International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
No.of | Mean | RMS | <2.0 | 2.0- | >50 | Max.
Area Compared (m) (m) m &0m m (ni)
Points v
O+C+V+A | 29,210,494 -1.21 4.80 | 60.796 | 16.896 | 21.390 | 4242
O+C+A 17,610,588 | -1.11 | 2.91 | 77.096 | 13.996 | 10.196 | 358.9
O+A 14,891,390 | -1.24 | 2.77 | 79.8% | 12.2% | 8.0% | 358.9
O 11,795,795 -1.00 1.28 190.396 | 8.5% 1.2% | 37.33
Table 9. Accuracy measures and error classes for the triplet. O-Open areas; C-City areas; V-Tree areas; A-Alpine areas.
No. of E
Area Compared M RME ae 2 "8e M
Points wh
O+C+V 20,336,024 | 0.45 4.78 15777935 213% {209% | 125.2
O+C 13,496,226 | -0.33 | 3.38 | 68.7% | 20.8% | 10.3% | 47.34
O 3,969,734 -0.97 1.54 83% | 15.0% | 2.0% 39.4
Table 10. Accuracy measures and error classes for the stereopdir. O-Open areas; C-City areas; V-Tree areas.
with cooperative texture. In fact in these areas, the matching
accuracy was close to that of LIDAR.
5. CONCLUSIONS
The presented results verify that 3D points can be determined
with a submeter accuracy which for the planimetry can be 0.5
m or less, if accurate GCPs are used. This was achieved also
with non-GPS GCPs and in mountainous areas with not very
well defined GCPs. The number of GCPs can be small, their
accuracy being the main point. GCPs can cover only a portion
of the image, although caution should be paid in areas with
large height differences. QB is not as linear as IKONOS and to
achieve equal accuracy, needs an affine transformation after
employment of RPCs. The simple models (3D and 2D affine)
do not always perform well, thus use of RPCs should be
generally preferred. IKONOS and QB orthoimages have been
generated with an accuracy of 0.5 — 0.8 m, for typical sensor
elevations of 65-75 deg. This requires, however, an accurate
DSM/DTM. Sophisticated matching algorithms have derived a
5 m DSM with an accuracy of 1-5 m without editing and under
very difficult conditions. In spite of that, accuracy in open
textured areas was 1m or below. This potential has been very
little exploited up to now, especially with IKONOS which is
more stereo capable than QB, and presents an interesting
alternative technology for deriving DSMs. Future work will
focus on refinement of these investigations and possibly
processing of new better quality images in the Thun testfield.
ACKNOWLEDGEMENTS
We thank the Swiss Federal Office of Topography and the
NPOC, Bern for providing in Thun the laser DSM and in
Geneva the HRS images, Swissimage and the DHM25, the
Canton Geneva for providing the 25-cm orthoimages and the
laser DTM and Space Imaging USA for the IKONOS images in
Thun and the RPCs of the Geneva IKONOS images. We also
thank Oliver Heller and Oliver Gut, students at ETH Zurich,
for contributing in this work, especially for the Geneva data.
REFERENCES
Baltsavias, E. P., 1991. Multiphoto Geometrically Constrained
Matching. Ph.D. thesis, IGP, ETH Ziirich, Mitteilungen
No. 49, 221 p.
Baltsavias, E., Pateraki, M., Zhang, L., 2001. Radiometric and
Geometric Evaluation of IKONOS Geo Images and their
use for 3D building modelling. Proc. Joint ISPRS
Workshop "High Resolution Mapping from Space 2001",
Hannover, Germany, 19-21 September (on CD-ROM).
Fraser, C., Yamakawa, T., 2004. Insights into the affine model
for high-resolution satellite sensor orientation. ISPRS
Journal of Photogrammetry and Remote Sensing 58(5-6) (in
print).
Fraser, C.S, Baltsavias, E., Grün, A., 2002. Processing of
IKONOS images for submetre 3D positioning and building
extraction. ISPRS Journal of Photogrammetry & Remote
Sensing 56 (3), 177 — 194.
Grodecki, J., Dial, G., 2003. Block Adjustment of High-
Resolution Satellite Images Described by Rational
Functions. Photogrammetric Engineering and Remote
Sensing 69(1), 59-70.
Gruen, A., 1985, Adaptive Least Squares Correlation: A
powerful Image Matching Technique. South Africa Journal
of Photogrammetry, Remote Sensing and Cartography 14
(3), 175-187.
Heller, O., Gut, O., 2004. Auswertung hochauflósender
Satellitenbildern | IKONOS/QUICKBIRD. Report for
Vertiefungsblock, ETH Zürich, March.
Otto, G. P., Chau, T. K. W., 1988. A “Region-Growing”
Algorithm for Matching of Terrain Images. Proc. 4^ Alvey
Vision Club, University of Manchester, UK, 31 Aug. - 2
Sept.
Zhang, L., Gruen, A., 2003. Automatic DSM Generation from
StarImager (SI) Data. Proc. 6" Conference on Optical 3-D
Measurement Techniques, Gruen, A., Kahmen, H. (Eds.),
September 22-25, Zurich, Vol. 1, pp. 93-105.
Zhang, L., Gruen, A., 2004. Automatic DSM Generation from
Linear Array Imagery Data. IAPRS, Vol. XXXV, Part B3
(proc. of this Congress).
1256
