摘要
针对钥匙是保险柜最不保险的根源 ,提出通过识别主人实时面部图像的方法来开启保险柜的新思想 .针对无钥保险柜系统高可靠性和实时性的特点 ,提出并实现了一种人脸识别方法 ,该方法将人脸的主分量特征与人脸的几何特征融合在一起 ,共同构成人脸的识别特征 .鉴于人脸识别问题的特殊性 ,将传统分类方法与人工神经网络方法结合起来 ,构造了一个混合分类器 ,从而极大地提高了识别率 .最后实现了一个无钥保险柜仿真实验系统 .该系统运行良好 ,效果令人满意 .
A safe can be made safe by making it key-free. Such a safe can be opened by recognizing the owner′s face feactures only. A new method is proposed to fuse the principal features with the geometric features of a human face. A hybrid classifier is formed by combining the traditional classification method with an artificial neural algorithm to significantly improve the rafe of recognition. Simulation tests demonstrated satisfactory operation of the key-free safe.
出处
《哈尔滨工业大学学报》
EI
CAS
CSCD
北大核心
2002年第2期270-274,共5页
Journal of Harbin Institute of Technology
关键词
无钥保险柜
人脸识别
主分量特征
混合分类器
钥匙
key-free safe
face recognition
principal component features
hybrid classifer
