在Naive Bayes Classifiers模型中,要求父节点下的子节点(特征变量)之间相对独立,然而在现实世界中,特征与特征之间是非独立的、相关的。提出一种预处理方法,实验结果表明,该方法明显地提高了分类精度。
In this paper, a new method, aerial images texture classification based on naive Bayes classifiers, is proposed. It is concluded that the new method performs better in overall classification precision than naive Bayes classifiers on the basis of the experiment results.