m accuracy for
fied. For this
ly 700mm and
rious locations
different spatial
1easured using
achine (CMM).
and cylindrical,
rement.
ig a sub-pixel
ellipse-shaped
ge coordinates
program which
st.
ed in different
meras, without
‘a and various
ns of cameras
images
ries between 2
s from different
ers show the
f observations.
Its of different
in deviation is
and maximum
| the calculated
d the nominal
a
N
;
3 obs > 3 obs
nts
era and rotations;
ages per point; 3
n three images
These results make clear that the maximum deviations
are always less than O.imm if more than three
observations are used for 3D point measurement. If the
minimum number of two images is used for point
measurement, the maximum differences are always less
than 0.2mm.
A similar interpretation of the results can be made for the
effect of different imaging configurations. Optimal ray
intersection angles are obtained if different cameras are
combined with different rotations. It is obvious that a
reliable and accurate point determination depends on the
intersection angle in space. For combined measurement
sufficient results can be obtained even with 2 or 3
images. Using other configurations the maximum
deviations are again less than 0.2mm.
The investigation of distance accuracy show that higher
accuracies (< 0.1mm) can be achieved if suitable
configurations of sensors and rotations are used. If less
accuracy is required (< 0.2mm), one can work with the
minimum number of images using 2 or 3 observations.
4. SUMMARY
The prototype installation of the Programmable Optical
Measuring System (POM) is equipped with three Rollei
RSC digital high-resolution cameras and a digital rotary
table. The calibration of the system and its accuracy in
this configuration were investigated.
System calibration and orientation is performed using a
non-calibrated testfield with retro-reflective targets
combined with calibrated spatial distances within the
measuring volume. With a set of 24 images the complete
calibration of interior and exterior orientation of the
cameras and the rotary table is provided. The self-
calibrating bundle adjustment leads to an RMS (sigma 0)
of 1.5um, and an accuracy of 0.04mm in object space.
Repeated measurements of targets have been carried
out in order to judge the inner accuracy and stability of
the RSC sensors. These tests have shown that a high
accuracy of < 1.0um in image space is permanently
achievable. This result is even confirmed if the rotary
table is involved. This inner accuracy leads to an
accuracy of < 30pm in object space.
The measurements of calibrated distances show that the
final accuracy of the system is less than the calibration
results from bundle adjustment if a spatial intersection is
applied under practical conditions. The specified
accuracy of 0.1mm for a distance measurement can be
obtained if a larger number of images ( > 3) or a
sufficient image configuration (different cameras and
turntable rotations) is used. With the minimum number of
observations (2-3 images) the resulting accuracy is
always better than 0.2mm.
The results from calibration process are better than those
from practical measurements, showing that there is still a
possibility of improving system accuracy. Additional
effects (temperature changes, longterm characteristics,
different algorithms, influences of light) will be
investigated in the near future.
It has been shown that an accuracy of < 0.1mm in an
object space of about 2.0m x 2.0m x 0.6m can be
obtained. The relative system accuracy ranges between
1:20,000 (for spatial intersection) and 1:40,000 (using
bundle adjustment) of object size.
REFERENCES
Andresen.K., Helsch,R., 1990: Calculation of Analytical
Elements in Space Using Contour Algorithms. ISPRS
Symposium Comm. V, Zürich.
Hinsken,L., 1989: CAP - Ein Programm zur kombinierten
Bündelausgleichung auf Personal-Computern. BuL 3/89.
Loser,R., Luhmann,T., 1992: The Programmable Optical
3D Measuring System POM - Applications and
Performance. ISPRS Congress, Comm. V, Washington.
Luhmann, T., 1990: An Integrated System for Real-time
and On-line Applications in Industrial Photogrammetry.
Symposium ISPRS Comm. V, Zürich.
Riechmann,W., 1990: The Reseau-Scanning Camera
System; Conception and First Measurement Results.
Symposium ISPRS Comm. V, Zürich.
Wester-Ebbinghaus,W., 1985: Verfahren zur
Feldkalibrierung von photogrammetrischen
Aufnahmekammern im Nahbereich. Arbeitstagung
Kammerkalibrierung in der photogrammetrischen Praxis,
Deutsche Geodätische Kommission, Reihe B, Nr. 275,
pp. 106 - 114, München.
Zhou, G., 1986: Accurate Determination of Ellipse
Centers in Digital Imagery. ASPRS Annual Convention,
Vol. 4, March 1986.
