适用于惯性动作捕捉与复现系统的轨迹捕捉算法及其实现

Trajectory Capture Algorithm and Realization for Inertial Motion Capture and Reproduction System

针对惯性动作捕捉与复现系统轨迹捕捉过程中由于零速检测不准确、姿态角误差发散造成捕捉到的轨迹不够平滑进而导致动画复现过程中人体模型发生严重抖动甚至位置跳变的问题,提出了一种基于改进零速检测的轨迹捕捉算法,可得到较高平滑度的轨迹。本文结合试验分析了零速检测误检源,并根据拉依达准则优化了检测算法,在此基础上构建了完整的轨迹捕捉Kalman滤波算法框架,其内嵌的姿态深度修正模块可有效抑制姿态角误差发散。最后通过实物试验,从精度和稳定性两方面验证了轨迹捕捉算法的可行性,试验表明,本文所述轨迹捕捉算法能得到较高平滑度的行走轨迹,因此能在一定程度上减小动画复现过程中人体模型的抖动,具有重要的工程意义。

Aiming at the problem that human body model has a serious jitter and even jump during animation reproduction,due to inaccurate zero-speed detection and severe dispersion of attitude error in trajectory capture process of inertial motion capture and recurrence system. A trajectory capture algorithm based on improved zero-speed detection is proposed,and a trajectory with higher smoothness can be obtained. In this paper,the zero-speed detection of misdetection source is analyzed in combination with experiments,and the detection algorithm is optimized according to Raida＇s criterion,based on this,a complete Kalman filter algorithm framework is constructed,the embedded attitude depth correction module can effectively correct the attitude angle. Finally,the feasibility of the trajectory capture algorithm is verified from the aspects of accuracy and stability,experiments show that the trajectory capture algorithm described in this paper can get a higher smoothness of walking trajectory. Therefore,it can reduce the jitter of the human body model in the process of animation reproduction to a certain extent,which is of great engineering significance.