Hybrid 3D face motion analysis system.
Figure 7.2. (a): The input video frame. (b): The snapshot of the geometric tracking system.
(c): The extracted texture map
The geometric deformation parameters determine the registration of each
image frame to the face texture map. Thus we can derive a face texture map
from each image frame, which are independent of geometric motion.
Here, is the coordinate system of the texture map. Figure 7.2(c) shows
the extracted texture map. From the texture maps, we extract appearance-based
features described in Section 1.1. These features are designed for subtle de-
tails of facial expression and independent of people's face surface albeo. We
then use the texture (appearance) features, together with shape (geometric) fea-
tures, to analyze face appearance variations based on semantically meaningful