bid 
Vi wd wl 
)0 
———— 
1000 
> to 
International Archives of the Photogrammetry, Remote Sensing 
After resampling the DSM to a pixel spacing of 10 m, the 
comparison with the DEM was evaluated in order to emphasize 
possible systematic or gross errors. As shown in the frequency 
histogram of figure 8 the distribution of differences shows a 
distribution with the 90% of pixel exhibiting a difference with 
the reference elevation of + 20m. 
FREQUENCY (%) 
co 
      
6 
4 
be 
ols ^ n 
$995 8v29-7-22 9259255535 
"eg oe agen egt ERN 
CLASSES (meters) 
Figure 8. Frequency histogram showing differences in the 
comparison between the satellite-derived DSM and the DEM 
used as reference 
4. ORTHOIMAGE GENERATION 
The orthorectification of the QuickBird image, based upon the 
discussed DSM, has been performed using PCI Geomatica 
software. The software adopts different geometric correction 
models, among them the parametric rigorous model and the 
rational polynomial model are the most accurate. 
The rigorous model can be applied through the knowledge of a 
certain number (around 20) of ground control points well and 
evenly distributed over the whole scene. 
The Rational Polynomial Model can be theoretically applied 
without knowing ground control points, but just using the 
coefficients (RPC, Rational Polinomial Coefficient) delivered 
with metadata. In this work the transformation between UTM- 
WGS84, that is the reference system in which RPC are 
computed, to the National Gauss Boaga grid system, that is the 
final required system, has been applied using a set of GCP in 
the latter system. 
It has to be noticed that the extension of the QuickBird scene is 
wider than the EROS one, so the orthorectification could be 
carried out only for the overlapping area. 
4.1  Orthoimage accuracy assessment 
Results of the orthorectification tests are generally expressed in 
terms of Root Mean Square Error (RMSE) along the East and 
North grid axis computed using a series of known and 
independent checkpoints which are clearly recognizable on the 
orthoimage. Several test have been experienced in the past by 
the authors (Barbarella, 2003). On the other hand, it is well 
known from literature that the accuracy achieved in the 
orthorectification may be fixed in 1 or 2 pixels. 
The overall quality of the final orthoimage map could be also 
checked by the comparison with cartography at the higher scale 
than possible. The "first look" comparison between the raster 
orthoimage and the vector map available, with its related and 
scale-dependent accuracy, would be useful in the identification 
of critical area where the procedure fails. In figure 9 the 
overlapping of raster orthoimages and vector layer is shown. 
and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004 
  
Figure 9. Overlay of portion of the raster orthoimage and the 
map layer. Comparison of the main features highlights 
differences between products 
5. CONCLUSIONS 
The generation of Digital Surface Models from satellite high 
resolution optical images and its use in the ortho-reprojection of 
external HR satellite images (we discussed the case of a 
QuickBird image) should be considered as a productive 
methodology when the accuracy requested in map production 
has to meet the requirements of cartography at scale as large as 
1:10000. In particular, within the spatial methodology for 
terrain modelling, the use of stereo-pairs acquired by the EROS 
constellation may be considered one the possible choice, 
considering the low cost of the raw data. Problems may arise in 
the correlation procedure (low data quality, cloud coverage, 
atmospheric effects or shadowing in densely urbanized area), 
producing lack of data in the DSM or erroneous elevations. 
REFERENCES 
Barbarella, M., Mancini, F., Zanni, M., 2003. Rectification of 
high resolution satellite data: evaluating accuracy for map 
updating. Proceedings of the ASPRS/MAPPS conference 
"Terrain data: application and visualization, making the 
connection", Charleston, SC, USA, Oct. 2003. 
Barbarella, M., Mancini, F., Zanni, M., 2003. Processing of 
high resolution satellite data for map updating. Proceedings of 
30th International Symposium on Remote Sensing of 
Environment, Honolulu, Hawaii, USA, Nov. 2003. 
Hirano, A., Welch, R., Harold, L., 2002. Mapping from ASTER 
stereo image data: DEM validation and accuracy assessment. 
ISPRS Journal of Photogrammetry and Remote Sensing, 57, pp. 
356-370. 
Jacobsen, K., 2003. Orthoimages and DEMs by QuickBird and 
Ikonos. Proceedings of EARSeL: “Remote Sensing in 
Transition", Ghent, 2003, pp.513 — 525 
Siva Subramanian, K.S., Amitabh, S., Manda, S., 2003. 
Evaluation of Digital Elevation Models created from different 
satellite images. Proceedings of Map India Conference 2003, 
New Delhi, India, Jan 2003 
http://www. gisdevelopment.net/proceedings/mapindia/2003/tt. ht 
m (accessed 27 Apr. 2004) 
1053