ig using
ave been
ive been
rlapping
positives
ents.
are with
Romuald Kaczynski
Two first parameters and a’priori standard deviation of photo coordinates measurement have been roughly calculated.
Both blocks have been then elaborated without any intervention of the operator. Block No.2 has been adjusted with self-
calibration parameters with BINGO-F software. The data are shown in Table 3.
Description Block No.1 Block No.2
most accurate most accurate
parameters spliting factor = 4 spliting factor = 4
of AAT No. of tie points per area = 14 | No. of tie points per area = 14
selfcalibration = no selfcalibration = yes
total No. of tie points 5616 3955
No. of tie points per photo 190 209
No. of clusters of tie points
total / connecting strips 522/230 347/108
No. of suspect areas 1 8
Table 3. Parameters of AAT
From table 3 can be seen that there is only one suspect area in block No.1 and 8 in block No.2.
A minimum of one tie point has been fully transferred for 85 % of clusters connecting strips in block No.1 and for 71%
of clusters in block No 2.
About 40% of tie points connecting strips have been eliminated during block adjustment with blunder elimination in
block No 1 and 30 % in block No.2.
5 RESULTS
Results of block adjustment are shown in Table 4. Accuracy of photo coordinates in AAT as compared with semi-
automatic AT is 10 % lower for block No.1 and 24% lower for block No.2. Lower accuracy in the block No.2 is
probably due to long strips acquired in different times and not complete removal of radial deformations by the
additional self- calibration parameters.
Independent accuracy check of two blocks was done using well-distributed checkpoints. Results achieved in semi-
automatic and automatic aerial triangulation are similar.
Accuracy of exterior orientation parameters calculated by AAT are about 25% better for block No.1 and 15% for block
No.2 than that calculated by semi-automatic method.
RMS of differences of exterior orientation parameters achieved by the two methods is shown in line 9 Table 4. For
block No.1 they are 77 % higher than could be expected from mean standard deviations of exterior orientation
parameters for both methods (shown in line 8). The reason is that block No.1 is not regular (from 7 to 14 photos in strip)
and the maximum differences appear for the ten photos on the edge of the block. Maximum differences do not exceed
2.1 m for coordinates Xo,Yo, Zo and 54 mgon for o, q, K.
RMS of differences of exterior orientation parameters for block No.2 are about 30% higher then expected.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 459