| of various 
erial 
| and of 
rial 
lable. 
al procudures 
in Hamburg, 
| refers to 
[11/1 
thematical 
, respectively, 
e rather 
hey are: 
)f photo- 
he total 
y control of 
th other 
Mathematical 
ystematic 
‘kshops on the 
| here. 
'OSS errors 
| Observed 
le thank here 
tion. 
2. Systematic image errors 
2.1. After an initial analysis it was decided that the WG would not go into new experimental 
tests on systematic image errors and their elimination by self calibration block adjustment 
methods. The reason was that new experiments could not be expected to lead much beyond the state 
of information reached by the work of the previous Working Group of E. Kilpelä |3|. The previous 
experiments have given good insight into the effectiveness of self calibration and the precision 
level aerial triangulation is capable of. | 
Also, after more than 10 years of development of additional parameters the method has reached an 
operational level and is being applied in practice. Thus practical experience is constantly 
gained, and the method does not need special stimulation any more. 
It can be stated that the method of self calibration has seen thorough development within a 
period of about only 10 years, leading to operational computer programs and practical application. 
Also the scientific investigations have clarified both capabilities and limitations of the 
method to the extent that application is relatively safe. 
2.2. It is astonishing, nevertheless, that research into selfcalibration seems to have lost 
interest. Photogrammetrists seem to be satisfied that the additional parameters take sufficiently 
care of unknown systematic image errors. But to what extent the method is effective at all or how 
far the success depends on overlap, control, density of points or on the particular set of para- 
meters has not really been worked out. Here, a vast field of investigation remains to be dealt 
with. We believe, however, that this kind of investigations is at present to be left to 
individual researchers rather than to be taken up by a working group. 
2.3. In spite of remaining questions it can generally be stated that by the application of self 
calibration methods in block adjustment a consistent precision level of gg < 3 um for image 
coordinates has been reached provided instrumental errors and point transfer errors are 
negligible (by using comparators or analytical plotters and signalized terrain points). 
Thus, an extremely high performance level has been reached which proves aerial triangulation to be 
a genuine high precision method for point determination. It should particularly be noted that such 
precision is being reached more and more in standard practical application and is not confined to 
special research results only. 
It must also be noted, however, that the main contribution for reaching such high precision comes 
from reducing intstrumental errors (by comparator-type instruments) and by avoiding point transfer 
In such cases and in connection with sufficient number of 
errors (by using signalized points). 
0 - 30% 
control points the further refinement by self calibration is relatively small (about 1 
only). Bundle adjustments without self calibration and even the independent model method can 
reach almost the same precision level of 3 um or better. This indicates the high geometrical 
quality of modern aerial photographs. An important conclusion is, therefore, that development of 
high precision point transfer methods, for instance with digital image correlation, is urgently 
required, as signalized tie-points to which high precision results are limited, so far, remain 
restricted to special cases of application of block adjustment.