he
in
ot
to
he
or
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
new and existing sensors, and a growing number of “RFM
enabled” commercial tools are made available.
This paper discussed the single image based 3-D feature
extraction and compared these results with those measured
using stereo images. It has also demonstrated the application of
single image processing for 3D city model generation. It should
be noted that single image 3D reconstruction methods are of
great importance in the case of the HRSI due to the somewhat
limited availability of stereo pairs.
Although single image height measurement and 3D
reconstruction offer greater flexibly and lower costs, human
measurement errors, such as misidentification of the base or
roof of a building in one image, can affect the accuracy of
height measurement. It is therefore important to carefully
design and provide utilities to assist users on performing 3D
extraction from the single imagery.
The paper also discussed the extension of the RFM to aerial
imagery. This included the generation of the RFM coefficients
based on the known exterior orientation using the terrain
independent scheme. Overall, these preliminary results suggest
that sub-meter accuracies can be achieved.
Map in GSC Single (m) Stereo (m)
Terrain Roof Height Height Base Terrain Heishi
elevation elevation relative to relative to Height elevation Height elevatio eS
No. (feet) (feet) terrain (m) base (m) (MSL) n comcetion
l 236 357 36.9 36.3 37.8 73.7 36.2 73.1 0.6
2 235 307 49.4 47.5 49.2 73.7 48.3 . 71.8 ].9
3 236 521 86.9 86.2 87.3 73.3 86.0 72.6 0.7
4 236 521 86.9 86.2 87.2 73.3 86.1 72.6 0.7
5 232 554 98.2 92.1 91.9 77.0 91.2 70.9 6.1
6 232 327 29.0 22.9 23.7 77.0 22.8 70.9 6.1
7 232 335 31.4 25.3 26.1 77.0 25.2 70.9 6.1
8 232 350 ui 36.0 29.9 31.0 7.0 30.0 70.9 6.1
9 240 353 34.5 33.2 33.5 75.4 33.1 74.1 1.3
10 240 378 42.1 40.8 41.9 783... A10 74.1 1.3
11 235 453 00.5 = 65.0 65.7 73.1 65.5 71.6 aan 3
12 235 470 71.6 70.1 68.9 73.1 70.2 71.6 1.5
Table 4. Building Heights obtained in three ways
Stereo vs. Single vs. Single vs. Grodecki, J., Dial, G., 2003. Block adjustment of high
Map Map Stereo resolution satellite images described by rational polynomials.
Mean 2.82 2,27 0.55 PE&RS, 6001); pp. 59-68.
Std 2.63 2.64 1.03
RMS 3.78 3.39 113 Hu, Y., Tao, V., 2002. Updating solutions of the rational
Max Abs. 7.00 6.30 1.80 function model using additional control information. PE&RS,
Table 5. Differences with systematic vertical shifts (m)
Stereo vs. Single vs. Map vs.
Corrected Corrected Corrected
map map map
Mean -0.01 0.56 2.83
Std 0.41 0.98 2.45
RMS 0.39 1.09 3.67
Max Abs. 0.90 1.70 6.10
Table 6. Differences without systematic vertical shifts (m)
5. REFERENCES
Croitoru, A. Tao, V., Hu, Y., Xu, J. Wang, F. and Lenson, P.,
2004. The rational function model: a unified 2D and 3D spatial
data generation scheme. ASPRS annual Conference, 23-28
May, Denver, Colorado, 11 pages
Fraser, C., Hanley, H., 2003. Bias compensation in rational
functions for Ikonos satellite imagery. PE&RS, 69, pp. 53-57.
1027
68(7), pp. 715-724.
Hu, Y., Tao, V., Croitoru, A., 2004. Understanding the rational
function model: methods and applications. ZJAPRS, 12-23 July,
Istanbul, vol. XX, 6 p.
NIMA, 2000. The Compendium of Controlled Extensions (CE)
for the National Imagery Transmission Format (Version 2.1).
(http://164.214.2.5 1/ntb/baseline/stdi0002 1 /stdi0002 1 .pdf)
Tao, V., Hu, Y., 2001. A comprehensive study on the rational
function model for photogrammetric processing. PE&RS,
67(12), pp. 1347-1357.
Tao, V., Hu, Y., 2002. 3-D reconstruction algorithms with the
rational function model. PE&RS, 68(7), pp. 705-714.
Tao, V., Hu, Y., 2004. RFM: an open sensor model for cross
sensor mapping. ASPRS annual Conference, 23-28 May,
Denver, Colorado, 9 pages.