CIPA 2003 XIX th International Symposium, 30 September - 04 October, 2003, Antalya, Turkey 
• « 
Figure 6: Two views of the rectified part of the façade that 
contains the 22 corresponding points. 
4. CONCLUSIONS 
The processing of images of the CIPA reference data set has 
demonstrated the capabilities and limitations of the developed 
methods for camera calibration, image orientation, and object 
reconstruction. Object reconstruction using the same four 
images as used for the orientation experiments is presented in 
(Hrabacek and Heuvel, 2000). Figure 7 shows a view of the 
reconstructed model that is represented in VRML format. All 
methods used have the following characteristics in common: 
The incorporation of knowledge on the construction of 
the building 
The use of image line features, manually or 
automatically extracted 
The use of a limited number of images 
Calibration of the camera was performed in a semi-automatic 
way with five images of the reference data set. Estimation of 
interior orientation parameters from single images showed that 
two zoom settings were present in the set of selected images. 
The quality of the results of single image camera calibration 
depends greatly on the image characteristics. For a full and 
accurate camera calibration, several images with different 
orientation relative the building are needed. Correspondence 
between the images is not required. 
The method for image orientation aims at automation. Tests with 
images from the CIPA data set show that correct relative 
orientation can be obtained, but as in the method for automated 
camera calibration, the chance of success greatly depends on the 
image configuration and characteristics. Especially the repeating 
structures (windows) in the façades lead to an ambiguous 
problem. Between two and three manually measured 
corresponding points were needed to obtain a correct and 
reliable solution, and at the same time allowed to determine the 
relative scale of consecutive models. The paper shows that 
automatic relative orientation of two images is possible. 
However, success depends on the settings of a few parameters. 
A partial object model results as a by-product of the methods for 
image orientation. 
Figure 7: Reconstructed model, partially texture-mapped. 
REFERENCES 
Heuvel, F.A. van den, 1998. Vanishing point detection for architectural 
photogrammetry. H. Chikatsu and E. Shimizu (Editors). 
International Archives of Photogrammetry and Remote 
Sensing, Vol. 32 part 5, Hakodate, Japan, pp. 652-659. 
Heuvel, F.A. van den, 1999a. Estimation of interior orientation 
parameters from constraints on line measurements in a single 
image. P. Patias (Editor). International Archives of 
Photogrammetry and Remote Sensing, Vol. 32 part 5W11, 
Thessaloniki, Greece, pp. 81-88. 
Heuvel, F.A. van den, 1999b. A Line-photogrammetric mathematical 
model for the reconstruction of polyhedral objects. S.F. El- 
Flakim (Editor), Videometrics VI. proceedings of SPIE. 
SP1E, Vol. 3641, San Jose, USA, pp. 60-71. 
Heuvel, F.A. van den, 2001. Object reconstruction from a single 
architectural image taken with an uncalibrated camera. 
Photogrammetrie Fernerkundung Geoinformation, 2001(4): 
247-260. 
Heuvel, F.A. van den, 2002. Towards automatic relative orientation for 
architectural photogrammetry. P. Patias (Editor). 
International Archives of Photogrammetry and Remote 
Sensing, Vol. 34 part 5, Corfu, Greece, pp. 227-232. 
Hrabacek, J. and Heuvel, F.A. van den, 2000. Weighted geometric 
object constraints integrated in a line-photogrammetric 
bundle adjustment, ISPRS congress. International Archives 
of Photogrammetry and Remote Sensing, Vol. 33, part B5, 
Amsterdam, pp. 380-387. 
Rottensteiner, F., Grussenmeyer, P. and Geneva, M., 2001. Experiences 
with the digital photogrammetrie program package Orpheus 
based on CIPA's "Zurich city hall" dataset for architectural 
photogrammetry, XVIII CIPA Int. symposium, proc, 
Potsdam, Germany, pp. 639-646. 
Streilein, A., Hanke, K. and Grussenmeyer, P., 2000. First experiences 
with the "Zurich city hall" data set for architectural 
photogrammetry, ISPRS congress. International Archives of 
Photogrammetry and Remote Sensing, Vol. 33 part B5, 
Amsterdam, pp. 772-779.