Modeling and compression of motion capture data

Abstract

Motion capture (MoCap) system uses sensors or markers, placed on human body joints, to record the movements of a human in space over time. Motion capture data is used in many entertainment applications such as in virtual reality environments to drive avatars, in video games to animate characters, in movies to produce CG effects, etc. In this paper, we present an efficient method for modeling and compression of motion capture data. The method uses quadratic Bezier curve fitting to smoothly model and compress the MoCap data. The temporal variation of MoCap data of each joint is approximated and parameterized using Bezier segments. Simulation results shows that our method uses smaller storage and better visual quality compared to other methods. The low degree of quadratic Bezier curve ensures computationally efficiency required for the realtime gaming applications.