单训练样本视频监控俯视人脸识别研究  

Video surveillance face recognition by single training sample

作　　者：胡晓[1] 廖启欣[1] 彭绍湖[1] 

机构地区：[1]广州大学机械与电气工程学院,广东广州510006

出　　处：《广州大学学报（自然科学版）》2015年第5期67-70,共4页Journal of Guangzhou University:Natural Science Edition

基　　金：广东省自然科学基金资助项目(S2013010013511);广州市科技计划资助项目(2014J4100127)

摘　　要：近年来视频监控已普遍应用于各行各业,因此基于监控视频人脸识别也成为了智能监控系统中重要的研究领域.然而,由于监控视频人脸通常是非正面人脸,传统性能优良算法应用于视频人脸识别时,其性能也明显降低.同时,单张训练人脸问题在监控视频人脸检测和识别是一个普遍问题.因此为了能有效地提高单训练多姿态人脸识别的正确识别率,文章提出了一种基于三维建模技术的人脸识别算法.该算法先由一张二维高清正面人脸生成一个三维人脸模型,然后再进一步在该三维人脸空间里产生多种姿态的人脸模型,并由此获得多张相应姿态下的二维虚拟人脸,最后利用原始正面样本和所得到的虚拟人脸来构筑训练人脸库.该算法用SCface视频监控人脸库中加以验证,与传统的PCA和LDA算法相比,该算法对监控视频人脸的识别率提高了13%.由此表明,文章介绍的算法是一种有效的人脸识别算法,能有效地提高对俯视人脸的识别率.Video surveillance has more and more been applied in recent years for security,video-based face recognition therefore became an important task in intelligence monitoring system.However,among these cap-tured video faces there are many non-frontal faces.As a result the art of state algorithms would become worse. On the other hand,only one training sample could usually be got.In order to effectively improve the correct rec-ognition rate of multi-pose face recognition with single frontal training sample,this paper proposed a face recog-nition algorithm based on 3D modelling.In the proposed algorithm,firstly a 2D frontal face with high-resolution was taken to build a 3D face model,and then several virtual faces with different poses were produced from the 3D face model.At last,both the original frontal face image and virtual face images were put into gallery set. The algorithm was evaluated on SCface database using traditional PCA and LDA methods.The result showed that the proposed approach could effectively improve recognition rate of looking-down faces.

关 键 词：三维建模 单训练样本 视频监控 PCA LDA 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]