visual cues in communication. Experiment shows that this approach achieves
higher PSNR and better visual quality in very low bit-rate conditions. The other
application is an integrated HCI environment for computer-aided education. In
this environment, face analysis techniques is used to understand the users' state
and synthetic face is used as interface to help engage users.
More generally, our face processing framework can be applied in many ap-
plications related to human faces. Besides the two applications described in this
chapter, other examples of potential applications include: (1) intelligent video
surveillance; and (2) diagnosis and rehabilitation tools for face-related medical
problems. In security-related video surveillance, human races provide valuable
visual cues to identify people and understand human activities. In face-related
medical problems, such as disorders of facial muscles and associated nervous
system, facial visual cues are important input for diagnosis of the problems.
Our face processing framework provides a possibility for automating the di-
agnosis. On the other hand, one of the rehabilitation techniques is presenting
appropriate audio-visual stimuli (e.g. videos of normal facial expressions and
talking faces) to patients. The face synthesis techniques can help to generate
and manipulate these stimuli more easily.