J4 ›› 2007, Vol. 29 ›› Issue (11): 72-75.
• 论文 • 上一篇 下一篇
方志刚[1] 鲍福良[1] 叶伟中[2]
出版日期:
发布日期:
Online:
Published:
摘要:
由于单生物特征认证往往难以满足实际应用的要求,本文在信息融合的基础上提出一个多通道生物特征认证模型,它采用基于PCA的人脸识剐方法和基于MFCC与VQ的说话人识别方法,在分数层使用多层线性分类器实现了人脸和语音的双通道融合。实验结果表明,在人脸识剐率和说话人识别率分剐为82.6%和75.9%的情况下,两个通道融合后的 识剐率达到了92.2%.
关键词: 多通道生物特征认证 人脸识别 说话人识别 分数层整合
Abstract:
Because single biometrics usually can not meet the requirements in practical applications, this article puts forward a model of multi-modal biometrics on the basis of information fusion. Based on the PCA face recognition algorithm and the MFCC and VQ speaker recognition algorithm, it realizes the hi-m odal fusion of face and voice on the score level using a multi-linear classifier. The experimental results show that the recognition rate of the fusion algorithm is 92. 2% while the face recognition algorithm is 82. 6% and the speaker recognition algorithm is 75.9%.
Key words: (multi-modal biometrics identification, face recognition, speaker recognition, score-level fusion)
方志刚[1] 鲍福良[1] 叶伟中[2]. 多通道生物特征认证融合算法[J]. J4, 2007, 29(11): 72-75.
0 / / 推荐
导出引用管理器 EndNote|Ris|BibTeX
链接本文: http://joces.nudt.edu.cn/CN/
http://joces.nudt.edu.cn/CN/Y2007/V29/I11/72