Ww MS M p, 
— 
a 
cal 
ti- 
19 
| _ 
S 
n 
Research has consequently moved to the empirical analysis of 
systematic image deformation or model deformation and to the 
methods of correcting such errors by selfcalibration. Vast 
experimental data have been obtained which prove signi- 
ficantly the existence of systematic image errors of noti- 
ceable magnitude in virtually all cases. Systematic errors 
have been found to be remarkably consistent, and their depen- 
dence on project parameters has been investigated, such as on 
camera, flight mission, flight direction, and measuring 
instrument. Those results imply that the correction of syste- 
matic image- or model-deformation by a suitable method improve 
considerably the accuracy of aerial triangulation, depending 
mainly on control and overlap. 
Related with the analysis of image errors (although developed 
parallel or prior to) appear the methodical developments and 
computer programs for the correction of systematic errors by 
suitable procedures, especially by selfcalibration. The test- 
field-calibration of image coordinates has proven to be re- 
markably effective. Main emphasis is, however, on the method 
of selfcalibration of photographs or stereo-pairs through 
blockadjustment. The principles have already been presented 
at the Ottawa Congress in 1972, together with reports on first 
experience and results. The subject has been thoroughly dis- 
cussed at the 1974 symposium of Commission III. The develop- 
ment of the method of selfcalibration and of associated com- 
puter programs is still going on, combined with efforts to 
solve the inherent numerical and statistical problems. There 
is no doubt that aerial triangulation with selfcalibration 
can achieve considerably increased accuracy and better relia- 
bility of its results. This development marks a new highlight 
and qualifies aerial triangulation as a geodetic precision 
method for point determination. 
2.4 Theoretical and numerical questions of computational 
techniques 
With the exception of the topic of selfcalibration there 
have been few isolated activities in the field of computatio- 
nal techniques proper for photogrammetric purposes. General 
development effects cannot be reported. 
The numerical solution of large photogrammetric adjustments 
seems to be stabilized on the level of presently applied 
techniques. There are indications, however, that block ad- 
justments can be considerably sped up once more. 
The problem of automatic detection and elimination of gross 
data errors has finally received some more attention, with 
first theoretical and methodical investigations, and program 
developments. The preliminary results justify some concern 
that complete gross error detection will not be possible with 
the conventional conditions of overlap, control, and redun- 
dancy. Solution of this problem has direct relation with the 
quality and economy of aerial triangulation. Therefore 
continued research is of primary importance.