摘要
本文提出了一种基于数字骨架结构特征的手写数字识别方法。该方法把细化等预处理后的数字的骨架,通过搜索匹配,提取出数字骨架的端点和交叉点等结构特征,然后再根据端点的位置特征来进行数字识别。本算法容易实现,且具有较高的可靠性和识别率,能够满足手写识别应用的要求。该方法已在表格自动录入和识别系统中得到成功应用。
In this paper, a new method based on skeleton structure features for recognition of handwritten numerals is proposed. Use the numerals skeleton after thinning and other preprocessing, and according search matching, extracted digital skeleton endpoint and cross-point structure features and then basis the location characteristics of the endpoint far the digital identification. The algorithm is easy to implement, has a high reliability and recognition rate, can meet the handwriting recognition applications. This method has been successfully applied in forms automatic entry and identification system.
出处
《微计算机信息》
2010年第16期191-192,165,共3页
Control & Automation
关键词
细化
手写数字识别
结构特征
Thinning
Handwritten numeral recognition
Structural features
