go- 
the 
int 
ne- 
da- 
/Se- 
reo 
reo 
'Te- 
ital 
ing 
ped 
ut- 
1et- 
nal 
am 
UM 
ters 
ari- 
Lain 
y of 
ılti- 
tive 
)po- 
"ing 
rne 
size 
ZZ 
I ZZ = E 
a 
: 
7 
  
  
Figure 1: Axonometric oblique view of the map- 
derived DEM indicated in the location map. 
e a panchromatic SPOT 1 HRV-2 image with a 
nominal pixel size of 10 meters and nadir-looking 
imaging mode (denoted as SPOT); 
e KFA-1000 stereo photographs (KFA-1, KFA-2) 
taken from the Sojuz space station at an alti- 
tude of approximately 285 km in a scale of about 
1 : 280 000. For digital mapping, the image da- 
ta were digitized with a nominal pixel size of 7 
meters on ground. 
e Airborne three-line scanner imagery using the 
MEOSS (Monocular Electro Optical Stereo Scan- 
ner, cf. Lanzl, 1986 [5]) scanner, taken at an 
altitude of about 11.3 kilometers, with a nomi- 
nal pixel size equivalent to 2 meters on ground. 
Three images, denoted as MEOSS-F, MEOSS-N 
and MEOSS-B, are acquired from one overflight 
in forward-, nadir- and backward- looking mode, 
where the off-nadir looking angles are +/-23 de- 
grees. 
e A scanned aerial image pair (AIR-1, AIR-2) tak- 
en from an altitude of about 5 kilometers above 
ground in a scale of about 1 : 30 000, with a pixel 
size equivalent to about 1.5 meters on ground. 
The coverage of the DEM (see Figure 1) and of the 
airborne image data is indicated in the location map 
shown in Figure 2. These only in part cover the DEM, 
185 
A Dachstein 
  
  
  
  
  
A 
/ Ds 
/ E 
/ e 3 = 
to Bes 
0) ; & = 
S / < < 
/ I 
/ / 
/ 7 
/ / 
Y y Schladming 
River ERES / | [^ 
Eu. NI / 
S / | 
  
L. 
4) 
= 
Digital Elevation Model 
  
  
  
Figure 2: Topographic location map indicating the 
DEM and the airborne image data coverage. 
whereas full coverage is available by any of the space- 
borne image data. Hence, subframes of this DEM have 
been resampled to 10, 5 and 2.5 meters, respective- 
ly, suitable for mapping purposes based on the multi- 
resolution image data. 
Ground control points (GCPs) were measured in to- 
pographic 1 : 25 000 map sheets, in existing 1 : 10 000 
ortho-photo maps and in the digital images as well. 
On the one hand, these points were then used to de- 
termine the geometric mapping parameters for the in- 
dividual images and, on the other hand, to investigate 
the pointing accuracy of single images and of stereo 
pairs. It is obvious, that this is restricted in image 
and on ground by the pixel size and the scale of the 
reference maps, respectively. 
3 DATA GEOCODING 
3.1 Evaluation of Pointing Accuracy 
  
Based on the GCPs measured on ground and in the 
images parametric imaging models were determined 
and optimized for each of the individual images us- 
ing a least squares parameter adjustment implement- 
ed in the software package RSG (Raggam et al., 1992 
[9]). Subsequently, the geometric performance of these 
imaging models was determined from image residual 
vectors, resulting from the transformation of control 
points from map geometry to the image coordinate 
system. 
Statistical entities of these residuals can be consid- 
ered as representative accuracy parameters, the re-