This article presents
a part of a system that is used to analyse and synthesize human
movement by means of a
color segmentation and matching process to track and reconstruct
the main aspects using a biomechanical
model of a person. In particular, we explain the application of
the Marching Cubes algorithm [Lorensen87]
obtained from a set of voxel information. A color segmentation criterion
is proposed. The segmented parts help
us reduce the space search in the matching process. The main purpose
of the system is to carry out a correspondence
between this graphic model (primitives) and the person in movement
in several real images (color
and grey level).
The original voxel approximation is used to roughly fit the volume
occupied by the person in the scene. The
process is intended to be non-invasive and automatic, although it
is currently used with the minimum manual
intervention of the user. The system works in a controlled environment.
The study of the movement of the human
body is applicable to many current fields of science and technology,
such as biomechanical study in
sporting areas, the integration of people in virtual worlds, and
others related to computer vision techniques, in
order to recognise or track people in scenes. The final result will
enable us to integrate the synthetic model
and its movement with the real person in a real or virtual world
adaptable to different applications.
|