An Automatic Solution Framework for Robust and Computationally Efficient Joint Estimation in Optical Motion Capture

Jianwei Hang,Joan Lasenby,Adrian Li

CSCI '15 Proceedings of the 2015 International Conference on Computational Science and Computational Intelligence (CSCI)(2015)

引用 0|浏览10
We propose an automatic and computationally efficient solution framework for addressing the joint estimation problem in marker-based optical motion capture. A fast joint estimator is presented which only requires an optimisation over 3 variables using marker-trajectory-bases (MTB). We also introduce the theory of solvability propagation to realise this automation. The framework acquires a 'hybrid' power making it able to deal with difficult cases where there are less than three markers on the body segments. It does this by combining the MTB-based and the joint-marker-variance optimisation methods. Computer simulations are used to examine the framework in terms of accuracy, speed and functionality. Results from these simulations show that the framework is robust and produces fast and accurate solutions.
Joint Estimation, Optical Motion Capture
AI 理解论文