基于多特征融合的票据分类技术及应用  

Bill Classification Technology Based on Multiple Characteristics Fusion and Its Applications

作　　者：陈增照[1] 何秀玲[1] 杨扬[1] 董才林[2] 

机构地区：[1]北京科技大学信息工程学院,北京100083 [2]华中师范大学最优控制与离散数学重点实验室,武汉430079

出　　处：《计算机工程与应用》2006年第9期202-204,共3页Computer Engineering and Applications

基　　金：湖北省科技攻关计划资助项目(编号:2003BDST004)

摘　　要：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。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.

关 键 词：票据分类 多特征融合 知识库 金融票据 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP391.13[自动化与计算机技术—计算机科学与技术]