个体相关的人体质量参数求解

Subject-specific Body Segment Parameter Optimization

在传统的基于物理的角色动画中,人体模型一般使用简单的多刚体模型进行近似。但是人体具有非常灵活的关节和几何变化,因此在传统的基于物理的角色动画中,使用简单多刚体模型估计的人体质量参数不够准确。针对这个问题,提出一种基于测力台数据和运动捕获数据的人体质量参数优化方法。该方法根据地面接触力信息和采集获得的运动信息,通过拉格朗日方程的约束,迭代优化获得人体质量参数。采集获得了多个角色的多种不同类型的运动。实验结果表明:该方法计算获得的人体质量参数信息与已有生物力学文献结果一致,而且能够有效的减少运动过程中存在的噪声(拉格朗日误差)信息。

In conventional methods of physics based character animation, human model are approximated by the rigid body models. However, the human body parts are not as easily described as rigid bodies with defined joints and inflexible geometry, daily varying mass, etc. A body segment parameter optimization method based on the force plate data and motion capture data was proposed. This method optimized the body segment parameter iteratively using the Lagrange constraints based on the contact information and motion capture data. Different subjects took part in the motion capture process with various kinds of motions. A comprehensive evaluation was done for the optimization of body segment parameter. The results show that the optimized body segment parameter is consistent with the biomechanics literature and the result can decrease the noise information in the motions.