International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B1. Istanbul 2004 
  
From the results in table 1, one can conclude mainly two 
things: 
l. The accuracy achieved after post-processing is within 
the expectations, given by IGI. 
If the data should be used for direct geo-referencing 
without integrated sensor orientation, calibration of the 
system is needed for each and every flight. Even if the 
IMU has not been detached (flight 15.07., morning and 
afternoon, e.g.), the results differ considerably. 
ts 
The project has been finished successfully. 
3.2 Project Pipeline 
For pipeline planning in Germany, ortho photos had to be 
produced. The total distance was about 770 km, including 
several variations in certain areas (see figure 4). As already 
mentioned, the time table was tight (six weeks) and the 
budget of course not very high. 
Using an image scale of 1: 6.000, 1770 images covered the 
complete project, flown in five consecutive days. Four 
calibrations fields have been established distributed over the 
area to minimise additional flying to reach the nearest one. 
Also this project was flown with 60% forward over lap, out 
of which only every second image was scanned. Also here, 
no overall triangulation took place, which reduced the 
number of ground control points from app. 250 to 24. 
The DTM could be bought from the local survey authorities. 
This way, about 500 ortho photos mosaics have been 
produced (nearly) in time. 
4. Conclusions 
For the first projects, Hansa Luftbild German Air Surveys 
had an signalised general calibration field in the vicinity of 
the home airport Münster Osnabrück. Laid out for an image 
scale of 1: 6.000, 22 ground control points have been 
permanently signalised. The idea was, to have a reliable 
calibration field for permanent using, which could easily be 
flown just after starting or before landing or even both. From 
the economic point of view, it was planed to be a mid term 
investment, which would reduce cost overtime for GPS/INS 
projects. 
  
  
  
     
   
   
    
  
  
  
  
  
  
  
EQ SV Baker 
LA 
Fig. 5. Layout ofthe general calibration field 
  
It was never the case, that due to clouds over the calibration 
field, this could not be used before or after a GPS/INS flight. 
But very soon it was found out, that different conditions in 
terms of image scale and therefore flying height caused so 
different meteorological conditions (temperature, pressure, 
e.g.) that this caused problems. For projects close by and 
similar project parameters concerning image scale, etc. the 
results have been within the expectations. But this was not 
the case for projects with image scales of factor 3 or more, 
(6.000 to 18.000, e.g.) or locations several hundred 
kilometres off. During quality checks for these ortho photo 
products, large discrepancies between neighbouring stripes 
have been found, which could not be explained by simple 
reasons. By computing a triangulation, it was shown, that the 
exterior orientation was the source. 
The calibration for these projects had to be repeated with 
data taken from the actual project area, which drastically 
improved the result. Consequently, the calibration field in the 
vicinity of the airport was abandoned and since then, the 
boresight alignment is done under the same conditions in the 
actual project area successfully. 
This shows that the process of calibration does not only serve 
the task to determine the offset between projection center and 
center of the IMU. Also influences of transformation to the 
local system, variations in the focal length due to influence of 
pressure and temperature during flight, e.g. are modelled in 
the triangulation process an therefore included in the 
calibration values. 
Direct geo-referencing for ortho photos can be used quite 
successfully — technically and economic — by using the 
calibrated results directly. The possible loss of accuracy 
compared to a triangulation has to be balanced against the 
gain in speed and higher economic performance. 
From the point of reliability of the GPS/INS system , there 
was absolute no problem — no loss of data or any other 
complication so far. 
Using direct geo-referencing for stereo restitution was tried 
only once in a very small project at Hansa Luftbild German 
Air Surveys. It was successful, but due to the structure of 
most of our contracts, usually the achievable accuracy does 
not meet the requirements of the stereo restitution projects. 
Using the GPS/INS data as additional observations in a 
digital triangulation process make sense in certain described 
cases, but in general, the matching process runs with the 
kinematic GPS data already quite well. 
Reference: 
AEROcontrol CCNS, technical information sheet, IGI, 
Kreuztal, Germany, http://www.igi-cens.com/ 
aerocontrol.html (accessed 07.05.2004) 
Cramer, M. and Stallmann, D.: System Calibration for Direct 
Geo-Referencing. Proceedings of ISPRS Comm. I 
Symposium, Graz (2002) 
   
Interi 
  
Cram 
triang 
Photc 
Dree: 
Refer 
(Phot 
Stuttg 
Schr 
Preci 
Refer