Three classes of format can be noted consisting 
of products the size and number of which are as follows 
- class 1, small format : products at an output scale of 1:25 K 
of size either 22 x 26 cm (full frame) or 12 x 26 cm (stereopair 
amounting to nineteen products 
- class. 2, medium format : products at an output scale of 
1:5 K, coming from photographs at 1:16 K-scale, of size about 
30-35 cm by about 50-65 cm, amounting to sixteen products 
- class 3, large format : products at an output scale of 1:5 K, 
issuing from photographs at 1:30 K-scale, of size about 
55-65 cm by 75-90 cm, amounting to twenty four products. 
  
The actual number 
of pointings which have been carried out is as follows 
IGN (F) : 53120 
IPTU (A) : 1752 
NRC (CDN) : 276 
UGL (GB)  : 509 
Total number of * 5657 
pointings 
This total number includes pointings carried out for determi- 
ning the pointing precision, and is thereby more that twice as 
much as the number of averaged measurements (2243), used as 
inputs in the further computations for assessing planimetric 
and altimetric accuracies 
PLANIMETRIC ASSESSMENT 
PLANE TRANSFORMATIONS ; 
FOUR ASSESSING METHODS As described hereafter, different plane trans- 
formations have been computed and applied to those measurements, 
achieving a conversion of the orthophotos co-ordinates and 
enabling a comparison with those from the general catalog, 
established in the IGN (F) Lambert grid system. 
Once these transformations performed, some residuals appeared 
excessive, requiring a cancelling procedure for eliminating such 
measurements to be considered as gross errors. It was decided 
to eliminate those measurements exceeding 2.58 times the value 
of the r.m.s.e. in any residual, which corresponds to the maxi- 
mum error with a 1 4 confidence level to be reached. Seven i- 
terative cycles were necessary for stabilizing the procedure, 
no extra measurement having to be rejected after this last 
computation. The number of cancels was limited to 230,namely 10 %. 
For that purpose, four methods have been set up, comparatively 
used and applied to most of the orthophotographs. They consis- 
ted of three similarities and one affinity, the effects of 
which were therefore those of translations, rotations and 
scaling either similar along x and y or different. Hence 
they permitted it to evaluate the accuracy of the different 
products by computing residuals, standard deviations and 
r.m.s.e, on GCPs and check points as well, first for each 
round then for each orthophoto as an average of the related 
rounds, weighted according to the number of points in the 
round. > 
Afirst similarity has been 
computed by using only two remote known points. These base 
points were chosen by a computer program as being those dis- 
playing the largest distance between them in the round, in a 
similar way as a user should have operated. 
It was stated once more that extrapolation outside a base 
acting as a control point line, should not take place, which 
every user should be aware of ,when scaling his orthophoto. 
  
602