International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B4, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
Site 1 - Vidy Bay
High population density area,
which acts as a major source of
micropolutants.
Site 2 - Deep Lake
introduced along the coastal
regions.
Deep water zones, which act as
the "sink" to the micropolutants
Lausanne
Villeneuve
Site 3 - Rhone Delta
Major source of water, sediments and
other materials, which determine the
environmental functioning of the lake.
Figure 2: Three representative regions surveyed in the course of the initial stage of the Élémo project (Adopted from www. elemo.ch).
v i X NS
Figure 3: Underwater trajectories of the MIR submersibles in the
Bay of Vidy obtained from the raw position data recorded by the
surface navigation system (Courtesy of J.-L. Loizeau).
Section 2.2 and finally outline the structure of the developed Web-
GIS platform in Section 2.4 before drawing our conclusions in Sec-
tion 3.
2 METHODOLOGY
2.1 Master navigation data
The first major challenge that we have encountered in our effort to
construct a comprehensive representation of the collected data was
the reliable estimation of the underwater trajectories of the sub-
mersibles Mir 1 and 2 throughout the course of the field campaign.
Evidently, an accurate position is essential for the correct interpre-
tation of the collected environmental data. Unfortunately, the po-
sition information recorded by the submarines' navigation system
was largely unreliable and had many obvious outliers. The raw
Figure 4: A typical frame from the video log enquired by the main
camera of the MIR submersible in the course of the field campaign
in lake Geneva.
trajectories of the submersibles in the Vidy bay are portrayed in
Figure 3.
Correspondingly, the navigation data describing the 3D trajectories
of the MIR submersibles has been comprehensively analysed and
consolidated. In particular, the reliability of the navigation data has
been improved significantly by the means of suppression of instan-
taneous outliers, as well as by non-linear interpolation of missing
GNSS data. The submersible navigation trajectories become par-
ticularly informative in the presence of high-resolution bathymetry
data, which is presently available for a considerable portion of the
dive campaign area.
2.2 Indexing of the video logs
The objective of video processing was twofold: to prepare daily
video clips suited for online streaming and to synchronise such
videos with navigation data. Video data was originally provided as
a set of 24 disk image files (68.04 GB) covering 50 daily missions
(a)
(b).
Figure
algori
(26 fr
record
softw:
were €
practi
scope.
the lei
Recog
Were
frame
quenc
(Open
tions {
video
form «
WE Cre
set of
Throu
est co
the sa
(b) res
obtain
formu
where
resent
nal im
digit ¢
Then
one re
(Comi
digital
groun
results
algori
was a
exam
total s
sequei
The re
Using
and to
23° 1]
While
carrie
curren
for we
to be «
age d:
