摘要
脱机手写字符自动识别是计算机光学字符识别 (OCR)领域的一个活跃课题 ,有着十分广泛的应用前景 .文中提出了基于BP神经网络的脱机手写中英文和数字混合字符集的识别方法 ,给出一种特征提取方法 ,通过实验说明如何选取网络隐含层神经元个数 ,以及如何选取网络连接权值的初值 .对由不同人手写的中英文字符的混合字符集做识别实验 ,结果表明文中所设计的神经网络分类器 ,不仅能保证识别精度和识别速度 ,而且能有效的识别混合字符集 .
The automatic recognition of off-line handwriting character is an active subject in the area of computer optical character recognition (OCR),which has a wide range of potential applications.An off-line handwriting Chinese character,English character and digital recognition method is put forward based on the BP neural network.Also discussed are the problem of feature extraction,the problem of determining the number of hidden layer′s neural nodes,and the problem of initialization of connection weight,Experiments have been conducted for an off-line handwriting character set.The recognition results show that compared with other recognition method.this designed can recognize Chinese character,English character and digital with same excellence recognition rate and speed.
出处
《山东大学学报(理学版)》
CAS
CSCD
北大核心
2001年第1期50-55,共6页
Journal of Shandong University(Natural Science)
关键词
神经网络
模式识别
OCR
特征提取
neural network
pattern recognition
OCR
feature extraction