首页 | 本学科首页   官方微博 | 高级检索  
     检索      

票据OCR系统的设计与实现
引用本文:赵莉,于承新.票据OCR系统的设计与实现[J].山东科学,2007,20(3):33-37.
作者姓名:赵莉  于承新
作者单位:山东建筑大学商学院,山东,济南,250101;山东建筑大学商学院,山东,济南,250101
摘    要:票据OCR系统可广泛适用于银行、税务等行业大量票据表格的自动扫描、识别及电子存储。本文设计的票据OCR系统在图像预处理部分采用了灰度分段线性拉伸、二值化、滤波去噪等处理;在特征提取中设计了字符投影特征和小波变换相结合的特征提取方法;在模板匹配时,利用小波变换的高频系数和低频系数进行了分类多模板匹配。并通过对照实验对设计的有效性进行了证明。

关 键 词:票据OCR系统  图像预处理  特征提取  小波变换  分类多模板匹配
文章编号:1002-4026(2007)03-0033-05
收稿时间:2007-03-04
修稿时间:2007-03-04

Design and Realization of Cheque-Note Optical Character Recognition System
ZHAO Li,YU Cheng-xin.Design and Realization of Cheque-Note Optical Character Recognition System[J].Shandong Science,2007,20(3):33-37.
Authors:ZHAO Li  YU Cheng-xin
Institution:School of Business, Shandong Jianzhu University, Jinan 250101, China
Abstract:The cheque-note Optical Character Recognition system can be applied to the institutions with stacks of cheque notes waiting for being automatically scanned,recognized and stored such as banking,taxation,etc.In the paper,we adopt grayscale stretching,two value conversing,median filter and other arithmetics in image pre-processing,design a method that combines two compensatory characteristics,projection characteristic and wavelet transform,in the part of characteristic extraction,and design assoeted multi-templet with the coefficients of approximation and particular during the templet mathching.The results of our control experiment illustrate the validity and practicability of our design.
Keywords:cheque-note Optical Character Recognition system  image pre-processing  characteristic extraction  wavelet transform  assoeted multi-templet matching
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号