le 
to 
re 
hy. 
an 
ic 
SS 
th 
T 
‚tion 
eir 
oach 
bundle would give a more rigid solution. In this way more pairs could be formed, 
thus providing more information for accuracy assessment. Additionally, the bigger 
the base, the more favourable the angle of intersection of homologous rays. The 
pairs which were obtained did not all provide the necessary requirements for a 
strong solution of the relative orientation problem. The results of the space 
intersection are summarised in Table 5. 
Two pairs were set up on a Zeiss (Oberkochen) Planicomp C100 (University of 
Bonn) and were measured and adjusted. In addition the photography was also 
resected using the Planicomp measurements and software. In Table 4 the Planicomp 
results are presented and moreover the resection results of both procedures are 
compared. These results will be discussed in the following. 
L. Assessment of accuracy 
  
The two pairs which were measured on the analytical instrument produced, after 
the orientation, & good stereoscopic model. Plotting, however, was not carried 
out, due to lack of time. In Table 4 the results of the adjustments are summarised. 
For the relative orientation natural detail points were also observed in addition 
to the control points. The value of the mean residual parallax could be improved, 
if desired, by repetition of observations on certain points. It was considered 
adequate, however, and absolute orientation was carried out using the known ground 
control points. The residuals after the absolute orientation refer, of course, to 
the control points used. For the small size of the negatives and the large rota- 
tions, the results of the Planicomp C100 orientations may be considered very 
satisfactory. Unfortunately, access to an analytical instrument is not common, at 
least nowadays. Thus the practical significance of these results is diminished. 
The observations on the analytical plotter were, however, used for off-line 
resection using software developed for that purpose. The exterior orientation 
elements of photograph 8535 using Planicomp C100 observations are also included 
in Table 4. They are compared with the corresponding results of the resection 
program. The observed differences have been caused by the use of different number 
of control points on each occasion. The difference in the magnitude of the 
residuals at image scale is probably due to the difference in the observing accuracy. 
It is, however, suggested that the improvement in observing accuracy and in the 
final result does not justify the additional cost, when negatives of this size 
are involved. It should also be noted that the ratio between x and y residuals 
is the same in both procedures. 
The space resection results, using the Co-ordinatometer observations are 
summarised in Table 3. The residuals at image scale are within the observing 
accuracy of the co-ordinates (0.030 mm). They are fairly consistent, varying 
according to the number of points used and the quality of the photography. No 
difference in accuracy has been observed between black and white and colour nega- 
tives. The calculation of ground residuals, which is arbitrary, only indicates 
the accuracy capabilities of the negatives. 
The exterior orientation elements of the eight resected photographs were 
used for the space intersections (Table 5). The coverage of the photographs 
favoured the formation of numerous pairs. The ground residuals are clearly 
affected by the base:distance ratio. Due to the differences in scale, the percent- 
age overlap is not in direct relationship to the base:distance ratio or to the 
number of common points. The observed difference in the residuals for the Z co- 
ordinate may be attributed to the normal angle photography. On certain occasions, 
however, this difference is minimised. A factor contributing to that is the 
favourable angle of intersection. Fimally, the internal accuracy of the procedure 
used for space intersection is proved to be well within the observing accuracy 
(0.030 mm). Hence, it may be again concluded that an improvement in the accuracy 
of the observing procedure would not produce significantly more accurate results, 
as the limitations are posed by the quality of the photography. 
211