2.2.3 Calibration: As control points a sufficient number of well 
recognisable object points were measured with the totalstation 
Leica TCRA 1101. The experience shows in this case, that the 
accuracy of the determination of the coordinates is about 
<= +/- 1 cm. 
The self calibrating triangulation could be carried out with 
Erdas Orthobase. After some few point measurements (the 
control points were measured in each image) some 20 
additionally tie points could be generated automatically. 
With the additional information of some initial values for the 
exterior orientation the triangulation could be calculated. The 
calibration of the camera was calculated with the parameters et 
of D.C. Brown. The further evaluation showed that the 
parameter for the radial-symmetrical distortion conveyed 
already sufficiently precise results. 
  
  
  
Property Value S.D. 
Focal length 20.373 mm +0.0261 mm 
Principal point x0 0.073 mm +0.0167 mm 
  
  
  
  
  
  
Principal point yO 0.086 mm +0.0220 mm 
  
Figure 5:. Results of calibration with radial distortion graph 
With this equipment, the camera positions practically possible 
and with the choice of control points (natural object points) it is 
to be accepted that a further accuracy increase can only be 
achieved through additional, considerable expenditure. 
3. IMAGE MAPS 
3.1 Basics 
The demand of the image maps was dependent of the different 
types of buildings. There were differences between the islamic 
installations and the ancient structures. 
For the islamic installations the image scale was fixed at 1: 50. 
Important was a safe and quick documentation before the 
dismantling during the excavation. The view of the objects was 
taken by digital photos with less laps. The geometric 
information was given by geodetic measurement. 
For the ancient structure the picture scale was fixed at 1: 10 or 
1:20 and a high degree of geometric accuracy was required. To 
structure the documentation work the temple was divided in 
sections. After the complete excavation of a section the 
photogrammetric documentation was started. Stereo-photos for 
3D processing were taken and evaluated. The geometric 
information was given by geodetically measured points and by 
photogrammetrically determined tiepoints To visualise the 
surface structure in small-scale, detail-photos were projected 
onto the large-scale image maps. 
3.2 Software 
For rectification different software-products were available. 
Erdas Imagine 
Erdas Imagine Orthobase Pro 
Eddi 2D (Fokus GmbH Leipzig) 
For further procedures of image enhancement the Software 
Adobe Photoshop 6.0 was used. 
3.3 Rectification with Eddi 2D 
With the simply operable software Eddi 2D the images were 
rectified by a projective transformation onto a given plane. This 
projection plane can be determined through the choice of two 
typical control points which were measured on the object 
surface. 
The use of this software was sufficient for the rectification of 
single images that do not have to be stitched together to an 
image plan. Overview displays and images in a specific scale as 
a basis for drawings could be produced easily. 
After a short training period the architects could carry out the 
equalisations autonomously. 
3.4 Orthorectification with Orthobase 
Orthobase from Erdas is a photogrammetric triangulation tool 
for the orientation of any type of image. Depending on the input 
data provided, photogrammetric techniques such as space 
resection, space forward resection, and a bundle block 
adjustment with self calibration are used to define the variables 
required to perform orthorectification, automated digital surface 
model extraction, stereopair creation, highly accurate point 
determination, and control point extension. 
3,5 Practical evaluations on the example of a typical facade 
The stereophotos had a photographic scale of 1:450 and were 
taken with the Fuji Fine Pix with the image resolution of 
2304*1536 Pixel. This resulted in a object coverage of 4.5 mm 
for each pixel in the imagery. The distance was about 9 m, the 
focal length 20 mm. Six control points were measured with an 
accuracy of about 1-2cm, depending on the chosen natural“ 
object points. 
The GCP's were measured in the images manually, and the 
triangulation result gave a first estimate of the exterior 
orientation. With the capability of tie point collection — based 
on the initial exterior orientation, a needed amount of tie points 
were automatically identified and measured. 
  
Figure 6: Overview of automatic tiepoint collection 
The following table lists the result of the triangulation process 
on the basis of the standard deviation of unit weight. The 
standard error is calculated based on residuals computed for the 
estimated and adjusted observations. 
—376— 
  
For the 
product 
product 
elevatic 
the helr 
For the 
betweei 
points) 
deviatic 
error W 
shadow 
with Er 
With t 
througl 
  
pax NOR V 
Thea 
comp: 
Sirwa