Image mosaic, calibrate and reflectance inversion calculation have been done about the ALI remote sensing data in the research area, and through analyzing the N-D scatter plot which is calculated by PPI (pixel purity index), sample spectrum curves are obtained and analyzed with spectrum comparison technique. On that basis information of rocks and minerals in the research area are collected by means of decision tree classification. The research shows that the characteristic distinction of various mineral spectrum curves of the sample area from RS image after MNF (minimum noise fraction) transformation is obviously better than that from RS image before MNF transformation. The decision tree classification method based on that can accurately identify different types of rocks.%对研究区ALI数据进行了图像镶嵌、定标与反射率计算,通过PPI运算与N-D散点图三维分析,获取了研究区岩矿样本波谱曲线,并采用波谱比对技术对样本岩矿进行了分析.在此基础上,采用决策树分类方法提取了研究区岩矿信息,研究发现:通过MNF变换,样本区的各类岩矿波谱曲线特征差异明显且远远大于MNF变换前的波谱曲线特征差异,而基于此的决策树分类识别方法能够较为准确地识别各类岩矿.
展开▼
