ee a 
Ee 
= == 
  
30 
/ ’ " " s 
"(dais — di; 15 — dv; nus d- de; 41) (93) 
t 
, "n "n "n / 
db, ; 51 =n (dx; ; 5 dæ;i+rs + dy; 1577 da, ; r13 + dx; ; , 13 ^77 
9 ,9 ‚5 
- 
"H 
"n ‘ / , 
a ar | 
m dy; iiis d d; |i4 = dX, iia zi dy; iia f= dx; iis "D 
dr da; is = dy; ie) (94) 
For the further treatment of the error transformation attention has 
to be paid to the performance of the measurements of the image coordi- 
nates. In using stereoscopic measurements of the image coordinates the 
pairs à — 1,5 and ¢,0 + 1 are measured independently. The image 
coordinates of the photograph i will in other words become measured 
twice independently. If the image coordinates are measured separately 
for each image in single image comparators the coordinates of each 
image will be measured only once. Hence the measuring method will 
play a certain role for the correlation within the strip. 
Below we assume the measurements to have been performed stereo- 
scopically and consequently independently for each pair of photographs. 
The errors of the basic image coordinate measurements «'v', u"v" 
propagate to the oriented image coordinates z'y', z"y" according to the 
expressions (30)— (33). 
In order to simplify the notations we will from now on use capital 
U and V for the image coordinates of the image pair 2,7 + 1 and small 
u and v for the coordinates of the image pair 4 — 1,7. U', for instance, 
means an image coordinate which is measured in the image i of the pair 
iA + 1. 
After introducing these notations we find from the expressions (30) 
—(33) as follows. 
Image pair © — 1,1 
Image à — 1 
, , , , 1! , 
dx, — du, — du, — dv, + dv, 
, , , 
dy, — dv, — dv, 
, , , , , 
A — d ju 3 — 3 
dx, = dug — du, + dv, — dv, 
, , , 
dys = dv, — dv,