基于Kinect深度相机的实时三维人体动画

J4 ›› 2014, Vol. 36 ›› Issue (08): 1544-1548.

基于Kinect深度相机的实时三维人体动画

陈姝1，梁文章2，伍靓1

(1.湘潭大学信息工程学院，湖南湘潭 411105;2.广西卡斯特动漫有限公司，广西南宁 530003)

收稿日期:2012-11-26 修回日期:2013-03-12 出版日期:2014-08-25 发布日期:2014-08-25
基金资助:
国家自然科学基金资助项目(61100139,61040009)；科技部科技型中小企业技术创新基金资助项目(国科发计［2011］242号)；广西千亿元产业重点攻关工程资助项目(桂科攻1110700613号)

Real-time 3D human motion animation based on Kinect

CHEN Shu1，LIANG Wenzhang2，WU Liang1

(1.College of Information Engineering,Xiangtan University,Xiangtan 411105; 2.Guangxi Cast Animation Co.,Ltd.,Nanning 530003,China)

Received:2012-11-26 Revised:2013-03-12 Online:2014-08-25 Published:2014-08-25

摘要/Abstract

摘要：

研究了一种基于HAnim标准的实时人体三维动画方法，首先对HAnim中人体肢体层次结构进行研究和分析，给出了进行人体动画的坐标变换方法；其次，基于OpenNI对Kinect获取的数据进行重新处理，采用逆运动学计算非根关节旋转矩阵；最后给出了系统流程及具体实验方法，采用OpenGL由实时获取的关节旋转矩阵驱动虚拟人实现了人体动画。实验结果表明，该算法可以较精确地提取人体三维姿态，实时重构人体运动。

关键词: 人体动画, Kinect, 逆运动学, H-Anim

Abstract:

An approach to animating 3D human motion based on HAnim standard is proposed. Firstly, the human body hierarchical structure in HAnim is studied and analyzed, and a coordinate transform strategy is given based on the analysis. Secondly, based on OpenNI, the data received from the Kinect is reprocessed, and the rotational matrix of nonroot joint is obtained by inverse kinematics. Finally, the entire data flow and the concrete implementation are described, and the human body animation is achieved by using OpenGL to retarget the realtime human pose to the virtual human model. The experimental results show that the proposed approach can accurately extract the human pose, and reconstructs realtime human body movement.

Key words: human motion animation;Kinect;inverse kinematics;H-Anim

陈姝1，梁文章2，伍靓1. 基于Kinect深度相机的实时三维人体动画[J]. J4, 2014, 36(08): 1544-1548.

CHEN Shu1，LIANG Wenzhang2，WU Liang1. Real-time 3D human motion animation based on Kinect [J]. J4, 2014, 36(08): 1544-1548.

参考文献［9］

［1］	Wu Ling, Hou Jin, Xu Fang. The realization of 3D human animation by BVH［J］. Journal of Shenyang Ligong University, 2012, 31(1):5256.(in Chinese)
［2］	Lü Zhiguo,Li Yan,He Hangen.Study on modelbased hand posture reconstruction of virtual human［J］. Journal of System Simulation, 2009, 21(3):665667.(in Chinese)
［3］	Chen Shu, Zou Beiji, Peng Xiaoning. 3D virtual human animation based on captured motion data in key video frames［J］. Journal of System Simulation, 2008, 20(21):58685873.(in Chinese)
［4］	Xiao Jun, Zhuang Yueting, Wu Fei. Computer vision and machine learning in 3D human animation:A survey ［J］. Journal of ComputerAided Design & Computer Graphics, 2008, 20(3):281290.(in Chinese)
［5］	Yu Xuesong, Tang Xianglong, Liu Jiafeng, et al. An algorithm for fast tracking of 3D arm motion based on hierarchy joint model ［J］. Chinese High Technology Letters, 2009, 19(6):596602.(in Chinese)
［6］	Chen Jia, Wu Xiaojun. 3D human body shape and motion tracking by LBS and Snake［J］. Journal of ComputerAided Design & Computer Graphics, 2012, 24(3):357371.(in Chinese)
［7］	Chen Shu, Zou Beiji, Peng Xiaoning, et al. Markerless 3D human motion tracking for monocular video sequences ［J］. Journal of ComputerAided Design & Computer Graphics, 2008, 20(8):10471055.(in Chinese)
［8］	Wang Zhaoqi.Study on synthesis of virtual human［J］. Journal of the Graduate School Academia Sinica, 2000, 17 (2):8998.(in Chinese)
［9］	Shu Bo, Mao Tianlu, Wang Zhaoqi. Motion display of virtual human based on D3D［J］. Journal of ComputerAided Design & Computer Graphics, 2004, 16 (11):15051510.(in Chinese)
［10］	Humanoid Animation Working Group. Specification for a standard humanoid［EB/OL］.［20030316］. http://www.hanim.org.
［11］	Wen G J, Wang Z Q, Xia S H, et al. From motion capture data to character animation［C］∥Proc of the ACM Symposium on Virtual Reality Software and Technology, 2006:165168.
	附中文参考文献：
［1］	吴玲, 侯进, 徐芳. BVH文件驱动三维人体动画的实现［J］. 沈阳理工大学学报, 2012, 31(1):5256.
［2］	吕治国, 李焱, 贺汉根. 基于模型的虚拟人手势重构方法研究［J］. 系统仿真学报, 2009, 21(3):665667.
［3］	陈姝, 邹北骥, 彭小宁. 基于关键视频帧跟踪数据的三维虚拟人动画［J］. 系统仿真学报, 2008, 20(21):58685873.
［4］	肖俊, 庄越挺, 吴飞. 计算机视觉与机器学习技术在三维人体动画中的应用综述［J］. 计算机辅助设计与图形学学报, 2008, 20(3):281290.
［5］	于雪松, 唐降龙, 刘家峰, 等. 基于关节层次模型的3D手臂运动快速跟踪算法［J］. 高技术通讯, 2009, 19(6):596602.
［6］	陈加, 吴晓军. 联合LBS和Snake的3D人体外形和运动跟踪方法［J］. 计算机辅助设计与图形学学报, 2012, 24(3):357371.
［7］	邹北骥, 陈姝, 彭小宁, 等. 适用于单目视频的无标记三维人体运动跟踪［J］. 计算机辅助设计与图形学学报, 2008, 20(8):10471055.
［8］	王兆其. 虚拟人合成研究综述［J］. 中国科学院研究生院学报, 2000, 17 (2):8998.
［9］	束搏, 毛天露, 王兆其. 基于D3D 的三维虚拟人运动显示［J］. 计算机辅助设计与图形学学报, 2004, 16(11):15051510.

[1]	刘国华1,2，段建春1. 基于Kinect相机的深度图像空洞修复及超像素分割算法[J]. 计算机工程与科学, 2020, 42(05): 851-858.
[2]	孟勃，刘雪君. 一种自动、快速的Kinect标定方法[J]. J4, 2016, 38(06): 1193-1199.
[3]	孟勃1，曹蕾2. 基于Kinect的机器人辅助超声扫描系统研究[J]. J4, 2016, 38(03): 494-500.
[4]	赵建敏,赵忠鑫,许晓伟. 基于Kinect的拳击虚拟训练系统[J]. J4, 2015, 37(09): 1736-1741.
[5]	曹小峰，程实，陈翔，陈晓红. 基于改进Kinect的点分布模型面部表情特征提取[J]. J4, 2015, 37(04): 802-805.
[6]	卢晓军李焱贺汉根. 一种基于自由空间法的虚拟人行走规划方法[J]. J4, 2005, 27(8): 60-61.