摘要： 研究步态跟踪的关键技术，提出一种比较鲁棒的多摄像机自校正跟踪方法，该方法结合卡尔曼滤波预测技术、标记点轨迹平滑性和多视角几何关系对标记点的运动轨迹进行跟踪。将该方法应用于三维人体步态重构，并与商业系统Vicon做了对比实验。实验表明，该方法可以获得较精确的步态数据，能够满足人体步态分析的需求。

关键词: 运动捕获, 摄像机定标, 运动跟踪

Abstract: This paper researches the key technologies of human motion capture. A robust self-corrected tracking method is presented in this paper. This method combines Kalman filter technology, the smooth of point’s track and multiple cameras’ epipolar geometry to track points. This method is used in 3D human gait motion capture, and is implemented in a gait capture system based on markers and multiple cameras. The effectiveness of this method is verified by experiments.

Key words: motion capture, camera calibration, motion tracking

中图分类号: 

• TP391.41