inbul 2004 
otography 
  
mage 
] case of 
measured 
yramic 
y and 
  
ments as 
es to be 
ts. 
  
ecognition 
Idwork of 
are marked 
dentified 
here shown 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
  
  
Figure 8. Natural control point No. 22 (Fig. 7) measured with 
tacheometer. Each pointing of natural control is documented 
simultaneously by photographing it. These documents are 
essential in order to identify and relocate control points exactly 
in survey images. 
  
Figure 9. Closer detail of marking of natural control point No. 
24 in Figure 7. 
  
Figure 10. Point No. 24 (Fig. 9) photographed for 
photogrammetric identification. The exact location is shown 
here by a measuring stake. It is essential that the viewing 
direction be close to the same than the one of survey images 
(Fig. 5). 
2.3 Orientation of images 
The photogrammetric processing of image orientations consists 
of measurement of control and tie points, and of block 
adjustment. This has been done with Pictran software (Pictran, 
2000). We use some 15 - 20 tie points for orienting one 
panoramic stereo pair. From these points less than half may be 
control points, which have been measured with tacheometer 
(Fig. 11). The minimum number of control points required for 
one stereo pair is four. The standard error of image observations 
after block adjustment has varied between 0.5 — | pixels. 
In case several panoramic sequences are taken from one survey 
area - but from different distances - we aim to adjust all images 
in one block. However, this may often cause singularity 
problems. Similarly, tie points further away cause problems, 
and in very front of both cameras the stereoscopic area splits to 
two different views. 
  
Figure 11. Geodetic control points (numbered squares) and tie 
points (circles) of the panoramic stereo model. The numbered 
points are the control points. 
2.4 3-D measuring 
Once the images are oriented, the remains of structures, which 
were identified and sketched on prints during archaeological 
survey, are measured in 3-D. If necessary, the digital elevation 
model (DEM) will be densified. 
The structures will be described and modelled as poly lines. The 
corresponding points are identified and measured in both 
images and the 3-D coordinates are determined by intersection 
in space. The 3-D measuring is performed point wise, since the 
software we used does not support stereoscopic measuring. 
However, the images can be at least viewed stereoscopically, 
which will ease remarkably the interpretation of the structure. 
Since a panoramic image sequence is rectified to a plane, 
geometrically transformed and resampled, the resolution is 
reduced (Fig. 5 and Fig 11). The stretching of the pixels will 
reduce the interpretability as well. Therefore, it would be better 
to use original images for all photogrammetric processing, or 
project them on cylinder instead of planes, but the software dos 
not support this either.