In table 6 the line widths and their standard devi- 
ations are presented for the different systems and 
different zoom settings. 
Table 6 and figure 6 show that when zooming in 2. 
with system 1, the line width decreases from 44 to 
20 microns. The reason is that superimposition 
takes place nearby the occulars. 
For the systems 2 and 3 zooming takes place near- 
by the photo and therefore the line width is con- 
stant, respectively 25 and 76y. It was not possible 
for all systems to use the same illumination factor. 3. 
Therefore it is possible that some lines have been 
measured relative thicker. The measuring results 
show that there is no relation between the line 
width and the line direction. To show this depend- 
ence the standard deviation of the measurement 
should be smaller. 
  
  
  
  
  
  
  
  
4. 
System Zoom 
1 10x 
20x 
40x 
10x 76 
20x 74 
Table 6. Line width at photo scale for different 5. 
systems and zoom settdings. n = number of lines; 
d = line width in microns; 0 = standard deviation 
of line width measurement (microns). 
width 60 
in u 
50 - 
A 
| 404 
| 304 = ET. 20 x 6 
| P s Es «t 
201 "^ t es N ALL 
= Dé 40 x 
10-1 
0 0 125 25 37.5 50 62.5 75 875 100 
—— ——»- line direction (grads) 
Figure 6. The line width of system 1 for different 1. 
zoom settings. 
2. 
4. CONCLUSIONS AND RECOMMENDATIONS 
1. The superimposition accuracy at the position of 3. 
the measuring mark through the whole stereo- 
model for the three tested systems varies from 
9 to 28 micron at photo scale (table 3). 
568 
If corrections are applied for systematic devi- 
ations then the accuracy varies from 5 to 22 
micron. 
Superimposition through the whole field of 
vieuw shows large systematic deviations (fig. 
4). If these systematic deviations can be 
removed by system calibration then the accu- 
racy for the three systems varies for magnifica- 
tion factor 10 from 10 to 24 micron at photo 
scale (table 5). 
The width of the superimposed lines for dif- 
ferent zoom settings is presented in table 6. 
For the systems 2 and 3 the line width is con- 
stant and is respectively 24 and 75 micron at 
photo scale. For system 1 the line width 
decreases from 44 to 20 microns when zoom- 
ing in from 10 to 40 times. 
The results of the research show that error 
detection of feature coordinates with 
superimposition is possible in a limited way. 
If the superimposition accuracy varies from 20 
to 40 micron in the photo than this accuracy is 
two to three times worse than the 
photogrammetric measuring accuracy of points 
which describe hard topography. One must 
keep in mind too that when using mono- 
superimposition the detection of errors is diffi- 
cult if errors are present in the Z-coordinates of 
the data base. These Z-coordinate errors are 
presented in the photo as x- and y shifts. 
In order to detect Z-coordinate errors stereo- 
superimposition is necessary. Through the high 
requirements of resolution, frequency and 
image synchronisation however,  stereo- 
superimposition requires much more processor- 
and memory capacity, which is also reflected in 
the price. 
A supplementary solution for the detection of 
errors which are present in the database could 
be to use the 'move to' routine of the analyti- 
cal plotter. When the operator does not trust a 
certain object (because f.e the superposition is 
not optimal) then he can check these points 
with this routine. The 'move to' accuracy lies 
between 3 to 10 microns at photoscale. When 
using the ‘move to’ routine with only X, Y data 
base coordinates the operator himself has then 
to set the measuring mark on the terrain. 
It is recommended to calibrate the superimposi- 
tion systems if the purpose is to use them for 
the detection of large database errors. This has 
been shown by the results as presented in 
paragraph 3.3. 
5. LITERATURE 
Persie, M. van, Superimpositie. Beschrijving, 
precisie toepassingen. Afstudeerscriptie, Fa- 
culteit der Geodesie, TU Delft, februari 1991. 
Witmer, R.C.J., Beeldinspiegeling: een hulp- 
middel in de fotogrammetrie. NGT Geodesia 
1991-7/8. 
Bonjour, J.D., Eital, G. Kóbl, O., 1988, Experi- 
ences with the Wild System-AP and its Stereo 
Injection Verification System, Proceedings 
ISPRS, Commission V, Kyoto.