4. Covariance Analysis 
The covariance functions are quite different between bundle block adjust- 
ments with or without additional parameters. But there is no significant 
difference between adjustments with the whole set of additional parameters 
(24) or the set, reduced with the tests described above (see fig. 2 and 3). 
ExCEENSs à | 
5e 
E | 
     
  
  
0% A fig. 3 normed covariance / cross covariance x - component 
.*2 3 CXII . 
= | block Jaemijaervi, adjustment with 24 additional parameters 
3 
E 
32 
Je 
. 1 
2 e | —= — = normed cross covariance function of neighboured photos 
... 
„22 \ — — normed cross covariance function of photos with photo number 
9 
js t differences larger 19 
le & 4 
„12 
le \F 9 
‚98 3 ? 
de i 2 8 x $ 
84 8 A a 
EE crimes M DA MU EAE, 
-.02 Vi ca a 8 n 2 
2.05 0 9 x 0 
2.06 $ 9 8 4 
2.00 * e 
«alo 
-.2 2 
le 9 
-l6 
clin 
-.20 ; 
EXCEEDS point distance [em] 
  
S l5 2.5 34.5 4.5 5.5 6,5 7,5 8.5 9.5 10.5 11.5 12.5 11.5 18,5 15.5 16.5 17.5 18.5 19.5 
EXCEEDS | i 
+50 
1 
se 9?! fig. 4 normed covariance / cross covariance X - component 
oe A I 
+2 3! block Jaemijaervi, adjustment with 14 additional parameters 
... H 
.38 \ 
„36 
136 | 
2:32 
230 i 
«28 et 
26 \ 
20 1 
‚22 
‚20 | 
la | 
16 i 
la L + 
H ! e 9 
98 is 7 c 
96 ic 8 € 4 8 c a K 
‚0. A € A à x 3 s 3 12 AN 8 e M 
2% A e ca esed = 2 ens A) et i 
0.00  -—— => PA = T 48390g229499922- aT Tee emmy = Ll 
-.02 731 8 = à + 3° © $ 4 € WV-.5^4*9 4 
2.04 » 0 78 3 > 2 4 t S3 " D a 
2.06 y + * 4 3 0 € 2 F 4 + ] 
-.08 ° s 8 2... " 9 $€ 
2.190 2 8 3 
-.2 
is ° 
“ult 
2.15 
-.28 
ELCEENS point distance [em] 
  
5 1.5 2.5 3.5 4.5 5.5 6,5 7,5 85.5 9.5 10.5 11,5 12.5 1305 16,5 15,5 16.5 17.5 18.5 19,5 
The auto covariance cannot be used for the analysis, because the residuals 
in the same photo are not independent. Neighboured points in the same photo 
usual have a negative correlation coefficiant, caused by the determistic 
and this is quite different from the cross covariance. The normed cross 
covariance of the residuals of a block adjustment with additional parameters 
to which the following explanations are limited, usual is small - in the 
case of Jaemijaervi it is between * 0.1 and - 0.1, in the mean 0.03. But 
there is the exception of the normed cross covariance of points in differ- 
ent photos, but with approximately the same location in the photo. The 
normed cross covariance for distances up to 1 cm is reaching a value of 
0.68 in the block Jaemijaervi. This cross covarance is depending upon the 
neighbourhood of the photos - it is a little smaller if the difference of 
the photo numbers is larger.