ighbouring, 
vn in figure 
tions. 
  
| mixes the 
eft), due to 
vegetation. 
considering 
/ low. The 
trongly im- 
5. 
hm reaches 
esidual val- 
ility in the 
m inappro- 
the image 
1S residuals 
in the RMS 
1 about 2% 
ignificantly 
Significant 
for 1:5,000 
information 
ctures. 
systematic 
iges of the 
igated, but 
be verified 
the digital 
for exterior 
workstation 
1tersections 
any kind of 
: automatic 
  
  
      
  
  
  
  
  
  
  
  
RMS differences RMS residuals 
Type/level Angles XYZ Image points | Control points H, V | Blunders 
[gon] [m] [um] [m] [^6] 
Reference 0.0000 | 0.000 2.7 0.046, 0.086 0.0 
Approx./starting val. | 2.6569 | 28.520 — — v 
Level 5, 480 um 0.4027 | 4.881 83.4 0.035, 0.030 15.0 
Level 4, 240 jm 0.1151 | 1.344 35.6 0.132, 0.170 2.1 
Level 3, 120 um 0.0477 | 0.639 39.4 0.088, 0.141 1.9 
Level 2, 60 um 0.0475 | 0.620 6.6 0.107, 0.144 10.0 
Level 1, 30 um 0.0263 | 0.355 4.1 0.148, 0.187 2.0 
Level 0, 15 um 0.0309 | 0.379 3.3 0.127, 0.106 2 
  
  
  
  
Table 2: Results for the image pyramid iterations. 'RMS differences' denote RMS differences between reference orientation 
parameters and automatically computed parameters in each image level. 'RMS residuals' denote RMS residuals in image arid 
object space, and is a result of inaccuracies in the digital map and the automatic image measurements. 
orientation and subsequent computation of orthophotos. 
7 CONCLUSION 
If the results in this test can be verified by further testing, 
the method presented in this paper may be profitably applied 
to map revision, orthophoto production, and thematic map- 
ping. Even better results may be reached by extending the 
matching algorithm to smaller object types, e.g. manhole 
covers and gratings. The potential benefit is to avoid signal- 
ization, surveying, aerotriangulation and any extra equipment 
like GPS/INS. 
The proposed method for exterior orientation has been ap- 
plied to 1:5,000 photographs and large scale digital topo- 
graphical/technical T3 maps. It seems evident that T3 maps 
are sufficiently accurate for this purpose. 
Problems may occur in case of many similar looking road in- 
tersections. In densely built city areas, occlusions and shad- 
ows may give rise to many blunders. Pavements, of which 
there are very few in the test area, has no specific code in the 
map data, and may be found difficult or impossible to handle. 
For rural areas, the presented approach should be applied to 
1:25,000 photographs and TO/TK1 map data. 
References 
[Ackermann, 1995] Friedrich Ackermann (1995). Digitale 
Photogrammetrie - Ein Paradigma-Sprung. Zeitschrift für 
Photogrammetrie und Fernerkundung, 3/1995, pp. 106- 
115. 
[Fórstner, 1988] Wolfgang Fórstner (1988). Model Based 
Detection and Location of Houses as Topographic Control 
Points in Digital Images. International Archives of Pho- 
togrammetry and Remote Sensing, 1988, Vol. 27, Part 
B11, Supplement, pp. 111505-111517. 
[Gülch, 1994] Eberhard Gülch, (1994). Using Feature Ex- 
traction to Prepare the Automated Measurement of Con- 
trol Points in Digital Aerial Triangulation. ISPRS Comm. 
Ill Symposium, München. International Archives of Pho- 
togrammetry and Remote Sensing, Vol. 30, Part 3/1, pp. 
333-340. 
[Heikkinen, 1994] Jussi Heikkinen (1994). Linear Feature 
Based Approach To Map Revision. Proceedings of the IS- 
PRS Commission IV Symposium, Athens, Georgia, USA, 
pp. 344-351. 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996 
[Hellwich et a/., 1994] O. Hellwich, C. Heipke, L. Tang, H. 
Ebner, W. Mayr (1994). Experiences with Automatic Rela- 
tive Orientation. ISPRS Comm. Ill Symposium, München, 
International Archives of Photogrammetry and Remote 
Sensing, Vol. 30, Part3/1, pp.370-378. 
[Kryzstek, 1995] Peter Kryzstek (1995). Generation of Digi- 
tal Elevation Models. Second Course in Digital Photogram- 
metry, lecture notes, Institut für Photogrammetrie, Univer- 
sitát Bonn, February 1995. 
Schriftenreihe des Instituts für Photogrammetrie der Uni- 
versitát Stuttgart, Heft 15, pp. 203-214. 
[Polis et al., 1995] Michael F. Polis, Stephen J. Gifford, 
David M. McKeown (1995). Automating the Construction 
of Large-Scale Virtual Worlds. Computer, Vol.28, Number 
7, pp. 57-65. 
[Schickler, 1992] Wolfgang Schickler (1992). Feature Match- 
ing for Outer Orientation of Singles Images Using 3-D 
Wireframe Controlpoints. International Archives of Pho- 
togrammetry and Remote Sensing, Vol. 29, Part B3, pp. 
591-598. 
[Schickler, 1995] Wolfgang Schickler (1995). Ein opera- 
tionelles Verfaren zur automatischen inneren Orientierung 
von Luftbildern. Zeitschrift fiir Photogrammetrie und Fern- 
erkundung, 3/1995, pp. 115-122. 
[Schwarz, 1995] Klaus-Peter Schwarz (1995). Integrated 
Airborne Navigation Systems for Photogrammetry. 
Fritsch/Hobbie (Eds.), Photogrammetric Week '95, 
Wichmann Verlag, pp. 139-153. 
[Tsingas, 1991] V. Tsingas (1991). Automatische Aerotrian- 
gulation. Schriftenreihe des Instituts fiir Photogrammetrie 
der Universitat Stuttgart, Heft 15, pp. 253-268. 
[Vosselman & Haala, 1992] George Vosselman and Norbert 
Haala (1992). Erkennung Topographischer PaBpunkte 
durch relationale Zuordnung. Zeitschrift fiir Photogram- 
metrie und Fernerkundung, 6/1992, pp. 170-176.