nt 
RESULTS OF THE GEOMETRIC TESTS OF THE OPTICALLY PROCESSED IMAGES 
Table (2) shows the results obtained after testing the survey optically 
processed image of the test area. 
For the control points, the planimetric accuracy ranged from 8 - *27m 
when the polynomial contains highest order terms to ó om 283Pm when only 
the similarity transformation is used. Witn the corresponding set of check 
points, the best planimetric accuracy is again obtained when the highest 
order terms are contained in the polynomial. Comparison of the accuracy 
obtained after using the linear conformal transformation with that obtained 
after using the affine transformation shows that a 3x improvement in 
accuracy has been obtained. This clearly indicates that this survey- 
optically processed image of the test area contains large affine scale 
errors. 
Table (3) presents the accuracy results obtained with the precision op- 
tically processed image of the test area. In general, the accuracy figures 
ranged from § =% 37 to p= 517 m for the control points and 8p = I 59 m 
to 6p = i 352h for the check points. In both cases, the highest accuracy 
was again obtained when the polynomial contains highest order terms; 
while the simple similarity transformation gave the lowest accuracy 
figures. 
Inspection of the results on this Table reveal the fact that substantial 
scale errors also exist on this image supplied by ERIM despite the fact 
that optical processing of this image has been carried out to higher 
standards. This is evident from comparison of the accuracy figures ob- 
tained after using the similarity and the affine transformations. 
RESULTS OBTAINED WITH THE RAE IMAGERY 
The accuracy results obtained after geometric transformation of the RAE 
digitally processed image of the test area shown in Table (4). 
For the set of control points, the best accuracy figures were obtained 
when the polynomial contained the full 11 terms i:e óy- $595 m. Sy = T 34m 
giving rise to a planimetric error of * 80m. The accuracy remained nearly 
the same until the sixth term of the polynomial was truncated. A 1.5X 
deterioration in accuracy was then experienced. Quite unexpectedly, the 
largest r- m. s.e value was obtained when the polynomial was restricted 
to the first three linear terms (i.e. to the affine transformation) where 
values of &, =f 112 m and §y=2 84 m were obtained in the X and Y directions 
respectively with a planimetric error of 6 -t 140m. A noticeable point is 
that the affine transformation results are not much better than those ob- 
tained with the simple linear conformal transformation. 
The corresponding set of check points behaved slightly differently. The 
accuracy figures remained nearly the same for almost all non-linear terms 
of the polynomial. When only the first three linear terms were used, the 
accuracy deteriorated considerably from Sp="T 162 m when 4 terms were used 
to 6,.2* 205 m when only the affine terms are utilized. Surprisingly, 
the simple linear conformal transformation gave accuracy figures much 
“1;