hul 2004
ad been
this long
nes may
ow parts
r. These
process
isted for
> scanner
21 um
ammetric
the error
the GIS-
ting from
into three
nd Block
the AAT
ght lines
ult of the
he above
yr control
3." 2003).
GPS. The
ts of the
ne Trento
acy of the
e used as
lation.
problems
e pictures.
tie points
images of
h residuals
entation of
blem, the
mage data,
acy of the
e expected
ith each of
10 and 30
4 Pixel)
‚4 Pixel)
‚0 Pixel)
),7 Pixel)
tic images
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004
Because of legal conditions AVT could not get the original
films. Therefore no investigations concerning the mentioned
geometrical problems had been possible.
MATCH-AT permits sub-block handling with free block
adjustment (see 3.3). To reduce the mentioned problems for the
automatic tie point detection within the matching process, the
following strategy was used:
1. Further splitting into sub-subblocks with similar
image scales, same flight date and flight time.
2. Using a rough DTM and approximate values of the
projection centres of each image (see 3.2) for the
matching procedure.
3. Additional manual tie point measuring in problematic
areas of neighbouring images to achieve a stronger
connection.
4. Connecting sub-subblocks by manually tie point
measuring.
5. Common adjustment of connected sub-subblocks.
3.2 Additional information for the block adjustment
As mentioned in 2.1 approximate coordinates for the projection
centres were given. The flight company informed us, that the
standard deviation of each coordinate is about 1.0 m.
As aforementioned, different photo scale, large height
differences in the terrain, different texture because of changed
vegetation within the images of neighbouring flight lines caused
some problems for the matching procedure. This process can be
improved by adding DTM-data. The “DEM 2000 — Digital
Elevation Model” data with a grid interval of 100 m was used
during the AAT. These data were submitted by the PAB.
3.3 Used AAT-Software
INPHO GmbH is offering two software tools for AAT -
MATCH-AT and inBLOCK (Saile, 2003). AVT is using
MATCH-AT, the automatic digital aerial triangulation software
for the computation of bundle blocks. The main features of this
adjustment software are:
- . Support for all standard requirements such as
processing of blocks of any size and any overlap
- . Sub-block handling with free block adjustment
- A Point selection, point transfer, measurement and
block adjustment are integrated into one process
without user interactions
- A Automatic interior orientation
- Tie points are automatically selected in so-called ‘tie
point areas (v. Gruber)’
- . The AAT procedure combines sophisticated image
matching techniques with an integrated robust
approach
- GPS and INS data handling
- . Graphical representation of results and analysis
support
Because of the very specific conditions of the presented project
an independent review was done. The complete block was
computed with the new bundle block software inBLOCK too.
This software offers the following additional features:
- More flexible mathematical models are used for
correction of systematic image effects. Available
635
models are ‘Physical’ (5 parameters), ‘Brown’ (16
parameters) and ‘Ebner’ (12 parameters).
- J Complete statistical information:
e Internal and external reliability values for all
observations and unknowns.
e Traditional values like residuals, RMS and
standard deviations.
4. AAT RESULTS
This section describes the aerial triangulation results which
have been gotten meanwhile. In Table 2 the a priori standard
deviations are listed, which have been used for the bundle block
adjustment with MATCH-AT and inBLOCK.
Full control points
Height control points
Projection centres
X 7 Y 7 0.05 [m]; Z= 0.07 [m]
Z- 0.15 [m]
X = Y = Z = 1.00 [m]
Table 2. Used standard deviations (a priori) for the block
adjustment (MATCH-AT and inBLOCK)
4.1 AAT results using MATCH-AT
Tables 3a — 3c show the AAT results of the three adjusted
subblocks. The required accuracy for the aerial triangulation
results and the succeeding photogrammetric restitutions in the
scales 1:5.000 and 1:10000 are fully met!
Strips / images 59 / 949
Full control points 181
Height control points 70
RMS full control points 0.110/ 0.124/ 0.113 [m]
RMS height control points 0.188 [m]
RMS projection centres | 1.342/ 0.277/ 0.433 [m]
RMS adjusted coordinates | 0.065/ 0.071/ 0.162 [m]
Sigma 0 4.8 [um]
Table 3a. Block South, results of the block adjustment
Strips / images 64 / 1009
Full control points 146
Height control points 106
RMS full control points 0.121/ 0.110/ 0.128 [m]
RMS height control points 0.236 [m]
RMS projection centres | 1.014/ 0.226/ 0.395 [m]
RMS adjusted coordinates | 0.068/ 0.077/ 0.160 [m]
Sigma 0 4.8 [um]
Table 3b. Block Middle, results of the block adjustment
Strips / images 66 / 1213
Full control points 86
Height control points 44
RMS full control points 0.132/ 0.112/ 0.150 [m]
RMS height control points 0.361 [m]
RMS projection centres | 1.153/ 0.212/ 0.245 [m]
RMS adjusted coordinates | 0.053/ 0.060/ 0.132 [m]
Sigma 0 3.9 [um]
Table 3c. Block North, results of the block adjustment
