Model-Based Head Pose Estimation

by

Axel Weissenfeld, Onay Urfalioglu, Kang Liu, Joern Ostermann (UHANN)

A model-based head motion estimation algorithm is designed. Algorithm estimates the head pose due to tracking feature points. Large out-of-plane rotations lead to pose errors, if feature points are not updated. We propose a new strategy for an automatic update of feature points.

Presentation (ppt)
Always Update Texture Information versus Proposed Strategy Video [352x288 133KB] (mpg)
Never Update Texture Information versus Proposed Strategy [720x576 19MB] (mpg)
Update_vergleich (png)
result_error (png)
result_error (eps)