tube contour. Optionally a light table can be installed on
the ground plate, which enables to measure any kind and
material of tube surfaces.
The parts fixture must hold the tube to be measured that
way, that a deformation free positioning can be realized.
Additionally the fixtures must be designed that way, that
the user of the system can easily put different kind of
tubes into the system without making any modification or
adjustments to the parts fixture. Therefore a net of elastic
material was chosen as parts fixture. By a low pre
tension of the net, it can be achieved that any point on
the net is supported according to the actual weight force
of the tube.
2.2 Electronic hardware
The CCD cameras incorporated in the system are of the
type JenaCam M-77m (by RJM). Those are monochrome
CCD matrix-cameras with a 2/3 inch interline transfer
sensor from Sony with 756 x 581 sensorelements and
pixelsize of 11 microns. The pixelsynchroneous mode of
operation makes this cameratype suitable for an
utilisation in measuring systems with highest accuracy
requirements (Bósemann 1990). As measurement and
system computer a ‘486 industrial PC is used, equipped
with three frame grabbers of the type BFP-AT/20 (by
Leutron). Each of the three frame grabbers is connected
to four cameras via an interface board JCI 2/4 (by RJM)
which allows a pixel synchroneous read out of the
camera images with a resolution of 768 x 512 picture
elements. Besides the VGA-monitor for the display of
user interface and measurement results the camera
images can be displayed on an additional video monitor.
3. SYSTEM CALIBRATION
3.2 Calibration during system setup
During the installation of the system in the production
environment the coordinates of the controlpoints on the
groundplate have to be determined at a higher level of
accuracy. Therefore the controlpoint field was recorded
with 26 digital images by a Kodak DCS 460. The images
"were processed with the digital photogrammetric station
DPA-WIN (Peipe, Schneider, Sinnreich 1993).
Fig. 2: Digital photogrammetric station DPA-WIN
56
The image measurement and the bundle triangulation
resulted in an image accuracy of 0.23 uim and an object
accuracy of 0.02 millimetre in the x- and y-direction and
0.04 millimetre in the z-direction. The results show a
discrepancy between the x- and y-directions and the z-
directions in the accuracy. This is caused by the very
narrow environment due to the sidewalls of the
measurement frame. Anyhow the accuracy is ten times
higher than the achievable accuracy of the online CCD
cameras. The DPA-WIN with the Kodak DCS 460 is also
used for periodical on-site calibration of the system which
has been performed approximately every twelve month.
3.3 System calibration
For the calibration of the whole system, the reference
marks on the ground plate are measured automatically
with high precision in the different camera images with
methods from digital image processing. The positions
and orientations of the cameras can be calculated
together with parameters of the imaging optics. This
calibration of the camera setup is the basis for the later
precise 3-dimensional determination of the shape in a
bundle adjustment of the tube.
In the calibration procedure the exterior orientation of
each camera has to be estimated with high precision.
Therefore the black circular reference marks on the
ground plate are automatically measured in each image.
The calibration menu give the opportunity to test the
calibration and to calculate new calibrations. In the
calibration test only the reference marks are measured in
the images and the differences to the calibration are
displayed. From this information the user can decide
whether a new calibration is necessary or not. Calibration
test and calibration running completly automatical
without any user interaction. Since the measuring and
calculation times for a new calibration are short it is
recommended to perform a new calibration at least
daily.
4. EVALUATION PROCEDURE
The digital images were processed by methods from
photogrammetry and image processing which were used
to derive threedimensional coordinates of the bend points
of the tube. In a second step these threedimensional
coordinates were used for a CAD comparison using a
socalled optical gauge.
4.1 Measurement principle
Spatially bended tubes can be described in several ways.
For the description and the display in CAD systems a
tube is reduced to its center line. The tube can be
described as a spatial polygon, with the theoretical
bending points given by the intersection of two following
straight parts and with the start- and endpoints of the
tube. From the coordinates of the polygon now the
theoretical bending elements can be calculated. This
bending elements are the distance of two following
theoretical bend points, the angle between two following
straight elements and a rotation angle calculated from
four following bend points. For the calculation of real
bending elements only the length between theoretical
bend points has to be calculated with tube diameter and
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B5. Vienna 1996
bending radiu:
can be calcul
following des
reduced on tt
image and obj
The measure
steps. First im
in the comput
automatically
reducing the
Straight eleme
measured tut
elements deli
points in the |
detected in e
theoretical ber
by intersectinc
Fig.
4.2 Optical gz
In this mode
elements of
compared to tl
coordinates of
readable form
adjust import
diameter or
measurement,
reference tube
the measuring
order to conti
measuring cel
be viewed on
measurement
tube could b
repositioning
repeated meas
be measured c
For the meas
ways can be :
tube exists in :
for example ti
adjusted mant
the measurem
