al si- 
p. 
is 77 
lereo 
jrads 
The new terrain coordinates have been determined 
by transformation. After some computations we 
had to our disposal: 
1. X, Y, Z-coordinates of the theoretical terrain. 
2. Orientationparameters of the photos, which are 
in this case two grid plates. 
3. Photo-coordinates of the terrain points. 
3.2. Test A. Superimposition accuracy at the 
position of the measuring mark through the 
whole stereomodel 
The aim of test A is to determine by measurement 
the superimposition accuracy at the position of the 
measuring mark through the whole stereomodel. 
First of all the intention was to use the following 
procedure: 
1. Realize the superimposition of the 77 data base 
points by using the known photo orientation 
data. 
2. Determine by measurement the  photo- 
coordinates of the superimposed data base 
points. 
3. Determine the superimposition accuracy by 
comparing the photo coordinates of step 2 
with the ideal photo coordinates. 
It appeared however that when using the 
superimposition software of the different available 
Systems recording of photo-coordinates instead of 
X, Y, Z coordinates is not possible and that for 
some systems the input of known orientation data 
give problems. Therefore the following procedure 
has been used: 
1. Input of database coordinates X, Y, Z. 
2. Interior, relative and absolute orientation of the 
grid plates. The coordinates of 6 data base 
points are used for the absolute orientation. 
3. Superimposition of the database on the left 
photo. 
4. Measurement of the superimposed points and 
recording of the terrain coordinates X', Y', Z'. 
5. Transformation of X, Y, Z and X', Y', Z' to 
photo-coordinates x, y and x‘, y with absolute 
orientation parameters. 
6. Determination of the superimposition accuracy 
by comparing the ideal photo-coordinates x, y 
with the photo-coordinates x’, y". 
When testing the three superimposition systems the 
database has been superimposed in the left ‘photo’ 
with a magnification factor of 10x. 
In order to facilitate the work of the 
photogrammetric operator the superimposed image 
has been shifted + 0.2 mm with respect to the 
photo image. 
The measurements have been executed forwards 
and backwards. When using system 2 only 66 of 
the 77 database points could be measured. 
After the transformation of X', Y', Z', to the 
geometry of the photo the obtained photo- 
coordinates have been averaged. 
Then for each system the standard deviations 0, 
and o, of the mean photo coordinates have been 
computed, see table 2. 
  
System 1 System 2 | System 3 
  
g. 4 7 6 
x 
g. 4 5 6 
y 
  
  
  
  
  
  
  
Table 2. Measuring accuracy of superimposed data 
base points. Magnification factor 10. Units in 
microns at photo scale. 
The superimposition accuracy o, and g, have been 
computed by using the formula: 
d 
y ido c (1) 
with d — (x - x!) for g, and d = (y - y!) for o,. 
n — the number of points. 
Computed standard deviations are given in table 3. 
  
System 1 System 2 | System 3 
  
  
  
0, 16 10 19 
0, 21 9 28 
n 77 66 77 
  
  
  
  
  
  
Table 3. Superimposition accuracy in microns at 
photo scale at the position of the measuring mark. 
n = number of points 
After plotting the discrepancies dx, dy for the three 
Systems it appeared that for the systems 1 and 3 
the superimposed image has a systematic distor- 
tion. For system 1 this is caused by a scale differ- 
ence with as origin the model centre and for system 
3 there is a distortion which is perpendicular to the 
model base and especially present at the left and 
right model border. 
After repetition of the computation with only the 
points which lie between the principle points H1 
and H2 (see Figure 2) the superimposition accurracy 
becomes as given in table 4. 
  
System 1 System 2 | System 3 
  
  
  
o, 12 10 13 
c, 21 9 22 
n 55 55 55 
  
  
  
  
  
  
Table 4. Superimposition accuracy in microns at 
photo scale with 55 points.