ments 
are 
with 
iat in 
photo 
| loss 
photo 
tional 
are 
tment 
ni: SZ 
% and 
Irrors 
del to 
matic 
ectly. 
block 
the 
ificial 
curacy 
)efore 
block 
lising. 
0 t0 3 
jucing 
50 the 
ed on 
j for 
mining 
ith 15 
17mm, 
indent 
onding 
known 
points 
-photo 
jround 
photo orientations cannot be used for set up of 
the model in an analytical plotter. Caused by the 
uncertainty of the photo orientations it is not 
possible to have an exact self calibration by 
additional parameters. The calculated systematic 
image errors are smeared like in the case of the 
block adjustment by independent models. In 
addition there is a negative influence of the 
relative orientation in the procedure. The 
distribution of the points in the models are 
different from model to model, so in the relative 
orientation remaining errors are influencing also 
the other points in the model. The same is 
happening with the systematic image errors. 
3. Empirical Bundle Block Adjustment 
with Model Coordinates 
The results of different bundle block adjustments 
with the original and the quasi photo coordinates 
have been analysed. The results of two blocks 
are shown more in detail. 
3.1 Block Jämijärvi 
The test block Jämijärvi is located in Finland. The 
photo coordinates have been used for comparison 
of data handling in bundle block adjustments by 
the WG III/3 of the ISPRS in 1978. For the points 
in the center of the block ground coordinates 
with standard deviations better than *1.5mm are 
available. 
photo scale: 1 : 3300 
focal length: 152mm 
side lap: 60% + crosswise also 60% 
number of used photos: 88 
number of photos/ground point: 
mean: 8 maximal: 17 
Table 1: technical data test block Jämijärvi 
  
  
  
x x 
; x 
Hos mum sie ; x i 
TK ---u i ! 
x mM. : 
; ¥ Tu... N x 
; i i "—€——— X : 
' > "S 1 
; : ; : on ed. 
s ï ; i ! DUX 
Ness ¥ x ; ; ; 
Kinn ; x x X 
1 1€ X. 1 : 
i : RN nm A 
: T + ne Me d 
2 ; : s ; LI 
NEU eei aud j foto faci x 
: - x 
i Y---*....d x 2 
: : Fax Fx * 1 
Populo wa eed 
Xen . 1 ; 1 + Xx 
En : X i x 
i * us m : : ! 
: . 1 pu : 
: SERA 
Mes : i x ? ! Home se emi 
ait + - ! : x x 
1 "M em ir. : x 
i Ferme Led 
: . 1 eres : 
1 ; x ! pter. A 
x i ; I cd dám 
1 : + x . : 
; x ! ; ! ; 
! ! / ! ; 
i ! = . : : 
; : 
Y mas T hs ni ; 
x : . : 
7 
: i i 
x x x 
a 
2 
-500 
1: 13857 
  
TWERTTSETVT 
Fig. 3: configuration of block Jämijärvi 
Following configurations have been used: block 
with 6 strips (E-W-direction 60% sidelap), 11 
strips (E-W-direction 60% sidelap + N-S-direction 
60% sidelap), both blocks with the control point 
configuration 9 (8 complete points at the periphery 
- 4 base length distance, + 1 vertical point in the 
center), with the configuration 13 (same complete 
control points like 9 + 5 vertical control points - 
raster 2 -4 base length distance) and with the 
configuration 36 (20 complete control points at 
the periphery - 2 base length distance, + 16 
vertical control points - raster 2 base length 
distance). 
If model coordinates (from analog or analytical 
plotters) are used, the original photo orientation 
for the transformation from the united models to 
the quasi-photo coordinates is not known. By 
this reason the model coordinates have been 
rotated before transformation to check the 
influenceoftheserotationstotheblockadjustment. 
A projection center 
   
ko 
que? RO 
  
original photo 
bundle of rays: 
Fig. 4: relation between original and quasi photo 
done without 
The transformations have been 
photo, with +42 
artificial rotation of the quasi 
grads, +5grads, *7grads and #10grads. 
  
rotation Jämijärvi 8 horizontal, 9 vertical control points 
  
  
of photos 6 photo strips 11 photo strips 
no self calib.| self calibration no self calibr. | self calibration 
Sx Sy Sz Sx Sy Sz Sx Sy Sz Sx Sy Sz 
reference] 30 17 88 11 10 2? 13 14 58 118 12 
0 29 18 85 11 12 29 14 16 60 10 8 156 
2 grads 29 16 86 12 12 28 13 15 61 10 8 28 
5 grads 26 20 98 19 14 46 17.16 65 15 10 27 
10grads | 35 20110 23 14 73 13 16 65 11 10 40 
Jamijarvi 8 horizontal, 13 vertical control points 
reference | 28 17 42 10 10 24 13 14 32 11 38 1i 
0 28 17 42 12 12 28 14 17 30 11 8 14 
14 16 32 10 8 14 
0 
0 
2 grads 28 16 43 12 12 29 
5 grads 26 19 48 19 14 42 17 16 36 15-1 
10grads | 34 20 52 25 14 43 13 17 34 15 1 
  
  
  
  
  
  
Jämojärvi 20 horizontal, 36 vertical control points 
reference| 20 1 8-8 15 9 921 76:11 
0 19 14 25 9 9 20 10 10 19 8 7.12 
2 grads 19 13 26 9 8 20 10 10 20 7 7 1 
5 grads 20 16 34 13 12 28 12 11 25 10 9 17 
  
  
Table 2: accuracy of the ground coordinates [mm] 
determined by bundle adjustment with the 
original (reference) and the quasi photo 
coordinates depending upon the rotation of the 
quasi photo coordinates 
587