Automated gap-filling for marker-based biomechanical motion capture data

Abstract

Marker-based motion capture presents the problem of gaps, which are traditionally processed using motion capture software, requiring intensive manual input. We propose and study an automated method of gap-filling that uses inverse kinematics (IK) to close the loop of an iterative process to minimize error, while nearly eliminating user input. Comparing our method to manual gap-filling, we observe a 21% reduction in the worst-case gap-filling error (p < 0.05), and an 80% reduction in completion time (p < 0.01). Our contribution encompasses the release of an open-source repository of the method and interaction with OpenSim.