基于多特征融合的票据分类技术及应用

对票据进行高速、准确的分类是实现票据自动处理的关键。论文分析了各种票据分类方法的优缺点,并根据金融、税务等行业所使用的票据特点,提出基于多特征融合的快速分类方法,可以将票据的几何尺寸、分类条码、标题、特征字符串、特征图像块、特征框线等特征有机地融合并进行分类,提高了票据分类的速度与准确性。利用该技术设计的银行票据处理系统的实际应用说明了该方法的有效性。

Bill Classification Technology Based on Multiple Characteristics Fusion and Its Applications

To quickly and accurately classify bill is a key to automatic bill processing.This paper analyzes the strong and weak points of various existing bill classification methods.Based on tbe distinguished features of the bill used in finance,tax and other industries,we have proposed a new method that can quickly classifies a bill according to the combination of its multiple characteristics,bill size,barcode, header,characteristic string,characteristic image,and characteristic pattern,etc.The new method increases the speed and the accuracy of bill classification.Its effectiveness is demonstrated by its application in the bank slip processing system.