for 
here 
es. 
tion 
ight 
ther 
and 
tric 
onic 
HAUS 
mage 
the 
| the 
can 
tric 
ig to 
;HAUS 
the 
well 
)85/, 
nates 
from 
are 
ed in 
uring 
s the 
, the 
ights 
| the 
rate 
   
iteratively. The standard deviation of the 
weight unit, the average unknown corrections, or 
the average residuals of the observations can be 
employed as characteristics. If the system does 
not converge, or if there are gross errors, the 
characteristics minima will not be reached, so 
that the set number of iterations is exceeded. 
This will stop the computation. 
In addition to those characteristics already 
mentioned, a correlation coefficient which 
serves as similarity rate is used in multi image 
matching. This coefficient defines the 
similarity between the gray dispersion of the 
object surface and the corresponding gray scale 
dispersion of the image; it is computed for each 
image according to equation (4.1) /SACHS 1973/. 
[01,5 * Gi} - [9g1.1] * [6i] /n 
  
ri = 
(C[gi, 53?1- [g1, 31? /n) *([G5 2]- [G5 ] 2/n)) ? 
(4.1) 
with: rij - correlation coefficient 
gi,j = image gray value 
Gj = object gray value 
n = number of surface elements 
i = image index 
j = surface element index 
Réseau crosses, total reflexion or lacking 
object structure may cause signal disturbances 
in the images. In this case, the correlation 
coefficient is considerably lower than the 
average of all coefficients. 
In the method that is being presented in this 
paper, images with signal disturbances will be 
excluded from further computations by means of 
the correlation coefficient. 
5. APPLICATIONS 
In the following, two examples will be presented 
to illustrate possible applications of digital 
multi image orthophotos and the method of its 
computation. 
5.1 Example 1 : Testfield 
A testfield with an affixed texture serves as a 
first example (see Fig. 5-1). In this case CCD- 
sensors were used to create digital images. 
  
   
   
  
  
  
  
  
   
   
   
   
ZT Sat xh € 
pis O^ x 5x 
goa RA^ TT 
Figure 5-1 : Digital images of the testfield 
The parameters of the interior and exterior 
orientation were defined during preliminary 
measurings. For the computation of the multi 
image matching, the size of the single surface 
elements was set at 1 x 1 mm2. The number of 
elements per mesh was 15 x 15. 
Figure 5-2 shows the operating display of the 
multi image matching. 
  
Figure 5-2 : Operating display 
The meshes of the single images and the result 
of one iteration (see SAMPLE) is shown in figure 
5-3. 
  
Figure 5-3 : Control display