Synthesis and Analysis Techniques for the Human Body: R&D Projects - 3D Modeling and Animation: Synthesis and Analysis Techniques for the Human Body - page 365

is developing an innovative system for recognition of human motion based on
image processing analysis/synthesis techniques. The system will be enhanced to
recognize and analyze other biped and no-biped beings, like for example pet
animals, robots, etc. In addition, it will be able to focus only in a part of the body
but with high detail, like for example the hands or the face. The fundamental aim
is to obtain a 3D model of the person or persons by means of a sequence of
images taken from different viewpoints. With this information the aim is to carry
out different tasks such as: realistic animation of a person, biomechanical study
of sports or dance movements, recognition of a person (face and movements),
integration of a virtual humanoid with real characters, interaction in a person and
humanoid immersed environment, robot tracking of a person, etc.
To ensure the widest range of applications, the individual recorded will not wear
any type of marker or special suit. To this end, biomechanical models will be
constructed using a hierarchical and articulated structure in order to establish a
correlation between each structural element of the biomechanical model with the
analytical characteristics of the images obtained using different views. Innova-
tive shape or part recognition techniques will be applied. The biomechanical
model will include a knowledge database to retain high-level information of the
motions.
To make the system usable it will also be necessary to develop specific
applications and plug-ins to integrate the animation into end users tools such as
digital TV production software, animation software and virtual environments like
a CAVE. The HUMODAN system is expected to allow the development of new
applications including: a) medical applications such as non-invasive diagnose of
neurological damages affecting the locomotor system, b) surgery training,
anatomy, traumatology, etc., training the use of tools or processes in dangerous
environments, virtual laboratories, sports, etc.
HUMODAN technical approach
HUMODAN is an on-going project being mainly developed based on the Avango
framework (Tramberend, 2000). Avango is a programming framework for
building distributed, interactive VE applications. It uses the C++ programming
language to define two categories of object classes. Nodes provide an object-
oriented scene-graph API, which allows the representation and rendering of
complex geometry. Sensors provide Avango with its interface to the real world
and they are used to import external device data into an application.
Avango's distribution paradigm is based on an object-based representation that
matches well with scenegraph-based virtual environment development. All
objects in Avango are field-containers. These field-containers represent the