基于奇异点邻域结构的三维人脸识别方法  

3D face recognition approach based on singular point neighborhood structure

作　　者：袁姮[1] 王志宏[1] 姜文涛[2] YUAN Heng WANG Zhi-hong JIANG Wen-tao(College of Business and Management, Liaoning Technical University, Huludao 125105, China College of Software, Liaoning Technical University, Huludao 125105, China)

机构地区：[1]辽宁工程技术大学工商管理学院,辽宁葫芦岛125105 [2]辽宁工程技术大学软件学院,辽宁葫芦岛125105

出　　处：《控制与决策》2017年第10期1739-1748,共10页Control and Decision

基　　金：国家自然科学基金项目(61172144);辽宁省教育厅科学研究项目(16-1048)

摘　　要：提出一种新的基于奇异点邻域结构的三维人脸识别方法.首先,在人脸纹理图像上分割目标区域,划分特征子区域,提取二维奇异点和奇异点邻域结构;然后,在人脸空间几何信息上标记三维奇异点及其邻域结构,并以奇异点和奇异点邻域结构的三维信息表征人脸特征;最后,采用奇异点邻域结构最近点方法识别人脸身份信息.实验结果表明,所提出方法在三维人脸识别方面具有较高的识别准确率和较好的鲁棒性能.A novel 3D face recognition approach based on singular point neighborhood structure is proposed in this paper. Firstly, by detecting feature sub-region in face texture image, two dimensional singular points and singular point neighborhood structure are extracted. Then three dimensional singular points and its neighborhood structure as facial features are marked in face vertex image. Finally, recognition results are obtained by using the method of nearest point of singular point neighborhood structure. Experiment results show that the proposed approach achieves good recognition results with higher recognition accuracy and good robustness.

关 键 词：三维人脸识别 纹理特征 几何特征 奇异点 奇异点邻域结构 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]