Facial Feature Detection Based on AAM for Facial Animation
by
Kang Liu, Joern Ostermann (UHANN) in collaboration with TUT
The improved AAM (Active Appearance Model) performs more
accurately and stable and achieves subpel accuracy of
detection and location. The detected geometrical features
are used for the synthesis of facial animation. Morphing is
used to smooth the transition automatically based on the
features resulting from AAM.