空间遥感数据的多粒度标记分类

摘要

以空间遥感数据为研究对象,研究多粒度标记数据的分类问题.首先,根据地表物的不同物理含义,可以得到一个多粒度的决策表.然后,通过知识约简发现隐藏在多粒度决策表中的if-then分类规则.最后,根据实际的遥感数据给出分类的精度.%This paper studied the discovery of classification rules for remote sensing data with multi-granular labels. According to the different physical meaning of land cover, we first obtained a multi-scale decision table, and then discovered the classification rules hided in the multi-scale decision table. Finally, we analyzed the accuracy of the classification.