International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004 
  
ORIMA APM module consequently. The exterior orientation 
parameters produced by the DGS was used as initial value 
during this process. The tie point pattern of 11 x 11 grid was 
used. GCPs and check points were observed manually and pre- 
adjustment was carried out to reject the blunders. Manual tie 
point observation was not carried out. Figure 9 shows tie point 
distribution of this block. 
  
Figure 9. Tie point distribution 
3. RESULT AND DISCUSSION 
. 
€ 9 9 oo «0000000 
9 900 9990 9 00000000000 
se... 99. 9.90004 
e 0 0 00000 
9 © 6 60 09090000 
. 0000000 0 0 
€ € © 00000 
0 00 00000 
® € € 09 00000002 
9€ 9090909900900 
€ 9009090 6 * 9 0 00000 
€ 0 0 00000 
. 0 0 0000000000 
. 0.0 0 00000000 
000.0. 000. 0008 .. © 00000 
9 0 9 00900 
. 0.0. 00.00000000 
0. 
° 00 00000 
. 0 0.00. 00000000 
ee eo... 
9e 9999 90 0 000 .. © 00000 
9 0.0 00000 
  
and all GCPs. The prior standard deviation values were 9° m 
for image measurement, 10cm for GPS position, 0.014 grad for 
phi and omega, 0.039 grad for kappa, respectively. Several 
cases were processed as written below: 
CASEI: Normal aerial triangulations 49GCPs, DGS was not 
used. * 
CASE? : Without GCPs 
CASE3 : 4GCPs without datum shift estimation 
CASEA : 4GCPs with datum shift estimation 
CASES : 1GCP without datum shift estimation 
CASE6 : 1GCP with datum shift estimation 
The result is shown in Table 10. By introducing exterior 
orientation parameters produced by DGS makes standard 
deviation value much smaller than normal AT (CASE 1 was 
compared with others). Comparisons among CASE 3 and CASE 
5, CASE 4 and CASE 6 show that the number of GCP do not 
effect for achieving the final accuracy. Even if the GCP is not 
used the result is almost same. CASE 4 and CASE 6 show that 
the datum shift estimation eliminates systematic error especially 
for the height and improve final accuracy. 
3.2 Result of misalignment estimation 
To confirm misalignment estimation accuracy, result of single 
strip, 2 strips, 3 strips and 4 strips cases are compared with the 
obtained results of all strips and utilized GCPs. The 4 GCPs lie 
at the corner of each block and are fixed. This result is shown in 
Tablel2. 
....... 
....... 
Table 10. Result of bundle adjustment 
3.1 Result of block adjustment 
Bundle block adjustment was performed consequently. Bundle 
block adjustment software BINGO (GIP) was used for all 
processing. Only 10 normal strips are used to evaluate block 
accuracy. The misalignment value was estimated using all strips 
862 
Even if it is a single strip, misalignment value estimated by 
adjustment is very close to what was estimated by using all 
strips and all GCPs. However if datum shifts estimation was 
introduced at least two or more strips were needed. It seems that 
more strips improve higher accuracy. 
In