Human Motion Description Based on GPDM and 3D Curve Moment Invariants in Latent Spaces
Motion description and analysis is important for automatically generating new realistic motions in computer animation and virtual environments. In this paper, we propose a new motion analysis method in low-dimensional latent spaces by using Gaussian Process Dynamical Model (GPDM) and 3D curve moment invariants. GPDM is used to mapping the highdimensional motion data into low-dimensional latent space. 3D curve moment invariants are used to describe the features of motion curves learned by GPDM. This method can be used to describe the characteristic of different motion. We verify our method using CMU motion capture database.
Zhenbo Li Jun Yue Hua Li Zetian Fu
College of Information and Electrical Engineering, China Agricultural University , Bei Jing School of Information Science and Engineering, LUDONG University , Yan Tai National Research Center for Intelligent Computing Systems, Institute of Computing Technology, Chine
国际会议
2011亚太信号与信息处理协会年度峰会(APSIPAASC 2011)
西安
英文
1-4
2011-10-18（万方平台首次上网日期，不代表论文的发表时间）