| 基于非刚性ICP的三维人脸数据配准算法 |
林源,梁舒,王生进( ) |
| 3-D faces registration via non-rigid ICP |
| Yuan LIN,Shu LIANG,Shengjin WANG() |
| State Key Laboratory of Intelligent Technology and Systems, Department of Electronic Engineering, Tsinghua University, Beijing 100084, China |
摘要:
| |||
| 文章导读 | |||
| 摘要三维人脸扫描数据的配准是三维人脸统计学习的基础和前提。为了解决离散点群数据的精确配准问题,该文提出一种融合弹性形变的非刚性最近点迭代(ICP)三维人脸数据配准算法。该算法利用稀疏的特征点逐步形变一个高分辨率的三维模板网格(模板三维人脸)以逼近三维人脸扫描数据(目标三维人脸)。此形变过程可以建模为一个关于模板人脸上每个顶点的仿射变换的优化问题。此外,该算法通过引入弹性形变在不损失配准精度的前提下加速了配准过程。在此基础上,该文提出一种融合弹性形变和三维配准的三维人脸的头部缺失数据补齐算法,使得三维人脸模型更具有真实感和完整性。实验表明,该算法比一种目前较为广泛应用的人脸头部补齐算法更具鲁棒性。 | |||
| 关键词 ：三维扫描,三维人脸,非刚性配准,最近点迭代法 | |||
| Abstract：3-D scanned face registration is the foundation and the prerequisite for 3-D faces statistical learning. To solve the problem of registering these digital points accurately, a 3-D faces registration method was developed using a non-rigid iterative closet point (ICP) algorithm combined with elastic deformation. By using the sparse 3-D feature vertices, the method fits high-resolution 3-D template meshes (3-D template faces) to 3-D face range scans (3-D target faces) through deformations step by step. This deformation process can be formulated as an optimization problem for the affine transformation at each vertex of the template face. In addition, this algorithm accelerates the registration process without sacrificing the registration accuracy by introducing elastic deformation. On this basis, a head missing data completion method for 3-D faces, combining the elastic deformation and the proposed 3-D registration, was developed to enhance the reality and the completeness of the 3-D face models. The experimental results show that this method is more robust than a widely used face completion method. | |||
| Key words：3-D scanning3-D facenon-rigid registrationICP | |||
| 收稿日期: 2012-03-19 出版日期: 2015-04-16 | |||
| |||
| 基金资助:国家自然科学基金资助项目(61071135);国家“八六三”高技术项目(2012AA011004) | |||
| 引用本文: |
| 林源, 梁舒, 王生进. 基于非刚性ICP的三维人脸数据配准算法[J]. 清华大学学报（自然科学版）, 2014, 54(3): 334-340. Yuan LIN, Shu LIANG, Shengjin WANG. 3-D faces registration via non-rigid ICP. Journal of Tsinghua University(Science and Technology), 2014, 54(3): 334-340. |
| 链接本文: |
| http://jst.tsinghuajournals.com/CN/或 http://jst.tsinghuajournals.com/CN/Y2014/V54/I3/334 |
图表:
 | |||||||||||||||||||||||||||||||||||
| 本文所提三维人脸配准算法流程图 | |||||||||||||||||||||||||||||||||||
 | |||||||||||||||||||||||||||||||||||
| 模板人脸模型及面部特征点 | |||||||||||||||||||||||||||||||||||
 | |||||||||||||||||||||||||||||||||||
| 头部补齐算法所用的三维模板头 | |||||||||||||||||||||||||||||||||||
 | |||||||||||||||||||||||||||||||||||
| 标准纹理图及头部纹理图 | |||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||
| 非刚性ICP配准时所用5组参数 | |||||||||||||||||||||||||||||||||||
 | |||||||||||||||||||||||||||||||||||
| BUJT-3D数据库中3个三维人脸配准结果 | |||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||
