Disney Research and Robotics Institute have developed a method which translates motions of actors into a 3D-face model.
320 Markers were applied on their faces to record sentences with emotional content, localized actions and random motions with a camera. These data were then analyzed by using a mathematical method to sub-divide the face into various regions, 13 in total. This enables animators to intuitively create the poses they need. It is envisaged to create a facial model which can be used by computer animators to quickly animate any character for movies, video games or exhibits.
http://drp.disneyresearch.com/projects/irblfm/irblfm.html