SAM和决策树结合的Hyperion数据分类方法

A Classification Model for Remote Sensing Images Based on SAM and Decision Tree

在地面实测波谱分析的基础上,采用决策树对波谱角(SAM)分类方法进行改进,自动地进行波谱角阈值选择,提出一种新的基于SAM和决策树相结合的综合分类模型。该模型用于云南鹤庆地区土地覆被信息提取,并与最大似然分类法(M IC)的分类结果进行比较。结果表明,就每一类型而言,SAM结合决策树分类的分类精度较高;最大似然法监督分类总体精度为79.4%,SAM结合决策树分类的综合分类模型总体精度为88.5%,比监督分类精度高9.9%。

To artificially define the threshold of SAM in the classification of hyperspectral remote sensing images often produce errors. In light of this, the paper proposes a new model for automatically classifying remote sensing images using SAM and decision tree and based on the analysis of field spectrum. Applied to the processing of images of Heqing region in Yunnan Province, the model is able to ensure the threshold of SAM and eliminate the disturbance and influence of unwanted objects effectively, so as to improve the classification precision. In comparison with the likelihood classification by field survey data, the classification precision of this model is 9.9% better.