使用RLE方法，能够快速提取印刷品缺陷的多种特征。针对不同缺陷特征，使用基于径向基神经网络的方法，依据缺陷目标的特征，对缺陷进行了分类。实验结果显示，径向基神经网络通过对训练样本的训练，对测试样本能够达到良好的分类效果。

Many kinds of characteristics of defects of printed matter can be extracted by RLE method. According to different characteristics of defects, the defects were classified using radial neural network method. Experimental results showed that radial neural network method can obtained good classification effect through training of specimen.