Yoshikawa Naoya, Suzuki Yasuyuki, Ozaki Wataru, Yamamoto Tomohisa, Nomura Taishin
Graduate School of Engineering Science at Osaka University, Toyonaka, Osaka 560-8531, Japan.
Annu Int Conf IEEE Eng Med Biol Soc. 2012;2012:4847-50. doi: 10.1109/EMBC.2012.6347079.
Human motion analysis in various fields such as neurophysiology, clinical medicine, and sports sciences utilizes a multi-rigid link model of a human body for considering kinetics by solving inverse dynamics of a motion, in which a motion capture system with reflective markers are often used to measure the motion, and then the obtained motion are mapped onto the multi-rigid link model. However, algorithms for such a mapping from spatio-temporal positions of the markers to the corresponding posture of the model are not always fully disclosed. Moreover, a common difficulty for such algorithms is an error caused by displacements of the markers attached on the body surface, referred to as the skin motion error. In this study, we developed a simple algorithm that maps positions of the markers to the corresponding posture of a rigid link model, and examined accuracy of the algorithm by evaluating quantitatively differences between the measured and the estimated posture. We also analyzed the skin motion error. It is shown that magnitude of the error was determined not only by the amplitude of the skin motion, but also by the direction of the marker displacement relative to the frame of reference attached to each segment of the body.
在神经生理学、临床医学和运动科学等各个领域的人体运动分析中，通过求解运动的逆动力学来考虑动力学时，会利用人体的多刚体链接模型。在这种分析中，常使用带有反光标记的运动捕捉系统来测量运动，然后将获得的运动映射到多刚体链接模型上。然而，从标记的时空位置到模型相应姿态的这种映射算法并不总是完全公开的。此外，此类算法的一个常见难题是附着在身体表面的标记位移所导致的误差，即皮肤运动误差。在本研究中，我们开发了一种将标记位置映射到刚体链接模型相应姿态的简单算法，并通过定量评估测量姿态与估计姿态之间的差异来检验该算法的准确性。我们还分析了皮肤运动误差。结果表明，误差的大小不仅取决于皮肤运动的幅度，还取决于标记相对于附着在身体各节段上的参考系的位移方向。
