Method of missing data recovery
for acquiring accurate hand movements
LI Yi1, LU Guang-ming1,2, JIN Shuai1,2, LUO Jian-xun1,2,CHEN Wei-dong1, ZHENG Xiao-xiang1
1. Qiushi Academy for Advanced Studies, Zhejiang University, Hangzhou 310027, China;
2. College of Computer Science and Technology, Zhejiang University, Hangzhou 310027, China
Using optical motion capture system for acquiring accurate hand movements data and gesture information can be effective for neural decoding and related applications. However, data missing often occurs and affects the accuracy and efficiency of neural decoding results. A method based on principal component analysis was proposed for missing data recovery and rebuild. This method takes benefit of expectation maximization (EM)-like algorithm to increase the accuracy of recovery data results by mapping between original data space and principal component space iteratively for principal component space construction and refinement. To evaluate this method, four different experiments were involved, namely, missing frame length, missing dimension number, cyclic feature and redundant data. Recovered data results were compared with those of the cubic spline interpolation and one iteration interpolation methods. The experimental results show that, this method is applicable to missing frame length less than 350, missing dimension number smaller than 13. The cyclic feature of hand movements contributes to the increased accuracy of recovered data, and the redundant data is also helpful in reducing recovery data errors. When comparing the data results with those of the cubic interpolation and one iteration interpolation methods, the average error of this method was less than 10 mm, only 50% or less of the other two.
LI Yi, LU Guang-ming, JIN Shuai, LUO Jian-xun,CHEN Wei-dong, ZHENG Xi. Method of missing data recovery
for acquiring accurate hand movements. J4, 2013, 47(6): 925-933.
［1］ VELLISTE M, PEREL S, SPALDING M C, et al. Cortical control of a prosthetic arm for self-feeding［J］. Nature, 2008, 453(7198): 1098-1101.
［2］ HOCHBERG L R, BACHER D, JAROSIEWICZ B, et al. Reach and grasp by people with tetraplegia using a neurally controlled robotic arm［J］. Nature, 2012, 485(7398): 372-375.
［3］ LU Guang-ming, LI Yi, JIN Shuai, et al. A real-time motion capture framework for synchronized neural decoding［C］∥ Proceedings of ISVRI 2011. Singapore: IEEE, 2011: 305-310.
［4］ LIU G, MCMILLAN L. Estimation of missing markers in human motion capture［J］. The Visual Computer, 2006, 22(9): 721-728.
［5］ HERDA L, FUA P, NKERS R P, et al. Skeleton-based motion capture for robust reconstruction of human motion［C］∥ Proceedings of Computer Animation 2000. USA: IEEE, 2000: 77-83.
［6］ KIRK A G, O’BRIEN J F, FORSYTH D A. Skeletal parameter estimation from optical motion capture data［C］∥ Proceedings of CVPR 2005. USA: IEEE, 2005: 782-788.
［7］ ZORDAN V B, VAN DER HORST N C, Mapping optical motion capture data to skeletal motion using a physical model［C］∥ Proceedings of SCA 2003. USA: ACM, 2003: 245-250.
［8］ TAYLOR G W, HINTON G E, ROWEIS S T, Modeling Human Motion Using Binary Latent Variables［C］∥ Advances in Neural Information Processing Systems. USA: MIT Press, 2007: 1345-1352.
［9］ PARK S I, HODGINS J K. Capturing and animating skin deformation in human motion［J］. ACM Transactions on Graphics, 2006, 25(3): 881-889.
［10］ HSU E, GENTRY S, POPOVI J. Example-based control of human motion［C］∥ Proceedings of SCA 2004. USA: ACM, 2004: 69-77.
［11］ CHAI J, HODGINS J K. Performance animation from low-dimensional control signals［J］. ACM Transactions on Graphics, 2005, 24(3): 686-696.
［12］ DORFM K, DORFMULLER-ULHAAS K. Robust optical user motion tracking using a Kalman filter［R］. Augsburg, Germany： Department of Computer, University of Augsburg, 2003.
［13］ ARISTIDOU A, CAMERON J, LASENBY J, Real-time estimation of missing markers in human motion capture［C］∥ Proceedings of ICBBE 2008. China: IEEE, 2008: 238-247.
