Game Development Reference
In-Depth Information
of facial expression and motion analysis on monocular images by easily
finding the right references to the detailed description of all mentioned
methods.
Introduction
Researchers from the Computer Vision, Computer Graphics and Image Pro-
cessing communities have been studying the problems associated with the
analysis and synthesis of faces in motion for more than 20 years. The analysis
and synthesis techniques being developed can be useful for the definition of low-
rate bit image compression algorithms (model-based coding), new cinema
technologies, as well as for the deployment of virtual reality applications,
videoconferencing, etc. As computers evolve towards becoming more human-
oriented machines, human-computer interfaces, behavior-learning robots and
disable-adapted computer environments will use face expression analysis to be
able to react to human action. The
analysis of motion and expression from
monocular (single) images
is widely investigated because non-stereoscopic
static images and videos are the most affordable and extensively used visual
media (i.e., webcams).
This chapter reviews current techniques for the analysis of single images to
derive face animation. These methods can be classified based upon different
criteria:
1.
the nature of the analysis: global versus feature-based, real-time oriented;
2.
the complexity of the information retrieved: general expression generation
versus specific face motion;
3.
the tools utilized during the analysis: for instance, the cooperation of a 3D
head model;
4.
the degree of realism obtained from the Face Animation (FA) synthesis;
and
5.
the environmental conditions during the analysis: controlled or uniform
lighting, head-pose dependence or not.
Table 1 depicts a rough evaluation of the techniques that we review in this
chapter by comparing these criteria, considering the data provided by the
referenced articles, topics and other bibliographical material, as well as the
judgment of the authors.
Search WWH ::
Custom Search