© 3D 
vertical 
horizontal 
Fige3 Distribution of photogrammetric controls 
required as input for the IMT, the measurement 
of six grid points/stereomodel was carried out 
in two projective planes by intersection method. 
During the IMT, a simultane spatial transforma- 
tion was executed for every models of the strip 
where the computational units were the formerly 
relative-oriented models with their projective 
center's data. The observations to be adjusted 
were the model coordinates, The model and plot- 
ting scale were set up 1/2000 and 1/1500, res- 
pectively. All computations were performed on 
the IBM 370/158 computer of the THD, The total 
CPU time was 38s for execution of the projec- 
tive center computation and for the IMT run in 
OS-HASP environment at 320 Kbytes virtual me- 
mory size, The adjusted control, bridge points 
as far as the baseline points and sections in 
20 metres interval were drawn by electronic 
flatbed plotter at scale 1/1500 serving orien- 
tation control for on-line plotting operated 
by the Wild A-lo autographe 
Data acquisition 
Setting up every single model, after the rela- 
tive and absolute orientation the following se- 
lective and adaptive sampling techniques were 
chosen for the surface representation: 
a/ Features as railroads, roads, bridges and 
boundaries of the band of interest, Number of 
registred points = 326, Digitized output data - 
Lo le 
b/ Grid points. Inside the band of interest a 
semi grid system was established. The inter- 
val in y direction was chosen constant lo met- 
res while in x direction the operator has made 
registration in time-interval carefully with 
the aspiration that a smoothing curve on the 
registred points on the real surface profile 
in the same vertical plane be within an accept- 
able interval of 1/3 metre, Number of regist- 
red points = 1736, Digitized output data = X, 
Y,2 /fige 4/o 
42 
    
  
   
grid model 
1 
step direction 
Fige4 Grid model 
c/ ICES profile points in sections, followed 
the preliminary drawn lines perpendicular to 
the predefined baseline with profiloscope. 
Interval was set up to 20 metres while 
density within the individual sections was 
variable according to surface characteristics. 
Handled as randomly distributed reference points 
grid pattern generator program can process this 
kind of data if desired, computing heights in 
given intervals with one of the method reviewed 
by Schut [5]. For this purpose a modified IBM 
software is at our disposal which uses second 
degree equation, in which the rightmost terms 
are rejected according to the number and location 
of the surrounding height references, The ave- 
rage number of points/section «17, is conform 
with the recommendation of the Swedish Road 
Administration for normal and partially rough 
type of terrain [1,4]. Number of scanned points 
= 1119, Digitized output data = station no., X, 
Yon / fige: 5/5 
All points were registred by Wild EK-8 unit 
interfaced with IBM 545 puncher on card media, 
The disadvantage of this configuration is the 
relative low registration speed and, at higher 
recording density & large volume of output cards 
would result; for mass recording only mag tape 
is practical. For further representation of 
the terrain based on string format storage 
concept, reference points located on contours 
as characteristic terrain features were regist- 
red. Vertical interval of contours at 1/2000 
model scale was 2 metres, in some cases with 
additional intermediate information. The num- 
ber of contour points registred in time-inter- 
val = 1260, The standard error both in plani- 
metric and elevation sense was less than ,1 
in 1/1000 of the flying height,