Jacobsen: 
Elassal: 
» 
Jacobsen: 
Elassal: 
Jacobsen: 
AUTHOR: 
TITLE: 
DISCUSSION: 
  
I have talked about high precision point determi- 
nation that means we have had well defined points. 
In general, I think this thing is a little differ- 
ent if you have only topographic features or if you 
have only well defined points. In our case, we have 
had well defined points and we have had no problems 
with the strips and in general, the number of blun- 
ders has been small. That is not a problem, because 
if you have a low reliability in a first double 
block, we have the check by the following measure- 
ment then. That means in the case of the final 
block we have very good reliability, much more than 
usual and so we have absolutely no problem in re- 
liability. 
I have another remark myself on the problem of 
transferring points between strips. We found in our 
organisation that operating on 9x18 Stages where you 
can mount four photographs simultaneously provides 
a very good facility to deal with this situation. 
Of course, this is a solution but we still do not 
have those stages. That is the problem. 
I know. I'm just proposing it can be a solution. 
I think this can be a solution. There is only the 
problem that, here in this case, also the photos 
should be regular more or less in the whole block. 
Usually we have not one photo exactly above the 
photo of the neighbour Strip... This will cause us 
a little problem, but I think, in general, this 
will be a solution. 
K.R. Holm (Norway) 
IEST OF ALGORITHMS FOR SEQUENTIAL ADJUSTMENT 
Grün (Switzerland): I think that these two last arguments about the 
Holm: 
saving of primary storage are not really important, 
because the number of elements that you have to save 
in such a little block is so little. And the other 
argument with respect. to parallel processing also 
holds for the TFU, of course. Also the TFU lends it- 
self favourably to pipeline or parallel processing. 
So those arguments are valid to both algorithms. 
Thank you for the comment. For this last point I 
have just looked at the pattern and seen that the 
elements are treated in a very sequential way in 
Givens transformations. I have not thoroughly looked 
at the TFU algorithm in this respect. 
58