'ertical direction
y, the system is
sed on the prin-
curring margins
from the system
le same is valid
y which is me-
scale spectrum.
ciple of réseau
e sensor via a
cen réseau- and
en the two sys-
urring from the
for example the
m, differing fo-
yossible rotation
with the sensor
becomes slow
emented which
utomatically: by
me mesh in the
o all sides is as-
rst mesh is cho-
orrected. When
iis means that at
1e digital partial
no crosses are
e algorithm sets
yund in the first
eing initialised.
ed in such way
1e corner of its
and width of a
ust be known in
the scanner is
quarter of the
ed to recognise
ted in such way
D-sensor
Mesh
that no identification problems are possible. Now, both
scanner distances and the corresponding frame grabber
positions at which the crosses were identified are
known. Now the scale can be derived from the quo-
tient of the scanner distances and the positioning dif-
ference on the image board. With this scale the scan-
ner position can be corrected easily that the mesh can
be focused centrally by the sensor. Then the scale that
had been dependent on the positioning accuracy of the
scanner can be identified through the distance of the
réseau crosses in the digital partial image and the di-
mensions of the réseaus.
* | (sc)
Tot + + + + + + + + + +
Figure 3: coordinate systems
Orientation of the measurement images
There are two basic possibilities when orienting mea-
surement images: on the one side, images can be ori-
ented with a preilluminated réseau, on the other side it
is also possible to orientate images with a wide mesh
réseau. In the first case, no réseau plate is necessary
on the scanner. The initialisation of the scanner as des-
cribed above is effected directly when orienting the
images. The two methods only differ in one feature: in
the first case, the meshes are taken for identifying
position and rotation, in the second case, the crosses
of the réseau are the reference. In principle, the two
methods are identical. From the identified signal (mesh
or cross) the rotation of the image system to the sensor
system is determined. With this identification together
with the information given on the réseau it is possible
to position directly above the second relevant signal.
With every further measured signal, the transformation
between the systems is being improved successively
until four signals are measured and the affin transfor-
mation can be calculated with an overdetermination.
This method allows the calculation of a transformation
between image- and reference system by giving only
one signal per image. Any other positions in the image
system are determined automatically. It is also possi-
ble, to orientate the images manually to make a mea-
surement of difficult images possible. The signals
which determine the orientation can be chosen in the
interpreter files.
MEASUREMENT
Optimising the Measurement
In order to assure a fast measurement, calculations,
sensor movements and switching the illumination
regulation have to be parallised.
In pictures without preilluminated réseau (Figure 4) it
is necessary, to keep the following measurement pro-
cedure for one point:
To select the réseau crosses and to be able to initialise
the meshes, it is necessary to change into the on-light
mode. During the light adjustment, the sensor is posi-
tioned above the mesh with the signal to be measured.
The image is then given in the frame grabber. Initiali-
sing the mesh in this partial image and changing into
the through-light mode is done simultaneously. The
time, necessary for measuring the target mark, is used
to prepare the scanner for further measurements. In the
on-light mode, the scanner receives the order to posi-
tion above the next mesh. When the measurement is
finished, the scanner is already placed above the next
mesh in the on-light mode. Therefore, on-light recor-
ding is possible without any delay.
Figure 4: Picture without preilluminated réseau
