Even without editing of the aerial triangulation the results are
feasible due to the selection of suitable parameter sets for this
kind of aerial imagery. It is clearly proven, that no project
should be flown without proper GCPs introduced into the aerial
triangulation. Results indicate that an internal quality of some
dm can be reached with a moderate number of GCPs well
distributed. This still needs further investigation. The external
flight conditions do influence the aerial triangulation which can
be solved by additional manual tie point measurements.
As an alternative a newly developed approach UAVision for
selecting and matching of tie points using Open Source
Libraries showed very promising results to further automate the
AT process. UAVision provides a good, highly automated
alternative for tie point selection, tie point matching in the
block of UAV imagery and provision of all necessary input data
for a final bundle block adjustment. The run time is still rather
high and needs certainly to be improved, but this was not the
objective in this research. The used Open Source Libraries are
not directly usable and do require substantial improvements and
adaptations for the purposes investigated here. The advantage of
the OpenCV library was the well prepared documentation,
which supported the enhancements and adaptations. The
intended usage of the sba library was not possible. The
investigations clearly showed the limitations of this library
which is not feasible for our purposes. The intention is now to
a) either look for another open source library, or to b) use the
results of UAVision tie point matching as input for the Match-
AT final bundle adjustment.
Further investigations will be needed to check the effects of
camera calibration in flight. Other researchers report very
similar effects with other systems. Another unsolved problem is
the usage for stereo compilation, as the problem is the high endand
side-overlap which does require a smart selection of stereo
pairs for meaningful processing.