cal positioning
ciple based on
ons for active
ation for every
sequences ac-
nd retro-target
targets
on, based on
rgets.
d for the “free-
ware functions.
e reduced to a
meras and pro-
st setups of a
tory: a tripod
cameras and a
n the fields of
evaluation and
s of a
oratory.
96
8.5 Applications and Results
For a measurement example an aluminum part with
specular surface was used (Figure 26). This part was
developed for testing purposes and has been measured
and reproduced by rapid prototyping from several institu-
tions [IMS94].
Figure 26 : IMS test-part, made of aluminum.
Figure 27 shows measured data from twelve different
views which were acquired to obtain one point cloud that
was suitable for model reconstruction (Figure 28). The
rms-noise was below 50 microns, which is approximately
0.025 % of the field of view.
calibration
orientation
transformation
"superposition of |
12 point clouds |;
e 1 ha
[=]
Figure 27: Measurement of IMS test part (single point
cloud and superposition of twelve point clouds).
345
Figure 28: CAD model based on point cloud.
4. CONCLUSIONS
There are many existing and new application areas for
“mechanic-free“ self calibrating, autonomously navigating
optical measurement systems:
e fast digitization of hand-made models,
e. measurement and documentation of tools,
e on-line production control (automotive parts),
e geometric control of turbine blades and wings,
e flexible programmable checking fixtures,
e collision control (e.g. for mobile robots).
Several sensor principles have to be combined to build
"free-flying" sensors for 3-D surface measurement. As an
example, a new sensor system has been shown which
integrates active optical and photogrammetric tech-
niques. It can produce 3-D coordinates for each camera
pixel (which can be 1 million or more in a few minutes). A
self-calibration principle with a simple hand-positioned
calibration plate is used for the camera/projector com-
bination. In this system a new orientation principle based
on 2-D and 3-D data is used for point-cloud registration
from several views. The registrated point clouds can be
sent to a CAD system for further 3-D processing.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B5. Vienna 1996