| 模板人脸与目标人脸之间的相对距离 | |||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||
| 形变模板人脸与目标人脸之间的相对误差 | |||||||||||||||||||||||||||||||||||
 | |||||||||||||||||||||||||||||||||||
| BUJT-3D数据库中全部人脸的三维配准结果 | |||||||||||||||||||||||||||||||||||
 | |||||||||||||||||||||||||||||||||||
| 改进前后的非刚性ICP的配准时间对比图 | |||||||||||||||||||||||||||||||||||
 | |||||||||||||||||||||||||||||||||||
| 本文头部补齐结果与文[6]头部补齐结果对比图 | |||||||||||||||||||||||||||||||||||
 | |||||||||||||||||||||||||||||||||||
| 纹理贴图后的三维头部补齐形状模型 | |||||||||||||||||||||||||||||||||||
参考文献:
| [1] | Spreeuwers L. Fast and accurate 3d face recognition[J]. International Journal of Computer Vision, 2011, 93(3): 389-414. |
| [2] | Tena J R, Dela Torre F, Matthews I. Interactive region-based linear 3d face models[J]. ACM Transactions on Graphics, 2011, 30(4): 76. |
| [3] | Besl P J, McKay N D. Method for registration of 3-D shapes [C]// Robotics-DL Tentative. International Society for Optics and Photonics, 1992: 586-606. |
| [4] | 罗先波, 钟约先, 李仁举. 三维扫描系统中的数据配准技术[J]. 清华大学学报: 自然科学版, 2004, 44(8): 1104-1106. LUO Xianbo, ZHONG Yuexian, LI Renju. Data registration in 3-D scanning system[J] J Tsinghua Univ: Sci & Technol, 2004, 44(8): 1104-1106.(in Chinese) |
| [5] | Allen B, Curless B, PopovićZ. The space of human body shapes: Reconstruction and parameterization from range scans[J]. ACM Transactions on Graphics, 2003, 22(3): 587-594. |
| [6] | Amberg B, Romdhani S, Vetter T. Optimal step nonrigid icp algorithms for surface registration [C]// Proc IEEE CVPR'07. Piscataway, NJ: IEEE Press, 2007: 1-8. |
| [7] | Blanz V, Vetter T. A morphable model for the synthesis of 3D faces [C]// Proc ACM SIGGRAPH. USA: ACM Press, 1999: 187-194. |
| [8] | 丁镠, 丁晓青, 方驰. L1约束的三维人脸稀疏重建[J]. 清华大学学报: 自然科学版, 2012, 52(5): 581-585. DING Liu, DING Xiaoqing, FANG Chi. L1 constrainted 3-D face sparse reconstruction[J] J Tsinghua Univ: Sci & Technol, 2012, 52(5): 581-585.(in Chinese) |
| [9] | Zhu C, Byrd R H, Lu P, et al.Algorithm 778: L-BFGS-B: Fortran subroutines for large-scale bound-constrained optimization[J]. ACM Transactions on Mathematical Software, 1997, 23(4): 550-560. |
| [10] | Desbrun M, Meyer M,Schröder P.et al.Implicit fairing of irregular meshes using diffusion and curvature flow [C]// Proc ACM SIGGRAPH. USA: ACM Press, 1999: 317-324. |
| [11] | Lin Y, Wang S, Lin Q,et al.Face swapping under large pose variations: A 3D model based approach [C]// Proc IEEE ICME 2012. Piscataway, NJ: IEEE Press, 2012: 333-338. |
| [12] | ZHU Jianke, Gool L V, Hoi C H S. Unsupervised face alignment by robust nonrigid mapping [C]// Proc IEEE ICCV 2009. Piscataway, NJ: IEEE Press, 2009: 1265-1272. |
| [13] | Beijing University of Technology. The BJUT-3D large-scale Chinese face database [EB/OL]. [2012-03-15]. http://www.bjut.edu.cn/sci/multimedia/mul-lab/3dface/home.htm. |
相关文章:
|
