ae í i i 
a jis CR x ‘ ; ' — 
ES uu T 
The mean errors of & computed distortion A for an arbitrarily chosen d 
m, = m We, | (85) 
Substituting (81) into (82) we obtain 
. remains to be computed. 
A=d Ic My. K, (d 20080. x, (a : | 
which may be written as 
A=d [rr * fk + £5, (86) 
denoting 
Al A 
fi £, = a(d” - da) 
f=|f whereby £, = a(a* „ ) 
f f, a . 2) 
we may write (85) es 
| T 1/2 
! my =m [fQ.f] | (87) 
where Ok 1s the sub-matrix of the inverse of the original normal equation 
system associated with the unknown distortion parameters Ky» K, and Kye 
VII. CONSIDERATION OF REFRACTION 
Similar to distortion, refraction causes the tangent to the actual light 
ray, at the center of projection, to deviate from a line expressing the con- 
dition of co-linearity between object point, center of projection and image 
point, . 
In order to compensate for refraction, it is Necessary to incorporate a 
corresponding correction in the computation of the x? and y. values in formulas 
(12) anà (15), respectively. Figure 10 shows the basic situation, assuming 
54 
tha. 
pl ar 
fur!