摘要
在研究伪二维隐马尔可夫模型(E-HMM)的基础上,结合优化观测向量,将离散余弦的优化方法加入到特征提取中,增强了系统对噪声、光照以及姿态的鲁棒性,提高了识别的效率。本文首先对二维人脸图像进行建模分析,然后对模型进行训练,最后利用训练好的隐马尔可夫模型完成人脸识别。实验结果表明,该方法识别率高,鲁棒性好,具有较高的实际应用价值。
In this paper,based on the studies of pseudo two-dimensional Hidden Markov Models(E-HMM),combined with the optimization observation vectors,the discrete cosine optimization method was added to the feature extraction,to enhance robustness for system noise,light and posture,and to improve the identification efficiency.In this paper,first,we model and analyze two-dimensional face image;then,train the model;finally,use the trained Hidden Markov Model to complete face recognition.The experimental results show that this method has high recognition rate and high practical value.
出处
《科技通报》
北大核心
2012年第12期137-139,共3页
Bulletin of Science and Technology
关键词
优化观测向量
隐马尔可夫模型
离散余弦变换
optimization observation vectors
Hidden Markov model
discrete cosine transformation