Summary of a paper for Commission III (Aerial triangulation) 
STEREOBLOCK ADJUSTMENT 
This paper outlines an approach to the problem of aerial 
triangulation suitable for a medium-sized electronic computer 
such as Pegasus, 
1. Size of store hinders simultaneous least squares adjustment 
of a large block of photographs; but individual stereogram 
adjustment, followed by a strip & block formation, does not fully 
utilise points comnon to several overlaps, and local distortions 
renain "frozen" in the final combined model, : 
The block therefore to be subdivided into "stereoblocks" of 
nine photographs (3x3) and each rigorously adjusted by least squares 
as a single unit, Adjusted stereoblocks can then be Joined together 
analytically (while retaining internal geometric structure), or 
would be immediately suitable for Jerie type adjustment, 
2 Stereoblock to be adjusted internally by variation of 
coordinates, Each observation equation has as nine variables the 
Cartesian space coordinates of two photo points and an air station, 
while its constant term is the discrepancy between the "observed" 
space angle in the photo system and that computed from provisional 
space coordinates, 
The nine coefficients are obtained by substitution in the 
formula: 
d s I |f$p-s Lt 
pexsep prp 7 8.8 T 
where r,8 are vectors (X-U,Y-V,Z-W) & (X'-U,Y'-V,Z'-W) from the 
air station (U,V,W) to the photo points (X,Y,Z), (X',Y',Z'). 
Constant term is: 
P. S 3 e T.08 
  
dr + 
  
  
2 
(20)? | Go metr ay rey)? - rx] 
(r.r)(s.8) | XX €yy €^ res 
where (x,y), (x',y') are photo coordinates and f is focal length, 
Hence, apart from two square roots, all operations are of 
simple multiplication, division etc. - very convenient for an 
electronic computer, 
  
  
3 lo preserve linearity of the expression for de, residuals in 
observation equations must be less than about 15' of arc. A method 
of arriving at provisional space coordinates sufficiently accurate 
for this purpose is discussed, It involves computing approximate 
coordinates and then improving these by a simple method of using 
the formulae for de (given earlier) and d^e in terms of dr, ds. 
4, For n photo points on a photograph, (2n-2) space angles for 
the (2n-3) observation equations must be so chosen that, if one of 
the residuals is large, then the observation may be rejected at this 
Stage with minimum disturbance to rest of scheme. A very convenient 
method using principal point as additional "spurious" photo point 
is indicated,