International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part Bl. Istanbul 2004 
  
His: histogram matching applied 
Metadata: only metadata has been used 
Auto tie: metadata and triangulation based on automated 
generated tie points 
Auto & manual tie: metadata and triangulation based on 
automated and manually collected tie points 
Auto tie & 6 GCPs: metadata and triangulation with auto tie 
points and 6 GCPs, derived after a particular procedure. 
2.3 Assessment of DEM generation 
Firstly, it is reminded that the results of table 1 concern the 
creation of a total DEM, that is to say, the DEM that occurred 
from the processing of the stereo - pair. In Figure 2 is presented 
the DTMI that concerns the basic solution. 
influence the results, since the radiometry of this stereo-pair 
was good. The mean elevation in all the solutions was 
determined in 367 m and only in the cases where the 6 GCPs 
were used, the mean elevation was calculated in 371 m. In these 
solutions they were determined the minimum (-37.8 m) and the 
maximum value (9.9 m) for the min point mass elevation. This 
fact was rather expected since the accuracy and the place of the 
GCPs was not suitable. 
2.4 DEM evaluation 
The accuracy assessment of the produced DEM; that generated 
by different experiments, was realized by using the reference 
data set that describes an area of roughly 6Km x 66Km. The 
statistical results of comparison are given in the table 2. 
  
   
     
  
   
    
  
   
  
    
   
  
  
  
  
  
  
  
  
  
  
  
  
  
   
  
  
   
   
  
  
   
  
  
  
  
  
  
  
  
  
  
   
  
  
  
  
  
  
   
  
  
  
  
Min Max 
Excellent mass mass Mean 
% point point elevation 
elevation | elevation (m) 
(m) (m) 
DEMI1 93 c LA 
No His 93.86 0.7 1397.4 367.3 
Metadata 
DEM? 93.71 -0.7 1401.9 367.2 
His 
Metadata 
DEM3 a 
$ 395 2 2 
No His 93.85 0.7 1395.6 367.3 
Auto tie 
DENY 93.71 1.50 1393.6 366.5 
His 
Auto tie 
DEMS 
No His 92 02 ATS 23 amio 
Auto tie 93.93 37.8 1402.3 371 2 
& 6 GCPs 
DEM6 
His A rA = A 
Auto tie 93.73 9.9 1399.1 370.8 
& 6 GCPs 
DEM7 
No his à = 
Anto d 93.85 -0.7 1395.5 367.2 
Manual 
tie 
DEMS 
His 
3. 4 1393. 366.5 
Auto& 93.71 1 93.8 366 
Manual 
tie 
  
  
  
  
  
  
  
   
   
   
  
   
  
  
  
  
Table 1. The results of DEMS generated in different cases 
The following analysis concerns the examination of the 
influence of different methods of processing to the final DEM. 
In all the solutions the matching was done with the same high 
precision. Thus, the localization of the 94% of points is 
characterized as excellent. Consequently, the preprocessing of 
the images with histogram matching does not appear to 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
Min Mean 
Max Mean error Absolute RMS 
error (m) error (m) 
m) (m) 
DEMI 418.7 
= 322 
No his 166.4 0.6 7.1 13.2 
Metadata 
DEM2 -119.3 ; 
His 169.8 -0.3 7.6 14.8 
Metadata 
DEM3 -118.9 
s aio 
No his 165.6 0.6 71 13.2 
Auto tie 
DEM4 -119.6 
His 170.6 -0.4 7.6 14.8 
Auto tie 
DEMS 
No his -116.2 ; 
Auto tie 165.0 -0.2 7.6 13.4 
& 6 GCPs 
DEM6 
His -114.8 - 
# 2 
Auto tie 177.9 0.0 8.1 15.2 
& 6 GCPs 
DEM7 
No his -118.7 
0.5 32 
Auto & 165.8 0.5 7.1 3 
Manual 
tie 
DEM8 
His -119.7 
= 8 
Auto & 170.5 0.4 7.6 14. 
Manual 
tie 
  
Table 2. The statistical results of comparison between extracted 
DEMs and reference DEM 
Obviously, in this table, DEMI, .., DEMS, are in regards to 
corresponding area of the reference DEM. 
In Figure 3, a part of the entire DEMI is given and the area of 
the reference DEM is indicated. In Figure 4a the reference 
  
  
Inte 
  
pro 
for 
the 
whi 
119