Analysis of human motion interpolation methods based on real captured data
Referat wygłoszony
Status: | |
Autorzy: | Smołka Jakub, Skublewska-Paszkowska Maria |
Rok wydania: | 2013 |
Język: | angielski |
Źródło: | International Interdisciplinary PhD Workshop 2013 Proceedings September 8‐11, 2013 Brno, Czech Republic |
Państwo wystąpienia: | POLSKA |
Efekt badań statutowych | NIE |
Abstrakty: | angielski |
During a motion capture session some markers become occluded for brief periods of time. This results in gaps in recorded motion. The gaps, however, can be filled in by the use of a proper interpolation method on the captured data. Frequently mocap data is recorded together with analog data such as ground reaction forces. Sampling frequency of the analog data and the frame rate of the motion capture data may be different. Interpolation methods can also be used to resample data in order to match these frequencies. Data resolution may be both increased and decreased. This allows for generating a smoother animation or making analysis and presentation easier. There are many various interpolation algorithms. Three interpolation methods are included in this comparison. These methods are: Catmull-Rom spline, modified Catmull-Rom spline and blended parabolas. They are analyzed in two respects mentioned above: filling in gaps in motion capture data and resampling captured data. In all tests performed for the purpose of this paper, a sample data file (Eb015pi.c3d) available on the 3D Biomechanics Data Standard site [3] is used. |