基于频率特征提取的脱机手写字符的快速识别  

Feature extraction based on frequency of fast identification of offline handwritten letters

作　　者：朱孝山 

机构地区：[1]山西大学商务学院信息中心,山西太原030013 [2]中国科学院心理研究所,北京100101

出　　处：《微型机与应用》2015年第8期56-59,共4页Microcomputer & Its Applications

摘　　要：针对目前脱机手写字符识别计算量大,对字体字形都有一定的要求,提出了主要以字符矩阵中01变换频率为基础,从字符的整体和局部特征出发进行分析识别的算法。此算法缩减了计算量并对所需局部信息进行放大,在一定程度上避免了传统采用分类器方法的错误传导,提高了字符的识别率,易于实现移植和扩展。At present the amount of calculation of handwritten character offline recognition is large, and there are a lot of requirements in font glyph. The algorithm proposed in this paper is mainly based on 01 transformation frequency of the character matrix, starting from the overall and local characteristics of the character to analyze and recognize. This algorithm reduces the amount of calculation, enlarges local information needed, to a certain extent avoids the traditional classifier methods of transmission error, improves the recognition rate of the characters, and is easy to realize transplantation and extension.

关 键 词：字符识别 脱机识别 手写字符 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]