Figure 9.1. (a) The synthesized face motion. (b) The reconstructed video frame with synthe-
sized face motion. (c) The reconstructed video frame using H.26L codec.
1.4 Summary and future work
As an application of the 3D face processing framework, we present an effi-
cient and robust very low bit rate face coding method via 3D face face tracking.
The facial motion parameters can be extracted from the video and transmitted
over the channel. Then the facial area residual errors and video background can
be coded using waveform-based coder at very low bit rates. Experiments show
that our method can achieve better PSNR around facial area than H.264/JVT
coder at about the same low bit rate and have better subjective visual quality.
The key issue of model-based coding is the 3D face tracker. We plan to
improve the accuracy and robustness of our 3D face tracking algorithm. Be-
cause our tracking system works in real-time, we can combine it with real-time
waveform-based coder to make the overall system real-time. Then it could be
used in a real-time low bit-rate video phone application. Finally, we plan to
model the face texture variation so that the residual can be further reduced.
2. Integrated Proactive HCI environments
Face tracking and expression recognition techniques help computers monitor
users' states in a human-computer interaction (HCI) environments. On the