• Stronger contrast by geometries than optical factors (illumi
nation, color pattern, etc.).
Some limitations of the current refinement method have been lo
cated at:
It cannot solve ambiguities caused by multiple lines with
similar geometry properties.
It cannot distinguish whether a model-image inconsistency
is caused by reconstruction errors or inaccurate exterior ori
entation.
Knowledge based reasoning of the image information is the key
to the first problem. The current matching stretchy is rather local.
Experiments show that the offset direction between the model
edges and their matched image lines are mostly same, which
is obviously caused by inaccurate exterior orientation. A globe
matching process (RANSAC over offsets for example) should be
able to estimate the correct exterior orientations.
7 CONCLUSIONS AND OUTLOOK
In this paper we present a model refinement method, which uses
the lines extracted from close-range images to improve building
models reconstructed from terrestrial laser point clouds. With
the refinement, several modeling errors caused by either gaps in
laser data or reconstruction algorithm, are corrected with image
information. Texturing is also improved after the refinement.
Nowadays it is more and more common for acquisition platforms
to acquire laser data and optical data simultaneously. Line ex
traction from images is very accurate, while laser points are more
suitable to extract planar features. Efficient fusing of laser points
and image naturally avoids many barriers for building reconstruc
tion from either sides. The attempt through our refinement method
shows promising future for automated building reconstruction by
fusing laser altimetry and optical methods.
Two directions of the future work: knowledge based image rea
soning and global matching, have been suggested earlier. Be
sides, nowadays the mainstream image acquisition systems usu
ally determine exterior orientations via GPS and IMU, but they
are not accurate. If we use the laser points as reference data,
and match image lines with model edges from laser point clouds
(similar to this research), there should be enough control points
for estimating the accurate exterior orientations for images.
ACKNOWLEDGEMENTS
The authors would like to thank Cyclomedia B.V. for providing
the Cyclorama data.
References
Brenner, C., 2005. Building reconstruction from images and laser
scanning. International Journal of Applied Earth Observations
and Geoinformation 6(3-4), pp. 187-198.
Canny, J., 1986. A computational approach to edge detection.
IEEE Transactions on pattern analysis and machine intelli
gence pp. 679-698.
Dick, A., Torr, P„ Ruffle, S. and Cipolla, R., 2001. Combin
ing single view recognition and multiple view stereo forarchi-
tectural scenes. In: Eighth IEEE International Conference on
Computer Vision, 2001. ICCV 2001. Proceedings, Vol.
Frueh, C., Jain, S. and Zakhor, A., 2005. Data processing al
gorithms for generating textured 3D building facade meshes
from laser scans and camera images. International Journal of
Computer Vision 61(2), pp. 159-184.
Frueh, C., Sammon, R. and Zakhor, A., 2004. Automated tex
ture mapping of 3D city models with oblique aerial imagery.
In: 3D Data Processing, Visualization and Transmission, 2004.
3DPVT 2004. Proceedings. 2nd International Symposium on,
pp. 396^103.
Pollefeys, M., Nister, D., Frahm, J., Akbarzadeh, A., Mordohai,
P., Clipp, B., Engels, C., Gallup, D., Kim, S., Merrell, P. et al.,
2008. Detailed real-time urban 3d reconstruction from video.
International Journal of Computer Vision 78(2), pp. 143-167.
Pu, S. and Vosselman, G., 2009. Knowledge based reconstruc
tion of building models from terrestrial laser scanning data (in
press). ISPRS J. Photogramm. Remote Sens.
Schindler, K. and Bauer, J., 2003. A model-based method for
building reconstruction. In: First IEEE International Work
shop on Higher-Level Knowledge in 3D Modeling and Motion
Analysis, 2003. HLK 2003, pp. 74-82.
Schneider, D. and Maas, H., 2003. Geometric modelling and
calibration of a high resolution panoramic camera. Optical 3D
Measurement Techniques VI (Eds. Gruen, A. and Kahmen, H.)
2, pp. 122-129.
van den Heuvel, F., Verwaal, R. and Beers, B„ 2007. Automated
Calibration of Fisheye Camera Systems and the Reduction of
Chromatic Aberration. Photogramm. Fernerkund. Geoinf. 3,
pp. 157.
Vosselman, G., 2002. Fusion of laser scanning data, maps,
and aerial photographs for building reconstruction. In: 2002
IEEE International Geoscience and Remote Sensing Sympo
sium, 2002. IGARSS’02, Vol. 1.