Synthesis and Analysis Techniques for the Human Body: R&D Projects - 3D Modeling and Animation: Synthesis and Analysis Techniques for the Human Body - page 346

interactive multimedia content to any platform over any network. Based on the
Virtual Reality Modeling Language (VRML) standard developed by the Web3D
Consortium, MPEG-4 has been under development since 1993 and today is ready
for use. The first generation of MPEG-4 content servers and authoring tools are
now available. Advances in the MPEG-4 standard still continue, particularly in
the area of 3D data processing, offering a unique opportunity to generate new
revenue streams by way of MPEG-4 and related MPEG activities standards.
The Animation Framework eXtension (AFX) (MPEG-4 AFX), for example, is a
joint Web3D-MPEG effort that will define new 3D capabilities for the next
version of the MPEG-4 standard. Similarly, the MPEG group has recently
initiated an effort to develop standards for Multi-User capabilities in MPEG-4
(MPEG-4, requirements for Multi-user worlds).
The issue of hand and body modeling and animation has been addressed by the
Synthetic/Natural Hybrid Coding (SNHC) subgroup of the MPEG-4 standard-
ization group. More specifically, 296 Body Animation Parameters (BAPs) are
defined by MPEG-4 SNHC to describe almost any possible body posture, 28 of
which describe movements of the arm and hand. Most BAPs represent angles
of rotation around body joints. Due to the fact that the number of parameters is
very large, accurate estimation of these parameters for luminance or color
images is a very difficult task. However, if depth images from a calibrated
camera system are available, this problem is significantly simplified.
MPEG-4 originally focused on video and FBA (Face and Body Animation)
coding. The MPEG-4 FBA framework is limited to human-like virtual character
animation. Recently, the FBA specifications have been extended to the so-called
Bone-Based Animation (BBA) (Sévenier, 2002) specifications in order to
animate any articulated virtual character (Preda & Preteux, 2002).
Relative R&D Projects Worldwide
Computer vision-based techniques are not mature enough to be used for
industrial applications and, thus, semi-automatic systems are usually adopted.
Traditional systems have evolved from the 90s on the analysis of applied human
motion (analysis of sports performance, choreographic movements, medicine
and health, robotics, etc.) towards current semi-automatic systems based on the
implantation of optical markers, which are afterwards digitalized by stereoscopic
images.
Nowadays, the technology of human motion capture and human body synthesis
has achieved a reasonable degree of maturity. This can be deduced from the fact
that, currently, there exist several commercial systems that permit the capture/