基于Dempster-Shafer 证据理论的矿山遥感分类

摘要

The paper analyzed the characteristics of mines,using the confidence interval to represent the verdict through Dempster-Shafter evidence theory. The paper calculated the uncertainty of land type distribution map and classifies the land mines on the base of the largest division in CBV pixel attribution. The results showed that D-S evidence theory used to mine land classification can get the higher accuracy than the maximum likelihood method.%本文分析了矿山用地的特点，利用Dempster-Shafter证据理论用信任区间表示判决结果的特点，计算地类不确定性分布图，以CBV为最大划分像元归属类别原则对遥感图像上的矿山用地进行分类。实验结果表明， D-S证据理论用于矿山用地分类的精度高于最大似然方法的精度。