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的人脸识别方法和基于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,在人脸识别率和说话人识别率分别为82.6%和75.9%的情况下,两个通道融合后的识别率达到了92.2%.
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-modal 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%.
出处
《计算机工程与科学》
CSCD
2007年第11期72-75,共4页
Computer Engineering & Science
基金
浙江省自然科学基金资助项目(602060)
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
multi-modal biometrics identification
face recognition
speaker recognition
score-level fusion
