Fua, Pascal 
  
  
  
(a) 
Figure 11: Frames 0, 5 and 10 of a walking sequence. Top row: Original sequence. Middle row: Tracking results using 
the simple model, overlaid with the 3-D points’ projections. Bottom Row: Final fitting results using the detailed model. 
In future work, we intend to extend our approach to capturing facial dynamics in addition to body dynamics. We will 
use this motion capture data to cliaracterize actions and derive animation and biometric models for specific motions, 
such as running, and specific facial expressions. Such a capability will improve our ability to visualize, analyze, edit 
and synthesize human motion. This will have applications ranging from movie making to sports medicine and athletic 
training. 
ACKNOWLEDGEMENTS 
We wish to thank Prof. Daniel Thalmann, Prof. Nadia Magnenat Thalmann, and Prof. Prem Kalra for having made their 
animation models available to us. We are also indebted to Prof. Grün for sharing with us his insights about least-squares 
technology. 
The work reported here was funded in part by the Swiss National Science Foundation and in part under European Esprit 
project Motion Capture (MOCA). 
  
266 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000.