To get an impression about the success rate of AMOR, we
present a test with a block of 195 aerial images which are
used for the orthophoto production.
In the test the size of the search area for every control point
model was 70x70 meters or 500x500 pixels. The exterior ori-
entation from AMOR was compared to the one of a bundle
| point block adjustment with the roof points as control points. Ta-
lel ble 2 shows the result of the selfdiagnosis of AMOR.
ged AMOR compared to bun- Selfdiagnosis of AMOR
dle block adjustment
correct | incorrect
same result (correct) 88 % 4%
other result (incorrect) | 0% | 8%
ation of”)
Table 2: Result of a test with AMOR
The expensive error type ll, a failed orientation classified as
r<type> correct by the selfdiagnosis of AMOR has not yet occured.
latabase 8% of the images could not be orientated. The success rate
of 88 % will vary strongly when one changes the size of the
search area.
an object Because of the need of orientation values in other applications
our approach seems to be interesting in other fields, too. It
is possible to generalize the ideas of the presented processes
because the used 3D-edges as features for the automatic ex-
f data terior orientation occur in nearly every application.
MByte
MByte
MByte
MByte REFERENCES
3Byte
[D.W. Embley, 1992] D.W. Embley, B.D. Kurtz, S. W.
tfalia (1992). Object-Oriented Systems Analysis. Yourdon Press.
[Lang et al., 1995] Lang, F., Loecherbach, T., and Schickler,
W. (1995). A One-eye Stereo System for Semi-automatic
e POST. 3-D Building Extraction. International Journal for Geo-
> for te matics, 9(6):6-8.
miner ato [Schickler, 1994] Schickler, W. (1994). Towards Automation
lass with in Photogrammetry - An Example: Automatic Exterior
The in- Orientation. Geodetical Info Magazine, 7(4):32-35.
the class
es of the
nes' and
n POST-
'ormation
e module
grated in
partment
ZA called
n directly
night for
models is
ram.
223
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B2. Vienna 1996
