Summary: | 碩士 === 長庚大學 === 醫療機電工程研究所 === 96 === Motion captures are techniques of measuring and recording human
torso motions in digital representations. The motion capture systems can
be used to recording and analyzing motions of athletes or rehabilitation
patients as well as producing movies and 3D games. Therefore, the
motion capture system is an important research topic. According to the
importance of the motion capture system, this work develops a
stereoscopic vision based human motion measurement system. A
stereoscopic vision module is configured with two parallel CCD cameras,
and they are used to capture the images of humans. The image processing
and recognition techniques are further used to identify the image
coordinates of all active orange markers which are mounted on the joints
of human torsos. At the same time, the identity of each marker is traced in
terms of comparing two sequential images frames in the time domain.
Finally, the spatial coordinates of all markers are calculated via
corresponded markers in two image coordinates. At the same time, a 3D
graphical library of OpenGL is used to display the torso plot of humans.
On the other hand, this work also implements a remote motion
representation system using the windows socket libraries to provide
online investigations of motions for remote users such as physicians and
therapists. Consequently, the works in this thesis consist of a stereoscopic
image capture platform, a 3D coordinate positioning module for spatial
markers, a motion measurement module, remote 3D torso representation
module, and finally system integration, test and evaluation. In the future,
the proposed system may combine with mobiles robots to serve as a
remote motion measurement platform for distance rehabilitations.
|