ikodate 1998 
| conditions, 
c system is 
anner and a 
d to discuss 
d by bundle 
y solved by 
iguration of 
ust enable 
  
  
tric 
scribes the 
ne following 
ogrammetric 
top image 
igitized with 
Finally, the 
ormed with 
e integrated 
mbines the 
nology with 
2 SYSTEM DESCRIPTION 
24 Image Recording and Digitization 
All projects have adopted the Rollei 6006 medium-format 
metric camera, with focal lengths between 40mm and 
150mm using ILFORD FP4 b/w film. Its geometric 
resolution is estimated around 100 Ip/mm. The original 
negatives have been enlarged to paper prints of 18cm x 
18cm (scaling factor 3). 
A simple desktop scanner, the HP ScanJet, has been 
used for image scanning. Geometric resolution (400 dpi), 
radiometric resolution (8 bit) and geometric accuracy (30- 
50um) do not match the performance of the camera/film 
system. However, scanning of enlarged reseau images 
yields to an appropriate pixel size (approx. 21yum in the 
original film negative) and enables the numerical 
correction of almost all geometric failures down to image 
accuracies of 10um (Bendler 1994). 
Recently a second image scanner, the Agfa DuoScan, 
has been used (600 dpi, 8 bit). As compared with the HP 
ScanJet is shows a significantly worse accuracy with a 
strong non-linear behavior. Later results of bundle 
adjustment have shown that the Agfa system does not 
fulfil the requirements even for reseau images. 
2.2 Image Archiving 
Large and complex image configurations (e.g. >100 
the images. For this purpose an image data base has 
been developed using MS-ACCESS which allows the 
storage of relevant image information either during image 
acquisition on-site, or during later scanning. The situation 
in the field is described by relatively simple imaging 
attributes (e.g. north, south, top, bottom ..) and the 
parameters of exterior orientation. The object itself is 
defined by a number of predefined object attributes such 
as ‘main arch’, ‘west entrance’, ‘dome’ and so on. 
Figure 2 and Figure 3 show examples of data structures 
and report generation. 
{Neben_zwei 
Abbildung 
Orientiert 
Objektteilbeschreib ansicht 
  
  
  
  
  
  
pd X 
Wie [welches_Objekttei [Art_der_Ansicht] 
  
  
  
Figure 2: Example of data relations of the image data 
  
images) require systematic management and archiving of base 
Bericht Bericht Hauptabfrage 
Bildnr: 1 Kamera: 150 Orientiert: Ja Abbildung 
Filmnr: 8 
Negativnr: 1 
Dateiname: F8B1.TIF Boe: 
CD-ROM: P ADOLPH2 
Obiektteibeschreibung  Hauptbogen 
ansicht: Seitenansicht 
Ost 
  
Bildnr: 2 Kamera: 150 
Filmnr: 8 
Negativnr: 3 
Dateiname: F8B3TIF lage: 
CD-ROM: P ADOLPH2 
Orientiert: 
Obiektteilbeschreibung  Hauptbogen 
ansicht: Seitenansicht 
Ost 
Ja Abbildung 
  
Bildnr: 9 Kamera: 150 
Filmnr: 8 
Negativnr: 14 
Dateiname: F8B14 TIF lage: 
CD-ROM: P ADOLPH2 
Orientiert: 
Obiektteilbeschreibung, Hauptbogen 
ansicht: Seitenansicht 
Ost 
Ja Abbildung 
  
  
  
  
HT Seite] PIM] [wi 
VEIT 
Figure 3: List of images matching the query statement 
2.3 Photogrammetric Processing 
The interactive multi-image compiling systems PHIDIAS 
(Benning & Effkemann 1991), PHIDIAS-MS (Benning 
199???) and Rollei CDW (Godding & Fellbaum 1995) 
have been used. In order to transfer data between both 
Systems special data conversion programs have been 
developed. The first version of PHIDIAS was restricted to 
237 
a two-image compiling mode.The newer version of 
PHIDIAS-MS is integrated into the Microstation 3-D CAD 
environment and is suited for interactive processing of up 
to 8 images at once. The CDW package benefits from 
robust estimation of approximate values which is 
otherwise an extremely time-consuming task in complex 
applications with a large number of images (Godding & 
Fellbaum 1995).