(e 
us ( 
u (e 
|n- 
ion 
as 
o1 
ne 
to 
rror 
e COR 
Al ZH" 
BEER 
e i Tee ee E 
nw 
n 
2: = 
ond "ZR" (Fig. 3, and Fig. 4 a) and 5 a)) points at film distortion to 
some extent, which is however well cancelled by polynomial 2 in case 
n7 R 5 
Un 
b 
m 
s 
Residuals at check points (Fig. 2 and 3 c), 4), and Fig. 6 a), 
b)) display some local systematics. Differences of these residuals, 
(e.g. "ZH" minus "W" in Fig. 9 &) point at & well discernible systema- 
tic for medium radial distances. These results, with polynomial 2 
applied, lead to the above-said conclusion, that this type of cali- 
brations does not cover the whole deviation of the photogrammetric 
system's performance from central projection. À joint use of e.g. 
polynomials 1 and 2 is therefore recommended. 
Some concluding remarks should be made in this paragraph concer- 
ning significance and correlation of parameters of polynomials. Sig- 
nificance may be checked by t-test, and it changes with variation of 
cases (e.g. case without and with réseau) without reasons clearly 
detectable. With polynomial 1 significance may be only simulated, as 
the parameters in r are highly correlated. With polynomial 2 signifi- 
cance of d and d is sometimes given despite application of réseau. 
19 01 
Terms C5 » 8nd dys were significant (< 90 %) in equal 
or ess. Lena 3 the conpuhsS standard cases with this polynomial. 
Correlations larger than 0,7 occured only in the typical case of 
angles and coordinates of outer orientation and - as mentioned before - 
between parameters a. of polynomial 1. 
Fig. 10 is equivalent to the former ones, as far as polynomial 2 
is applied to standard case with réseau and lab. calibration to the 
flights W, ZH, and ZR. There are standard error ellipses of polyno- 
mial vectors at the chosen 25 points added, i.é. these ellipses show 
standard errors of partial field calibration. 
Fig. 10 b), d), f), showing these results plus error ellipses for 
polynomials 2 applied are matched with Fig. 10 a), c),e), after appli- 
cation of polynomial 1. Besides quite different shapes of calibration 
results due to different parameterization and constraints there are 
very large differences in shape and size of error ellipses to that ex- 
tent, that at the corner points of the photographs after applying po- 
lynomial 1, they degenerate practically to lines in tangential direc- 
tion. This is mainly caused by the comparatively large standard errors 
of the direction of symmetry axes and amount of asymmetric distor- 
tion for large radial distances. 
6.3 SIMULTANEOUS CALIBRATION OF SEVERAL PHOTOGRAMMETRIC SYSTEMS 
The results of calibrations resemble very close those of standard 
cases. There is an overall increase of o_, as all photogrammetric ob- 
servations are introduced with unit weight. Weighting according to e.g. 
9, of standard procedures would not alter results of calibration con- 
siderably. 
141