International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
  
Figure 9: The recovered camera poses of the Rollei images used 
for the modeling of the rock cliff. The spheres represent the 
used control points. Top: a view from the top of the strip. 
Bottom: a view from the side of the empty niche of the Great 
Buddha. 
Because of some software problems with the Analytical Plotter 
we could not perform stereo measurements with the instrument. 
Therefore the images were scanned at 20 micron resolution and 
the modeling of the cliff surface was done by measuring the 
points with the commercial software Photomodeler. Then a 
mesh with a 2.5D Delaunay triangulation was generated and a 
textured model of the cliff was produced (Figure 10). 
  
  
Figure 10: Two different views onto the 3D textured model of 
the Bamiyan rock cliff. 
Because of the network configuration and the complex shape of 
the rock facade, the recovered geometric model is not complete, 
in particular in the upper part. In some areas it is not possible to 
find corresponding features, because of occlusions, different 
lighting conditions and shadows. This is not such a big 
problem, because the cliff model is not meant to be used alone, 
but in a next step it will be integrated into the DTM of the 
larger environment. 
4. MODELING OF THE TWO EMPTY NICHES 
The modeling of the two empty Buddha niches was performed 
using digital images acquired with a Sony Cybershot F707. The 
image size is 1920x2560 pixels while the pixel size is ca 34 
um. 
Both niches are now a “national monument with unique 
importance to humankind” and are safeguarded by UNESCO, 
the Japanese Government and the Afghan Ministry of 
Information and Culture (Figure 11). 
       
    
   
  
— Re Ea ORIS da D d Tu EAS) Nec d a: 3 & 
Figure 11: The sign in front of the empty cave of the Great 
Buddha that declares it a protected national monument. 
4.1 The empty niche of the Great Buddha 
For the 3D computer reconstruction 5 images were used (Figure 
12). The camera parameters were recovered with a self- 
calibrating bundle adjustment, measuring the tie points semi- 
automatically by means of Least Squares Matching (LSM) 
[Gruen, 1985]. The final average standard deviations of the 
object coordinates are 0, — 0.014 m, o, — 0.017 m, o, = 0.021 
m. 
Figure 12: The empty niche where the Big Buddha once stood, 
as seen in August 2003. 
Afterwards, distortion-free images were generated and imported 
into the VirtuoZo stereo digitize module [VirtuoZo NT, 1999]. 
Three stereo-models were set up and points were measured 
along horizontal profiles, while the main edges were measured 
as breaklines. Thus a point cloud of ca 12 000 points was 
generated. The recovered camera poses and the measured points 
c 
are displayed in Figure 13. 
    
  
  
      
  
  
  
  
  
  
    
    
   
   
   
  
   
     
   
   
  
  
  
  
  
    
   
  
   
   
    
  
  
  
  
  
  
    
  
  
Figu 
mea: 
Grea 
The 
metl 
Am 
surf: 
[Rai 
Figu 
  
Figi 
visu 
The 
com 
befc 
  
  
bet