Kohei Arai 
KEYWORDS: 
  
DEM Estimation with Simulated Annealing Based 
on Surface Reconstruction Method 
Dept.of Information Science, Saga University 
1 Honjo, Saga 840 Japan, arai@is. saga-u. ac. jp 
Algorithms, DEM/DTM, Theory, Cartography 
  
  
ABSTRACT : 
A method for Digital Elevation 
Mode | (DEM) estimation based on 
Surface Reconstruction with 
Simulated Annealing is proposed. 
The proposed method allows us to 
designate the region for minimizing 
the difference so that the estimated 
DEM is optimum in the sense of 
minimizing all the difference 
between actual radiance and 
estimated radiance based on the 
models for the region of interest. 
The existing surface reconstruction 
method focuses on the pixel of 
interest, not the region. When the 
pixel of interest, however, is 
suffered from occlusion, then the 
estimation accuracy of DEM get worth. 
Turns out, the proposed method takes 
into account the region, not the 
only one pixel of interest, 
minimizing the total difference in 
the region so that a good estimation 
accuracy is expected in such case. 
1. INTRODUCTION 
Surface Reconstruction method is one 
of well known methods for DEM 
estimation (Ref. 1). Surface 
Reconstruction method, however, does 
not work so well for the areas with 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996 
a variety of local minimal in the 
least square method which is 
featured in Surface Reconstruction 
method because of non-linearity of 
the equation. Further the previously 
proposed Surface Reconstruction 
method does not take into account a 
geometric relationship among the sun, 
target surface and two sensors which 
makes stereo pair of images results 
in degradation of DEM estimation 
accuracy. 
In Surface Reconstruction method, 
actual  radiance at sensor is 
compared to estimated radiance based 
on a geometric relationship among 
the surface of interest, sensor 
onboard satellite and the sun, and a 
surface roughness model. The 
estimated radiance is a function of 
not only the geometric relationship 
and surface model but also DEM. Then 
DEM is estimated to minimizing the 
difference between the actual 
radiance and the estimated radiance 
based on the models. In order to 
minimize the difference, Least 
Square Methods are used in general. 
It, however, has a non-linear nature 
in the process on the estimation of 
the model derived radiance so that 
singular DEMs sometime are estimated 
through Sur face Reconstruction 
method, in particular, for the case 
that there are many local minimal in 
the least square based estimation. 
On the other hand, non-linear least 
square method such as  Levenberg 
36 
    
  
    
   
  
  
    
    
  
     
   
   
   
  
   
  
   
    
    
        
  
  
  
   
  
    
   
    
  
  
  
  
  
  
   
   
   
    
  
   
   
    
also 
Two . 
propo 
overc 
situa 
The 
desig 
the « 
DEM 
minin 
betwe 
estin 
mode | 
The 
methc 
inter 
pixel 
suff« 
estin 
Turn: 
into 
only 
minit 
the 
accul 
2. Sl 
SIM 
2.01 
for 
Hein 
Heip 
Surf 
esti 
Mode 
imag 
rela 
inte 
pixe 
inte 
the 
diff 
inte