€7 
C 
C 
mework of 
d from the 
ince. It is a 
e HRS data 
xamine the 
stereo-pair 
as led us to 
ereo model 
points that 
points that 
) have been 
ction of an 
ed that the 
pic (HRS) 
de satellite 
quasi real 
ellite, with 
any ground 
HRS study 
this Study 
duction of 
{ Cadastre, 
University 
ive as Co- 
s itis a 
[he height 
tense. The 
gion about 
2002 (Fig. 
nal images 
(lach scene 
> accuracy 
> data set 
Figure 1. HRS-SPOTS stereo-pair on test site Manosque, France 
This means that it describes approximately only 595 of the area 
covered by the stereo-pair (Fig. 2). The reference DEM has a 25 
m grid spacing and Z accuracy better than 1m. Neither GCPs 
nor any kind of digital map were available. The software Leica 
Photogrammetry Suite was been utilized for the processing of 
the stereoscopic images. 
  
Figure. 2. The area of the reference DEM in the frame related 
to the test area DEM 
2.2 Procedure 
In this project many different DEM generation experiments 
have been carried out in order to examine the influence of 
different parameters and methods of processing to the final 
accuracy of the produced DEM. Eight (8) different approaches 
are presented here (table 1). 
The first criterion that has been examined was the radiometry of 
the stereo pair. The simultaneous acquisition of both images 
provides the radiometric resemblance of the images, in order to 
facilitate the automatic correlation process. Because the fact 
that the images are captured by two different sensors has led us 
to apply the histogram matching technique for image pre- 
processing, in some experiments. 
Sensing and Spatial Information Sciences, Vol XXXV, Part B1. Istanbul 2004 
    
  
  
  
  
  
  
  
   
  
  
    
    
   
  
  
  
  
   
  
  
  
  
  
  
  
  
   
    
  
   
  
  
   
   
  
  
  
   
   
   
   
    
   
   
  
   
    
   
  
Thus, 4 solutions concern a DEM generation from the original 
data (DEM1, DEM3, DEMS, DEM7) and 4 solutions (DEM2, 
DEM4, DEM6, DEMS) concern the generation of a DEM after 
an applied histogram matching technique. 
The second basic advantage of the HRS SPOT-5 images is that 
is possible to know, with absolute precision, the location of 
them. The information included in the metadata-ephemeris data 
leads automatically in the generation of a DEM, without the 
need of GPC’s. In this study, except this basic solution, other 
experiments have been carried out as well. These tests concern 
“semiautomatic” solutions in which a triangulation technique 
with automatic or semiautomatic methods of tie points 
determination have been applied before the generation of the 
DEM. Thus, the georeferencing and the stereo model set-up of 
the images was established in four different ways: 
a. By using only the metadata (solutions DEM 1, DEM2). 
b. After the use of the metadata, a triangulation was performed 
without any GCP. The tie points have been generated 
automatically (solutions DEM3, DEM4). 
c. The same as the above b solution but in the triangulation 
process 6 GCPs (XYZ coordinates) were used. For the 
extraction of the GCPs an additional procedure has been 
performed, which consisted basically of the production of an 
orthoimage based on the reference DEM. Draping the rectified 
image over the reference DEM, 6 GCPs, in a stereoscopic way, 
were collected for the triangulation procedure. (solutions 
DEMS, DEMO). 
d. The same as the above b solution but for the triangulation 
process except the automatically generated tie points, some tie 
points derived manually, in a stereoscopic way, has been used 
as well; the latest were points situated in critical positions such 
as in the steepest and the highest areas (solutions DEM7, 
DEMS). 
The results of the different experiments are represented in the 
table 1. The characteristics of the different experiments are 
interpreted below: 
No his: no histogram matching applied