The general task, however, of establishing a comprehensive stochastic model for photogrammetric 
point determination still remains to be solved. Related with this general task is the delimitation 
of the stochastic error properties against systematic image erros and gross observation errors. 
It was felt, however, that the problem formulation is not yet clear enough, and that it was too |2 
early for the WG to rush into experiments. Instead,the efforts were first to be directed towards 
a thorough theoretical foundation and preparation. Therefore, the WG decided to organize a 
seminar (together with OEEPE Commission A) on the basic questions of mathematically modelling 
the total system of geodetic or photogrammetric point determination. The seminar was held on 
26 and 27 November 1981 in Stuttgart. A number of outstanding experts in the fields of |/ 
statistics, geodesy and photogrammetry were invited to present their views on the problem. 
The proceedings are published in |1|. 
4.2. The 9 papers presented at the seminar concentrate around the concepts of robust etimation, |! 
stochastic models including criterion matrices and variance-covariance estimation, evaluation of 
functional models, gross error detection and sequential procedures. | 
It can be said in general that on the geodetic side no concepts are ready which would clearly and 
directly suggest a certain strategy for setting up experiments for the assessment of the | 
stochastic model of photogrammetry. On the.other hand, however,.the presentations and discussions 
cleared considerably the theoretical situation. And a number of concepts were emphasized or 
emerged. These are in particular the concepts of robust adjustment, criterion matrices, the 
separation of gross and systematic errors, and a concept for assessing the frequency of small 
gross errors. 
In the mean time studies were continued about how the stochastic properties of series of photo- 
graphs could be described and assessed without requiring unduely large experimental efforts. 
Here, particular attention is drawn to the approach by R. Schroth |7| who models series of 
photographs by an autoregressive stochastic process which is defined by relatively few 
statistical parameters. Their assessment will not require too much experimental effort. 
4.3. The WG III/1 did not attempt any more to set up experimenta] investigations about the 
stochastic properties of photographs and series of photographs. It is suggested, however, that 
the theoretical and methodical basis for treating photographs as a stochastical process has been 
sufficiently clarified and that an experimental program might be launched by this or another 
working group during the next congress period. 
It is not expected that a refinement of the stochastic model of photogrammetric point deter- 
mination will result in spectacular accuracy improvements. Nevertheless it is necessary to 
establish a complete and realistic mathematical model. It will serve for the evaluation of 
the accuracy capability of photogrammetry in general. It will also allow performence prediction 
for special application and promote the further development of computational methods. 
Acknowledgement 
In closing the activities of WG III/1 I want to express my appreciation and my gratitude to all 
members and colleagues for their valuable contributions, their active participation, and their 
friendly cooperation.