6 ACCURACY VERIFICATION AND RELIABILITY 
It is important for the user to get accurate and reliable 
measurements under realistic, industrial environments. 
The accuracy of MNS can be tested and verified by 
different methods as listed in the following chapters, either 
in dedicated laboratory setups, or as a part of the day to 
day operation. This guarantees the reliability of the 
measurement results. 
6.1 Camera calibration 
The camera calibration is a very thorough procedure, 
taking into account both global effects like lens distor- 
tions, and local effects on the pixel level. The entire 
sensor, each individual pixel is mapped to take care of 
pixel variations and defects. 
The reference instrument used for camera calibration is 
traceable to The US National Institute of Standards and 
Technology. 
6.2 On-site accuracy verification 
The Reference Bar used for determination of the exterior 
orientation has a length that is certified to an accuracy of 
3 um. Each setup procedure is concluded with a display 
of the accuracy in this setup, i.e. how well the measured 
bar distances match the certified value. 
Measuring the length of the Reference Bar at any chosen 
location at any time after an initialization is a verification 
that the calculated spatial camera positions and orienta- 
tions are still correct. It is not possible to achieve correct 
distance readings at any arbitrary position within the 
measurement volume if any errors have occurred that 
influence the accuracy of the cameras or the camera 
setup. 
At each Light Pen measurement the Light Pen geometry 
is checked against its nominal value. Any discrepancies, 
e.g. caused by accidentally moving a camera, will prevent 
the operator from making further measurements. 
6.3 System test according to VDI guidelines 
The light pen allows the system accuracy to be verified by 
methods used for CMM's, e.g. according to the German 
VDI guidelines (VDI/VDE 2617). MNS is the only photo- 
grammetry system having this feature. This feature 
allows for full verification of 3D accuracy. 
7 ACCURACY TEST RESULTS 
Thorough accuracy verifications has been made as parts 
of the technical cooperation with Volvo Car Corp., and 
related to tests and deliveries to other customers. 
The test results below are related to a setup geometry 
optimized to give a measurement volume of minimum 1.0 
x 1.0 x 1.0 m?. The camera intersection angle is 60 to 90 
deg. 
The test results are given as Ugg values, i.e. 20. 
  
  
  
  
  
  
Fig. 4. 
232 
Setup for Uncertainty of Length Measurement determination