基于粗糙集分类器设计与应用研究

The classifier designs and application study based on rough set theory

针对从粗糙集能更好的处理不确定性信息的特点,粗糙集理论在遥感影像分类中取得一定成果的基础上,采用了三种粗糙集理论的算法进行粗糙集理论遥感分类的研究,一般粗糙集算法、启发式粗糙集算法和互信息动态粗糙集算法,采用记事本的形式把分类规则和分类结果记录下来,并用混淆矩阵对遥感影像进行了分类精度的评价,分类精度提高了1.1403%,试验证明粗糙集理论算法在遥感影像的分类处理取得良好的自动化分类效果,简化了分类的判别过程,试验结果是可行的。

Characteristic specifically for being able to handle uncertainly information more well than others. Rough set theory has been getting certainly achievement in portrait classification in remote sensing. This paper have adopt three kinds rough set theory al- gorithms to carry out classed research of image mining, the general rough set algorithm, heuristic method of algorithm and dynamic reduct rough set algorithm. This paper adopt text files format to store the classification regulations and classification results. The accuracy of classification on image mining was estimated by using confusion matrix. Classification accuracy has improved 1.1403%. A great deal of experiments have approved that the algorithms of rough set theory can perform well in image mining portrait classification, Test results demonstrated the implement of rough set theory in the remote sensing classification is feasible.