Database Reference
In-Depth Information
regular trajectories as outliers. An adaptive process should be therefore consid-
ered in order to detect a new pattern and possibility remove an outdated one.
Finally, the approach described could be applied or extended to other kinds of
moving objects evolving in open spaces, especially those having 3D trajectories
(e.g., underwater vehicles or planes that behave quite similarly to ships).
11.4 Bibliographic Notes
Several maritime projects worked to enhance the tracking and monitoring of
vessels. This is portrayed for example in
MarNIS
(
2009
). These monitoring
systems use ARPA and AIS sensors as input.
Bole et al.
(
2012
) describe the
ARPA system in detail. In a similar way, the Association of Marine Aids to
Navigation and Lighthouse Authorities describes the AIS in
IALA
(
2004
). These
new tracking and monitoring systems are parts of e-Navigation defined by the
International Maritime Organization in
IMO
(
2008
). e-Navigation relies on
Electronic Navigation Chart (ENC), defined by the International Hydrographic
Organization in
IHO
(
2000
).
If the reader needs additional information about some special technical points
of this chapter, several articles can be read. For the filtering part,
Meratnia
and de By
(
2004
) serves as the base for the filtering process presented in this
chapter. For the similarity measure between trajectories, Frechet distance has
been selected.
Alt and Godau
(
1995
) explains why this measure is better for this
kind of data.
Devogele
(
2002
) describes the algorithm for discrete partial Frechet
distance. Matching process based on dynamic time warping is also possible; see
Sakoe and Chiba
(
1978
). Results are very similar but this later process can align
only whole trajectories. Some details about our architecture are introduced in
Bertrand et al.
(
2007
). Finally,
Etienne et al.
(
2012
) details the clustering process
and the spatio-temporal pattern based on box plot. This representation is defined
in
Tukey
(
1977
).
