bul 2004
] of FGI.
(CN
7183,
14.108
:16 000
:8 000
:8 000
:16 000
ture and
principal
) & p2).
physical
ble 2 for
lad quite
p and p2
relations
high (y0
not used
bility of
be accu-
red also
nd other
did not
ecially c
y all the
Ebner's
he block
rameters
included
ny para-
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part Bl. Istanbul 2004
Blo-
ck € x0 yO | kl k2 k3 pl p2 BI B2
do | 0.03 | 0.00 | 0.44 |-0.01 | 0.00 | 0.00 | 0.01 | 0.59 | -0.01 | 0.00
I dp | 0.02 | -0.43 | 0.00 | 0.02 | -0.02 | 0.02 | -0.61 | 0.00 | -0.02 | 0.02
dk | -0.01 | 0.02 | -0.02 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.01 | -0.08
do | 0.04 | -0.01 | 0.16 | -0.06
| dp | 0.06 | -0.13 | 0.00 | -0.02
dk |-0.01 | 0.02 | -0.02 | 0.00
lo 0.02 | 0.46 | 0.02 | -0.02 | 0.01 | 0.06 | 0.60
2 dp -0.49 | -0.06 | 0.05 | -0.05 | 0.05 | -0.60 | -0.07
ik 0.00 | -0.02 | 0.01 | -0.02 | 0.02 | -0.01 | 0.00
do -0.01 | 0.10 | -0.04
2 dp -0.10 | 0.01 | -0.03
dk 0.02 | -0.03 | 0.01
Table 2. An example of correlations of boresight unknowns
with physical image deformation parameters. Blocks: 1: the full
block with full GCPs, 2: I-block with 0 GCPs.
meters, and the parameters got instable. The use of first order
radial term and affinity parameters did not affect instability.
4.2 Comparison of various block structures
The results of the boresight calibration with various block
structures are shown in Figure 3. In Figure 3a the differences
of the boresight parameters of the reduced blocks and the full
blocks are given. Differences in dk were mostly 2-4 mgon. In
do and dq the differences were smaller than 2 mgon. The
apparent systematic difference is most likely caused by the
small stripwise systematic errors of GPS/IMU-attitude
observations. The standard deviations of the boresight
parameters of 10 calibration blocks were similar, so only their
averages are shown in Figure 3b. The average standard
deviation in the full block was about 0.7 mgon in do and de
and 1.2 mgon in dk, while in the I-block these values were
almost doubled. The results of I-blocks with and without GCPs
were practically the same.
The effect of the block structure on the interior orientation
corrections is shown in Figure 4. In Figure 4b the averages of
the standard deviations of interior orientation corrections for
different principal distances, imaging scales and block
structures are shown. The standard deviations were naturally
higher with the reduced blocks. The standard deviation of the
principal point was better than 3.5 pum in all the cases, which
can be considered acceptable. The standard deviation of the
principal distance with the I-block with 2 GCPs varied between
8-10 um, which is quite poor accuracy. The block structure did
not affect the values of xO and yO significantly (Figure 4a). The
variability of the principal distance was larger, typically 10-20
jum, as could be expected based on high standard deviation va-
lues. The determination of principal distance was advantageous
only with the optics 7183, which had >20 jum corrections.
In most of the blocks appeared about 10 cm global shift in X-
direction, thus also a global shift was determined for the blocks
with GCPs. It is noteworthy that with I-block with 2 GCPs
standard deviations of the global shift parameters were in dX
and dY 5-10 cm and in height 8-16 cm, so their determination
was not very accurate. In the full blocks the standard deviations
of the shift parameters were 2-3 cm.
4.3 Image deformations
The effects of the Ebner's parameters were evaluated for each
optics. The maximum effects of the Ebner's parameters on the
a) Boresight differences to full block Odo b) Average boresight
5 = = iB dp stdev
23 |n dk 5 ye sd doi
9 2 D ul = 4 gsd dp
E | = = um di
80 n Hnln |$ sd ek
2 ET ET ci T T ET o 3 4 ron
§ 1 HH HE E
S 24 | |] |; = se = 324
= 3: EI EIE omm Em 3
4 L] Li u
5 | pp A pe T Sa RT 14
LT pz E353 i | FT | l I I | I | 1 | l |
E] | | luci Rips be ud 9
29072 9^2 peperere tas 2-| 0 2|o 2 02] |
| | | |
2121 | 2136 | 209 | 2120 | 2124 | 2137 2128 2134 2129 | 2135 12 |?
Figure 3. Effect of block structure on boresight parameters. a) differences between the full block and the reduced blocks, b)
average boresight standard deviations of 10 calibration blocks. Symbols below the graphs indicate block structure (I), number of
GCPs (12, 2, 0) and block name (2121, etc.).
a) Interior orientation corrections [He b) Standard deviations =
20.00 rrr REET in SER aa epee Ee ee eR Tn mr -ójg xO 12:00 em = ” urnes pre
@ 10.00 + li Pe 1 ee fa su znmyol | 7100 = {myo
© os © m TE
Ë 0.00 cl Ibat mam ee er di ett errr s 8.00 1——gp- el m
Eo JUL LU LEE 3 E d JH lE E Joo
s Gi >
PT m visit TT bel] hh ® 0.00 A Bn IT 11 UL Ii 1
| BEE TINTE | : Fi T ul us
pa | HE | $ | d f dier rir nni vf
| | | | lo |12]2|0 | 2| 2 |o |o] 2 lola |o 2| olia 2 |o | | F d |
riii 1220 | | haie otio pohapel :2/2]o|12/2]o [12|2 0 |12|2]0
2120 | 2124 | 2137 | 2128 | 2194 | 2128 | 2135 | | à
mai | 213 | 2119 | | | | | | 153mm | 153mm | 214mm | 214mm,
| 43153, c=153 mm 7183, c=214 mm 13026, snm | 7163, c=214 mm 1:8000 | 1:16000 | 1:8000 | 1:16000
Figure 4. Effect of block structure on interior orientation corrections. a) Interior orientation corrections calculated using reduced
and full blocks, b) average interior orientation standard deviations for each scale/principal distance-combination. Symbols below
2, 0), block name (2121, etc.) and optics (13153, etc.).
graphs indicate block structure (f=tull, I), number of GCPs (12,
169
