基于信息融合的人脸识别方法  

Face Recognition Based on Information Fusion

作　　者：张芝华 徐丽萍[1] 李美莲[1] ZHANG Zhi-hua;XU Li-ping;LI Mei-lian(School of Electrical and Electronic Engineering,Anhui Sanlian University,Hefei,230601,Anhui)

机构地区：[1]安徽三联学院电子电气工程学院,安徽合肥230601

出　　处：《蚌埠学院学报》2019年第2期81-84,共4页Journal of Bengbu University

基　　金：安徽省教育厅高校自然科学重点项目(KJ2018A06601);安徽三联学院重点科研项目(KJZD2018001);安徽三联学院校级科研项目(KJB2019001)

摘　　要：三维人脸数据对光照、姿态及表情变化具有较强的鲁棒性,而二维人脸图像纹理特征的高鉴别性又是三维人脸图像所不具备的。为了更好地将二维和三维人脸信息有机结合,将二维人脸图像特征和三维人脸图像特征进行融合并用复数表示,然后利用复数域流形学习算法C-LOGMAP进行降维。在Texas三维人脸数据库上验证了算法的有效性,实验结果表明该算法识别率高于单一图像信息的识别率。Three-dimensional face data is robust to illumination,pose and expression changes,while the high discriminability of texture features in two-dimensional face images is not available in three-dimensional face images.In order to better integrate two-dimensional and three-dimensional face information,it fused two-dimensional face image features and three-dimensional face image features in the paper and used complex representation,then used manifold learning algorithm C-LOGMAP in complex domain to reduce dimensionality.The validity of the algorithm was validated on Texas three-dimensional face database.The experimental results showed that the recognition rate of the algorithm was higher than that of single image information.

关 键 词：人脸识别 复数域 信息融合 复数域流形学习算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]