2. COVERAGE PLANNING AND STEREO- 
PREPARATION 
The photogrammetric survey consisted of views to be 
worked on separately (about 80 photographs) or by 
stereoscopic pairs (about 70). 
The problem of organizing such a patchwork found a 
rational solution with a rectangular network. 
Then we divided the field into parts which would be 
exploited in basic views, and we chose the way to 
connect each one to the others, that is to say, to find out 
the best compromise in the positions of the control 
points. 
According to the technical characteristics of the camera, 
a few geometrical patterns had been prepared in order to 
cope with any situation likely to occur. On a given part of 
the site, the most suitable pattern was then chosen and 
repeated in a sequence for as long as necessary . 
The geometrical patterns were marked out on the 
ground by spots staked out at regular intervals; when 
operational, they were equipped with targets. The 
network was set with regard to the ancient square town- 
planning (two major perpendicular axes : "cardo" and 
"decumanus") . 
Flat areas were recorded on single views (fig. 6) with 
little overlapping (10 to 20 96). 
  
  
9 ground nadir point 
: + side ground point 
A position of the 
support 
  
  
  
overlopping area 
Fig. 6 : pattern used for the central yard 
A stereoscopic coverage (60 % end lap and 20 % side 
lap) was used for ruins with important relief (up to a few 
meters). After signalizing and measuring the 3D 
positions of the targets, the survey was worked out into 
strips along the prominent structures and along 
excavations, in order to avoid hidden parts (fig. 7) . 
  
  
  
  
  
  
  
  
A A A A A 
+ + + + + 
+ 
4 
+ 
Y 
  
  
o ground nadir point 
+ side ground point 
A position of the support 
Fig. 7 : targets location and ground nadir point on the 
eastern tribunal 
International Archives of Photogrammetry and 
3. PHOTOGRAMMETRIC PLOTTING 
3.4 Bundle adjustment with the ORIENT software 
Before starting the plotting, a photogrammetric 
triangulation method was needed for balancing the set of 
control points over each strip . The method used is 
derived from aerial triangulation. We used the. ORIENT 
software from the Institute of Photogrammetry and 
Remote Sensing (Vienna University of Technology), 
[Kager, 1991]. 
The strips were recorded in the ZEISS Planicomp P33 
analytical stereoplotter with the ATM program. 
  
  
  
  
  
  
  
  
  
  
number of negatives (24mm x36mm) 24 
camera Ricoh KR10M 
number of points for the adjustment 184 
number of control points 35 
control points r.m.s.e. (X,y,2) «0.01m 
tie points : 
r.m.s.e. (Xy) «0.02m 
r.m.s.e. (z) «0.03m 
number of observations 995 
number of marked observations 14 
adjustement standard r.m.s.e. 8 um 
  
  
Tab. 2 : results of the bundle adjustment with ORIENT, 
(curia area , four strips) 
The parameters of the calibration were computed before 
and after the field work on a set of targets at ENSAIS, in 
order to check the stability of the camera. 
  
  
  
  
  
  
Calibration of the Ricoh KR 10 M camera 
Ro=14mm | c(mm) | X'o(mm) | y'o(mm) a3 a4 
pre- 
calibration | 28.128 -0.002 -0.165 | -0.401 | 0.065 
calibration 
after field | 28.117 -0.004 -0.136 | -0.366 | 0.049 
work 
  
  
  
  
  
  
Tab. 3: inner orientation and radial distortion parameters 
The equations used for the adjustment of the distortions 
in Orient are : 
  
where R? = x? +y? and dR = dx" + dy;* (2) 
The radial distortion equation is : 
dR = RRR -R) egi ROC -R) 8) 
The distortion graph is shown on fig.4 . 
202 
Remote Sensing. Vol. XXXI, Part B5. Vienna 1996 
3.2 Sterec 
The stereo 
slides, were 
software. 
Microstatiol 
equipped 
interpretatic 
backgrounc 
It has provi 
classical st