摘要
针对目前研究中对数据手套硬件要求较高的现状,研究汉语手指字母流的识别分类问题,提出一种低要求、高识别精度的数据手套方法。该方法使用BP神经网络结合马尔科夫模型,按照汉语拼音规则进行输入拼音序列概率的估计,输出汉语拼音。仿真实验结果表明,采用该方法能获得91%以上的单个手指字母识别率,并能有效输出汉语拼音。
This paper studies the Chinese finger alphabet flow recognition and classification.As there is a problem that higher hardware requirements on data gloves are needed for the current study,a kind of lower hardware requirements and higher recognition rate method is proposed.This method uses BP neural network and Markov model and estimates the input sequence probability by Chinese alphabet spelling rules to output Chinese alphabet flow.Experimental results show that this method can get more than 91% recognition rate and output alphabet flow effectively,which proves the method is effective.
出处
《计算机工程》
CAS
CSCD
北大核心
2011年第22期168-170,173,共4页
Computer Engineering
关键词
BP神经网络
马尔科夫模型
数据手套
手指字母流分类
在线识别
BP neural network
Markov model
data glove
finger alphabet flow classification
online identification
