International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
  
The images are evaluated with the digital photogrammetric 
system PHIDIAS. At the beginning of any test a portable field 
with control points is photographed from various directions to 
perform the orientation of the camera assembly (figure 5). The 
field of control points contains several points, whose distances 
are previously determined using an interferometer. These 
distances guarantee superior accuracy and are introduced to the 
calculating of the orientation as additional observations. The 
orientation parameters are determined by bundle adjustment in 
combination with self-calibration. 
The targets are measured automatically using image processing 
with PHIDIAS. The centres of the targets are determined with a 
standard deviation of about 0.4 microns. The three-dimensional 
coordinates of the targets are subsequently calculated by 
forward intersection using the image coordinates of two or three 
cameras. The scale of the images is 1 : 7 to 1 : 10, that means 
the targets on the specimens may be determinable with an 
accuracy up to 3 microns (perpendicular to exposure direction). 
  
Figure 5: Assembly of measurement cameras 
3.1 Use of cameras 
Up to three cameras are available for the test series. For a rapid 
overview one camera perpendicular in front of the measured 
object is sensible. But a displacement of the specimen 
perpendicular to the measured plane cannot be recognised. 
Thus at least /wo cameras are used. À second camera allows to 
determine three-dimensional coordinates. Then in opposite to 
the one-camera-measurement deformations perpendicular to the 
surface can be detected. 
The assembly of /wo cameras has an important effect to the 
quality of result, as well as using three cameras instead of two. 
The difference between a measurement with the outer two 
cameras (assembly of all cameras see figure 5) to a 
measurement of one outer and the middle camera is between 4 
microns and 6-8 microns to the measurement of three cameras. 
In the border area far from the cameras the noise raises (figure 
6) 
  
  
a plate test 
4. MEASUREMENT OF DEFORMATIONS AND 
AUTOMATIC CRACK ANALYSIS 
I. The evaluation software PHIDIAS calculates from the images 
in every epoch three-dimensional coordinates of the targets and 
its standard deviations. The deformations can be determined by 
the difference of two measurement epochs. They show global 
displacements like the deflection of a beam (figure 7a). Only 
the difference of two adjacent displacement vectors focus 
attention on the cracks (figure 7b). 
  
       
  
  
         
AE 
I aT 
enti 
nem 
    
  
  
b 
  
Figure 7: a) Displacement of targets of a shear test 
b) Difference of neighboured vectors (vertical direction) 
c) Original image (cracks drawn) 
d) Schematic drawing 
Cracks are extracted from the displacements. For this the 
following method is developed and implemented. An overview 
shows figure 8. 
   
    
  
    
I Interface to PHIDIAS 
Measurement Input of point coordinates 
method - — [Measured value jt, 
Standard deviation © 
  
II Displacement 
x Crack probability 
—— ———— 
i 
IH Controlling 
Manual or 
Expert system 
  
  
  
  
  
Ye mm at 
^ e - . 
i >| IV Local intensify 
A Rising contrast 
     
1 
| 
| 
Analysis — 
I 
| 
| 
| 
  
| | XV Following algorithm another iterative 
E \ Snap rack : € dun 
\ Snap global crack structures optimization 
4 
  
  
  
VI Finite element analysis 
Fox cutie Output by ORFEUS / MicroStation | 
  
  
  
Figure 8: Way of crack analysis 
I. By the comparison of the measured change of length 
between two targets with the statistically distributed material 
parameter, which is necessary for the creation of a real crack 
(primary crack width), the existence of a crack can be 
calculated as a stochastic value (figure 9). 
     
      
     
    
    
   
   
    
    
    
    
   
   
    
    
    
     
    
   
    
    
    
EN 
PY A mt 03 ON OOS OSC em 009) wl) 
rent —e