基于无线惯性传感器的人体动作捕获方法

为解决光学式动作捕捉设备成本高昂和操作复杂的问题,提出一种通过佩戴在用户身体上的无线惯性传感器进行人体动作数据捕获的方法.在用户身体的各个运动部位绑定多个由加速度和磁通传感器构成的无线惯性传感器单元,传感器通过无线信号发送运动传感数据到计算机端,应用优化算法计算惯性传感单元的三维朝向信息,最后将四元数与动画角色的骨骼绑定后生成人体动作数据.为了解决运动过程中的行走导致的骨架根节点移动问题,利用地形参数反向计算和调整角色骨架根节点位置,使生成的动作符合地形和环境要求,达到真实自然的运动效果.实验结果表明,使用无线惯性传感器进行人体动作捕获得到的动作数据准确度高,朝向计算方法运行速度快,能够满足实时性应用的要求,同时显著降低动作捕获的成本和使用复杂度.

Human motion capture using wireless inertial sensors

Human motion capture is an important technology in the three-dimensional character animation production.Since the motion capture process is very expensive and complex,in this paper we introduce a new method that captures the human motion data from the inertial motion sensors attached to the user’s limbs.The user wears multiple wireless inertial motion sensors with accelerometer and magnetometer,and the sensor data are sent to the computer using the wireless sensor network.Levenberg-Marquardt based optimization algorithm is applied to estimate the three-dimensional orientation of the inertial sensor and then bind the obtained orientation quaternion to the skeleton of the animation character.The root position of the animation character is modified inversely from the relative status of the supporting leg and the ground in order to make the walking motion more realistic.The experimental results show that our method can capture the user’s motion effectively and produce accurate and realistic motion capture data.