(前几个波段)可分性远远大于其在变换前图像上的波谱可分性,基于此的决策树分类方法能够识别出岩矿。
The pure pixel was extracted by using ALI data in the research area. Through analyzing the N - dimensional scatter plot of pure pixel, different end members were selected and classified, these data can be used as reference samples in later stage data processing. The decision tree method was used to discriminate rocks and minerals, it is shown that the partibility of reference sample' s spectrum from RS image after MNF transformation is better than that from RS image before MNF transformation. The method of decision tree classification based on above can discriminate rocks and minerals.
出处
《地质与勘探》
CAS
CSCD
北大核心
2009年第4期456-461,共6页
Geology and Exploration
基金
成都理工大学青年基金(编号:2006QJ17)
科技部科技创新基金项目(编号:04026215100796)