International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
Figure 6 gives a typical example of the effects of the post-
processing operations. It demonstrates the necessity of both,
the choice of elevation data for delineating buildings rather
than image data (where shadows often extend or falsify the
outline; indicated by the upper arrow), and the dilation
operation which leads to a rather good coincidence with
manually digitised edges. However, remaining errors occur
which are mainly due to the input data quality (indicated by the
lower arrow). It has to be noted that so far only few and simple
post-processing operations have been applied.
5. CONCLUSIONS
Multi-sensor data yield the advantage of producing more scene
features, however, the crucial point in the interpretation
workflow is still the big, application dependent gap between
these features and the related object characteristics. Hence, the
central goal of this contribution was to present a methodology
that bridges this gap for an important application, the
extraction of topographic surface edges with the emphasis on
building edges. In contrast to other methodologies for object
edge extraction we propose an algorithm into which several
features of both, the multiple reflections of the laser scanning
system and the multi-spectral imagery, have been introduced.
With that not only geometrical but also semantical information
are used. As we concentrate on the derivation of building
edges, we applied a region growing algorithm, which by-
passes the problem of linking detected edge pixels to
connected lines.
So far we have presented first qualitative results of our
approach. It could be shown that the simultaneous usage of
geometrical and semantical information definitively improves
the classification and delineation of object edges compared to
the single use of only one source. However, we have still to
state some imperfect results which are mainly due to the
quality of the input data in terms of the point density of the
laser scanning elevations and the image acquisition date which
has led to rather less predictive NDVI values.
In future several improvements of the methodological
components will be addressed. For instance, we will apply a
multi-scale segmentation instead of the use of a single segment
layer. Additionally, the edge image matching process will be
further developed by introducing even more features. Finally,
the methodology will be extended in order to extract further
surface edge classes (like embankments or ditches).
REFEERENCES
Baatz, M., & Schipe, A., 2000. Multiresolution segmentation —
an optimization approach for high quality multi-scale image
segmentation. In: Strobl, J., Blaschke, T. & Griesebner (Eds. ):
Angewandte Geographische Informationsverarbeitung XII,
Wichmann, pp. 24-29.
Briigelmann, R., 2000. Automatic breakline detection from air-
borne laser range data. In: Int. Archives of Photogrammetry
and Remote Sensing (Amsterdam), 33(B3), pp. 109-116.
Brunner, K., 1986. Bruchkante oder Geländekante? Bild-
messung und Luftbildwesen, 54(3): p. 132.
Chakreyavanich, U., 1991. Regular Grid DEM Data
Compression by Using Zero-Crossings: The Automatic
Breakline Detection method. PhD these (Ohio State
University), Report OSU-DGSS, no. 412., 90 pp.
Gaisky, D., 2000. Geomorphologic improvement of DTMs
especially as derived from laser scanner data. In: Int. Archives
of Photogrammetry and Remote Sensing (Ljubljana), 32(6W8).
Kraus, K. & Pfeifer, N., 2001. Advanced DTM Generation
from Lidar Data. In: Int. Archives of Photogrammetry and
Remote Sensing (Annapolis, Maryland), 34(3/W4), pp. 23-30.
Loffler, G., 2003. Aspects of raster DEM data derived from
laser measurements. In: Int. Archives of Photogrammetry and
Remote Sensing (Dresden), 34(3/W13).
Lohr, U., 2003. Precise LIDAR DEM and True Ortho Photos.
In: Fritsch, D. (Ed.): Photogrammetric Week '03, Wichmann,
pp. 111-116.
Pavlidis, T. & Liow, Y.-T., 1990. Integrating Region Growing
and Edge Detection. /EEE Transactions on Pattern Analysis
and Machine Intelligence, 12(3), pp. 225-233.
Petzold, B., Reiss. P. & Stossel, W., 1999. Laser scanning —
surveying and mapping agencies are using a new technique for
the derivation of digital terrain models. ISPRS Journal of
Photogrammetry & Remote Sensing, 54, pp. 95-104.
Pfeifer. N. & Stadler, P., 2001. Derivation of digital terrain
models in the SCOP++ environment. Proceedings OEEPE
Workshop on Airborne Laserscanning and Interferometric SAR
for Detailed Digital Elevation Models (Stockholm).
Schenk, T., 2003. From Data to Information and Knowledgde.
In: Schiewe et al. (Eds.): Challenges in Geospatial Analysis,
Integration and Visualization ll. Proceedings of Joint ISPRS
Workshop, Stuttgart (CD-ROM): 105-110.
Schenk, T. & Csatho, B., 2002. Fusion of LIDAR and aerial
imagery for a more complete surface description. In: Int.
Archives of Photogrammetry and Remote Sensing (Graz),
34(3A), pp. 310-317.
Schiewe, J., 2003. Integration of multi-sensor data for land-
scape modeling using region-based approach. ISPRS Journal of
Photogrammetry and Remote Sensing, 57 (5-6), pp. 371-379.
Schiewe, J., 2004. Status and future perspectives of the
application potential of digital airborne sensor systems.
International Journal of Applied Earth Observation and
Geoinformation (in Review).
Wild, D., Krzystek, P. & Madani, M., 1996. Automatic
breakline detection using an edge preserving filter. In: /nt.
Archives of Photogrammetry and Remote Sensing (Wien),
31(B3), pp. 946-952.
ACKNOWLEDGEMENTS
Presented data are courtesy of TopoSys GmbH -
(www.toposys.com).
610
KEY
ABS!
Curre
Geom
organi
comp:
increa
detect
of aut
when
recogi
and e:
proces
Two «c
demor
The C
Natura
initiati
Progra
Chang
conduc
spatial
mappit
extract
correct
involve
Nowad
expect
This is
techno.
continu
size of
mappir
cost in
the cur
higher
Autom:
extracti
*
Cort