适用于银行票据手写数字串切分的滴水算法  

Segmentation Algorithm for Unconstrained Handwritten Numeral Strings in Bank Check Reader System

作　　者：张闯[1] 蔺志青[1] 肖波[1] 郭军[2] 

机构地区：[1]北京邮电大学电信工程学院,北京100876 [2]北京邮电大学信息工程学院,北京100876

出　　处：《北京邮电大学学报》2006年第1期13-16,共4页Journal of Beijing University of Posts and Telecommunications

基　　金：教育部科学技术研究重点项目(02029);教育部跨世纪人才基金项目

摘　　要：在连通域提取与属性判别的基础上,得到不固定长度的粘连字串,利用波形分析的方法对字串个数进行判断,用具有指导信息的改进滴水切分算法解决了覆盖、粘连等非约束不确定位数的手写数字串切分问题.该切分算法在实际的银行票据自动识别系统中取得了实用化的效果.How to improve the strings segmentation process was investigated. The components connected were detected based on isolated components analysis, and the length estimation of the components using the feature of lower contours was discussed, the problem of segmenting overlapping or touching characters settled by the improved ＂drop-falling＂ algorithm was also studied in details. This method has been incorporated into a complete bank check character recognition system. It shows promising results.

关 键 词：票据识别 数字切分 字串个数判断 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]