Graphics Reference
In-Depth Information
Chapter 9
In this chapter, we discuss applications of our 3D face processing framework.
Two application will be described in more details. The first application is
model-based very low bit-rate face video coding. The other application is an
integrated human-computer interaction environment. Finally, we conclude this
chapter and discuss other potential applications that could benefit from our 3D
face processing framework.
1. Model-based Very Low Bit-rate Face Video Coding
1.1 Introduction
Facial motions convey very important visual cues which are valuable for
human-to-human communication. When the volume of video data is over-
whelming and stable high-capacity bandwidth is not available, very low bit rate
video coding can be a solution for teleconferencing.
To accommodate the needs of transmission of large volume of video data over
limited channel, several video coding standards, such as MPEG-1, MPEG-2,
H.261, H.263 and H.264 have been proposed. Because these approaches only
utilize the spatial-temporal redundancy statistics of the video waveform signal
without a prior knowledge of the semantic content of the video, they are well
applicable for general purpose video data compression where the scene in video
frame is arbitrary. In the mean time, due to the difficulty to extract redundancy
from video, a high coding rate usually also accompanies a high coding latency
for certain video quality. This hinders these approaches from applications where
real-time video transmission is needed.
For the applications where human face is known as the major focus of at-
tention, model-based coding has been proposed to improve coding efficiency
[Aizawa and Huang, 1995]. One example is the MPEG-4 face animation stan-
Search WWH ::

Custom Search