PT Unknown
AU Haase, K
TI Maritime Augmented Reality with a prioriknowledge of sea charts
PY 2013
PU Christian-Albrechts-Universität zu Kiel
WP https://macau.uni-kiel.de/receive/diss_mods_00011787
LA en
DE Augmented Reality; Topology S-57; ENC; Object Detection; Bayesian Network; Topologie S-57; Objekterkennung; Bayes'sche Netzwerke
AB The main objective of this thesis is to provide a concept to augment mar-
itime sea chart information into the camera view of the user. The benefit is
the simpler navigation due to the offered 3D information and the overlay onto
the real 3D environment.
In the maritime context special conditions hold. The sensor technologies have
to be reliable in the environment of a ship’s ferrous construction. The aug-
mentation of the objects has to be very precise due to the far distances of
observable objects on the sea surface. Furthermore, the approach has to be
reliable due to the wide range of light conditions. For a practical solution, the
system has to be mobile, light-weight and with a real-time performance. To
achieve this goal, the requirements are set, the possible measurement units
and the data base structure are presented.
First, the requirements are analyzed and a suitable system is designed. By
the combination of proper sensor techniques, the local position and orienta-
tion of the user can be estimated. To verify the concept, several prototypes
with exchangeable units have been evaluated. This first concept is based on
a marker-based approach which leads to some drawbacks.
To overcome the drawbacks, the second aspect is the improvement of the sys-
tem and the analysis of markerless approaches. One possible strategy will be
presented. The approach uses the statistical technique of Bayesian networks
to vote for single objects in the environment. By this procedure it will be
shown, that due to the a priori information the underlying sea chart system
has the most benefit. The analysis of the markerless approach shows, that the
sea charts structure has to be adapted to the new requirements of interactive
3D augmentation scenes. After the analysis of the chart data concept, an
approach for the optimization of the charts by building up an object-to-object
topology within the charts data and the Bayesian object detection approach
is presented. Finally, several evaluations show the performance of the imple-
mented evaluation application.
PI Kiel
ER