摘要
通过对手写体数字识别技术的研究,本文建立了一个脱机手写体数字识别系统,对手写体数字的识别提出了一些新的思路,并对识别过程中所采用的关键算法进行了阐述。本文提出了二次毛刺去除法对手写体数字图像进行预处理,采用骨架搜索法来提取字符的孔洞特征,使用模板匹配加骨架搜索的方法来确定不同方位的端点,利用模板匹配法来提取字符的横线及竖线特征,采用特征匹配法对字符进行识别。实验证明,本文所建立的基于形体特征的手写体数字识别系统能较全面地反映手写体数字各方面的特征,总的识别率达到95.5%,有着较好的识别性能和十分广泛的应用前景。
Studying on handwritten digital recognition technology,the paper attemped to establish a number of off-line handwriting recognition system which made some new ideas for recognition of handwritten digits,and elaborated the key algorithm used in the process of identification.In this paper,the second glitch removal methods on handwritten digital image is preprocessed using the skeleton search method to extract the pore characteristics of characters,using a template matching plus skeleton search method to determine the endpoints of the different orientations,using template matching method to extract characters’horizontal and vertical line features,using feature matching method for character recognition.Experimental results show that the establishment of this article based on physical characteristics of handwritten numeral recognition system can reflect more fully the various aspects of handwritten digits characteristics,the overall recognition rate of 95.5%,which has a good recognition performance and a very wide range of applications.
出处
《微计算机信息》
2010年第22期187-189,共3页
Control & Automation
关键词
手写体数字
预处理
骨架搜索
模板匹配
特征匹配
handwritten numeral
pretreatment
frame search
template matching
feature matching
