3. PARTICIPANTS IN THE OEEPE TEST 
In order to attain the scope of the working group it ap- 
peared most appropriate to invite in a first phase the 
manufacturer to scan the test material and to proceed on 
an analysis of the scanned images. The members of the 
working group intended to run different test on the mate- 
rial and should be rather free to develop their proper 
strategy. The synthesis of these analysis should then 
allow to develop standard procedures for testing of scan- 
ners. However the scanning of the material took much 
longer time than planned. As only one test set was avail- 
able composed of originals, the delay within one only or- 
ganization extended seriously the scanning phase. 
Therefore the analysis of the test material remained very 
limited and only results of the pilot centre (Technical Uni- 
versity of Lausanne) can be presented up to now. 
The following firms have participated up to now in the 
scanner test, which is highly appreciated : 
Agfa, Mortsel, Belgium with the Agfa Horizon Plus 
Intergraph, Huntsville USA with the PS1 
Wehrli & Assoc, Valhalla USA, with the RM1 
Helava, San Diego USA with the DSW200 
Complementary scanning have been made by the follow- 
ing organizations : 
ICC Barcelona with the PS1 
Cetop, Lausanne with the PS1 
Institute of Photogrammetry of the ETH-Zirich with 
the Agfa Horizon 
Institute of Photogrammetry of the EPF-Lausanne 
with the DSW200 
4. ANALYSIS OF THE TEST MATERIAL 
Although the invitation for the test scanning was sent out 
in spring 1995, we disposed only beginning of February 
1996 of some representative test material. The time for 
the analysis was therefore very short and is by far not 
finished. Earlier publications concentrated on the analy- 
sis of black and white photographs and showed how to 
determine the image noise, image resolution and the dy- 
namic range (cf. [1],[2]). The current tests included reso- 
lution patterns, photographed with aerial cameras on high 
resolution film (test 6), colour test patterns (test 3) and 
the gray wedge (test 2). It appeared of great interest to 
accelerate the analysis of the image resolution and of the 
colour reproduction, as these criterias have not been 
treated earlier too intensively. 
4.1 Image resolution 
A lot was already discussed on image resolution and very 
often the pixel resolution is considered as limiting factor. 
However when analysing the reproduction of small ob- 
jects in photographs, like signalized points one realizes 
that the cut off frequency is not at all decisive, much 
more important is the loss of contrast for larger frequen- 
cies. This loss of contrast is given by the modulation 
transfer function (MTF). In order to determine the MTF, 
one should know the object function and of course the 
image function, but which is anyhow given by the digitized 
image. By computing the Fourier transforms one can then 
determine the spread function or its Fourier transformed, 
the modulation transfer function. By some simplifications 
the modulation transfer function can also be deduced 
58 
from the contrast reduction of a rectangular pattern, a 
way which was chosen here. 
  
Fig. 1 
Resolution pattern (test 6) and density profile scanned on 
the DSW200. 
Figure 1 shows a reproduction of the resolution target 
(lower part) as negative. A density profile was determined 
with the software tools of ISI of the ImageStation of Inter- 
graph. One recognizes that the contrast diminishes with 
the increasing frequencies and the signal vanish for fre- 
quencies of about 50 lines/mm. 
Contrast x 10 in % 
A 
10 | 
| N 
  
  
  
  
  
Sa 
0 > 
0 2 5 10 20 50 100 
Frequency in lines/mm 
  
  
  
  
  
  
  
  
Photoscan PS1 7.5 um 
Rastermaster RM1 12 um 
Agfa Horizon Plus — 20 um 
Helava DSW200 12 um 
PhotoScan PS1 15 um 
Helava DSW200 photo 4x enlarged resampled 40 um corresponding to 10 um on original 
Helava DSW200 photo 4x enlarged 10 um 
ommoou» 
Fig. 2 
Modulation transfer function for different scanners. 
Figure 2 gives an overview of the determined transfer 
functions for the different scanners tested and different 
image configurations. The best curve was obtained for a 
simulated pixel size of 2.5 um for the Helava DSW200. In 
this case the original photograph was enlarged 4 times on 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B6. Vienna 1996 
  
Te 
fui 
gi