基于图像识别技术的手写数字识别方法  

Handwritten Digit Recognition Based on Image Recognition System

作　　者：吴忠[1,2] 朱国龙[1,2] 黄葛峰[1,2] 吴建国[1,2] 

机构地区：[1]安徽大学计算机科学与技术学院,安徽合肥230039 [2]安徽大学软件学院,安徽合肥230039

出　　处：《计算机技术与发展》2011年第12期48-51,共4页Computer Technology and Development

基　　金：安徽省高校省级重点自然科学基金(kj2010A023);安徽大学青年科学研究基金(2009QN0019B)

摘　　要：随着世界经济越来越发达,各国经济交流日益加深,人们每天要处理大量的票据。手写数字在这个领域是必不缺少的,例如人们要处理许多支票、发票、货物单等数据,这些都要大量与数据打交道。传统的方法是利用光学的光电变换原理对数字进行识别,该方法对环境要求比较高,成本较高,识别率和实时性比较低。文中提出基于图像识别的对0至9手写数字识别技术。实验表明,对2,3,4,5,8,9字符准确识别,而对1,7,0,6,数字基本识别无误,该方法提高了对数字识别率和实时性。As the world economy is more developed,countries deepening economic exchanges,people has to deal with a lot of notes each day. Handwritten digits in this field is necessary ,for example, to deal with a lot of people check,invoice,goods such as single data,which have a lot to deal with data. The traditional method is to use the optical principle of the photoelectric conversion to identify digit,it is relatively high for environrncntal requirements,high cost,low recognition rate and real-time. In this paper,handwritten numeral recognition technology from 0 to 9 is proposed based on image recognition. Experiments show that accurately identify 2,3,4,5,8,9 characters, while the 1,7,0,6, basic identification number is correct, this method improves the recognition rate of digital and real-time.

关 键 词：数字识别 图像识别 手写数字 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]