kodate 1998 
dimensional 
dard video 
e the object 
> at least 4 
ameras and 
values have 
d iteratively 
\pproximate 
e controlled 
given under 
accuracy of 
ace can not 
ral edges of 
measured 
nitted. After 
equivalent 
med. 
2.5m 
  
  
jS 
al 2-D CAD 
t may have 
s additional 
1). On the 
one hand the CAD data is used to generate approximate 
values for the later edge detection in the images. On the 
other hand the data represents the nominal shape of the 
object used for the quality check. 
Besides photogrammetric measurement the system 
should also be able to archive images and measurement 
results according to standards in quality assurance (DIN 
EN ISO 9000 ff.). 
  
   
Production line N NS 
50m 50m 
Figure 2: Principle system setup 
1.3 Solutions 
Figure 2 shows the principle system setup. One or more 
cameras are mounted above the production line. The 
system controller receives a signal and a part 
identification code from the production controller in order 
to start the measurement. 
The objects are assumed to be flat enough for a single 
image compilation. If at least four control points are given 
each image can be rectified onto the reference plane of 
the control points using the well-known perspective 
transformations: 
y.2ot4 xta y 
LG Ak Ca x 
yoboth x+b, y 
ICH 
Equation (1) may only be used if the images have no 
distortion effects. Therefore the image coordinates x,y 
have to be corrected by the parameters of interior 
orientation which have to be calibrated in advance. 
2 SYSTEM CONCEPT 
2.2 Hardware Components 
Image acquisition is performed by a standard Ye“ CCD 
video camera (European video standard CCIR, 25Hz, 768 
X 572 pixel, frame transfer). It is equiped with a standard 
TV lens (f=8mm) which is not designed for high-quality 
photogrammetric purposes. The camera is protected 
against dust and oil by a closed camera housing. 
The analog video signal is transfered to the computer 
system (NT workstation, Pentium-2, 128 MB RAM) which 
consists of a video frame grabber ITI AM-VS. The 
digitized images can be displayed in real-time either on a 
video monitor or directly as a window in the graphical 
user-interface. 
* 
A 
4 diffuse reflectors —— — — —— 
Figure 3: Diffuse illumination technique 
  
  
Object illumination is performed by 300 W halogene 
spots. The light sources are used for indirect light 
reflection in order to generate a bright diffuse illumination 
(Figure 3). Direct light sources would yield to bright spots 
on the object surface, thus leading to CCD overflow in the 
sensor. In addition the optical representation of object 
edges is better with diffuse lighting. 
  
Figure 4: Testfield with control points 
243