一种人脸图像的匹配识别方法  

Way of matching and identifying the image of face

作　　者：胡迎春[1] 张增芳[1] 梁数 

机构地区：[1]广西工学院计算机工程系,广西柳州545006 [2]柳州经济信息中心,广西柳州545001

出　　处：《计算机工程与设计》2003年第2期44-45,56,共3页Computer Engineering and Design

基　　金：广西自然科学基金资助项目(桂科自0146007)

摘　　要：为进行人脸图像的识别,需要提取面部图像的特征点,通过提取鼻尖、眼角、鼻角、嘴角等特征点,并且以鼻尖作为基准点,分别与两只眼睛的左右眼角以及左右两个鼻角和两个嘴角等构成11维特征向量,作为标准人脸特征向量存入数据库。在分析某人脸图像时,按上述方法提取特征点构成11维特征向量,并与数据库中的特征向量进行对照匹配,当数据接近或匹配时,则认为是同一个人。实验证明,该系统的识别率达到设计要求。In order to identify the image of face, it introduced all dimension eigenvectors which constituted of some feature points such as nose, eyes, mouth and their edges, etc. All these features are based on the top of nose. A same man was thanght if the feature points which be picked up from the face on the way mentioned above can match the 11 dimension eigenvectors which be put into the database when analyzed the image efface. Experimentations have proved that this system was satisfying of the rate of identification.

关 键 词：人脸图像 匹配识别 图像识别 自动识别 特征向量 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]