<s>
Data	B-General_Concept
fusion	I-General_Concept
is	O
the	O
process	O
of	O
integrating	O
multiple	O
data	O
sources	O
to	O
produce	O
more	O
consistent	O
,	O
accurate	O
,	O
and	O
useful	O
information	O
than	O
that	O
provided	O
by	O
any	O
individual	O
data	O
source	O
.	O
</s>
<s>
Data	B-General_Concept
fusion	I-General_Concept
processes	O
are	O
often	O
categorized	O
as	O
low	O
,	O
intermediate	O
,	O
or	O
high	O
,	O
depending	O
on	O
the	O
processing	O
stage	O
at	O
which	O
fusion	O
takes	O
place	O
.	O
</s>
<s>
Low-level	O
data	B-General_Concept
fusion	I-General_Concept
combines	O
several	O
sources	O
of	O
raw	O
data	O
to	O
produce	O
new	O
raw	O
data	O
.	O
</s>
<s>
The	O
expectation	O
is	O
that	O
fused	O
data	O
is	O
more	O
informative	O
and	O
synthetic	B-General_Concept
than	O
the	O
original	O
inputs	O
.	O
</s>
<s>
For	O
example	O
,	O
sensor	B-General_Concept
fusion	I-General_Concept
is	O
also	O
known	O
as	O
(	O
multi-sensor	O
)	O
data	B-General_Concept
fusion	I-General_Concept
and	O
is	O
a	O
subset	O
of	O
information	B-General_Concept
fusion	I-General_Concept
.	O
</s>
<s>
The	O
concept	O
of	O
data	B-General_Concept
fusion	I-General_Concept
has	O
origins	O
in	O
the	O
evolved	O
capacity	O
of	O
humans	O
and	O
animals	O
to	O
incorporate	O
information	O
from	O
multiple	O
senses	O
to	O
improve	O
their	O
ability	O
to	O
survive	O
.	O
</s>
<s>
In	O
the	O
mid-1980s	O
,	O
the	O
Joint	O
Directors	O
of	O
Laboratories	O
formed	O
the	O
Data	B-General_Concept
Fusion	I-General_Concept
Subpanel	O
(	O
which	O
later	O
became	O
known	O
as	O
the	O
Data	B-General_Concept
Fusion	I-General_Concept
Group	O
)	O
.	O
</s>
<s>
With	O
the	O
advent	O
of	O
the	O
World	O
Wide	O
Web	O
,	O
data	B-General_Concept
fusion	I-General_Concept
thus	O
included	O
data	O
,	O
sensor	O
,	O
and	O
information	B-General_Concept
fusion	I-General_Concept
.	O
</s>
<s>
The	O
JDL/DFIG	O
introduced	O
a	O
model	O
of	O
data	B-General_Concept
fusion	I-General_Concept
that	O
divided	O
the	O
various	O
processes	O
.	O
</s>
<s>
Currently	O
,	O
the	O
six	O
levels	O
with	O
the	O
Data	B-General_Concept
Fusion	I-General_Concept
Information	O
Group	O
(	O
DFIG	O
)	O
model	O
are	O
:	O
</s>
<s>
Despite	O
these	O
shortcomings	O
,	O
the	O
JDL/DFIG	O
models	O
are	O
useful	O
for	O
visualizing	O
the	O
data	B-General_Concept
fusion	I-General_Concept
process	O
,	O
facilitating	O
discussion	O
and	O
common	O
understanding	O
,	O
and	O
important	O
for	O
systems-level	O
information	B-General_Concept
fusion	I-General_Concept
design	O
.	O
</s>
<s>
In	O
the	O
geospatial	O
(	O
GIS	B-Application
)	O
domain	O
,	O
data	B-General_Concept
fusion	I-General_Concept
is	O
often	O
synonymous	O
with	O
data	B-General_Concept
integration	I-General_Concept
.	O
</s>
<s>
In	O
a	O
much	O
more	O
complicated	O
application	O
,	O
marine	O
animal	O
researchers	O
use	O
data	B-General_Concept
fusion	I-General_Concept
to	O
combine	O
animal	O
tracking	O
data	O
with	O
bathymetric	O
,	O
meteorological	O
,	O
sea	O
surface	O
temperature	O
(	O
SST	O
)	O
and	O
animal	O
habitat	O
data	O
to	O
examine	O
and	O
understand	O
habitat	O
utilization	O
and	O
animal	O
behavior	O
in	O
reaction	O
to	O
external	O
forces	O
such	O
as	O
weather	O
or	O
water	O
temperature	O
.	O
</s>
<s>
But	O
through	O
the	O
use	O
of	O
data	B-General_Concept
fusion	I-General_Concept
,	O
all	O
data	O
and	O
attributes	O
are	O
brought	O
together	O
into	O
a	O
single	O
view	O
in	O
which	O
a	O
more	O
complete	O
picture	O
of	O
the	O
environment	O
is	O
created	O
.	O
</s>
<s>
Hugh	O
Pederson	O
of	O
the	O
University	O
of	O
Tasmania	O
used	O
data	B-General_Concept
fusion	I-General_Concept
software	O
to	O
fuse	O
southern	O
rock	O
lobster	O
tracking	O
data	O
(	O
color-coded	O
for	O
in	O
yellow	O
and	O
black	O
for	O
day	O
and	O
night	O
,	O
respectively	O
)	O
with	O
bathymetry	O
and	O
habitat	O
data	O
to	O
create	O
a	O
unique	O
4D	O
picture	O
of	O
rock	O
lobster	O
behavior	O
.	O
</s>
<s>
In	O
applications	O
outside	O
of	O
the	O
geospatial	O
domain	O
,	O
differences	O
in	O
the	O
usage	O
of	O
the	O
terms	O
Data	B-General_Concept
integration	I-General_Concept
and	O
Data	B-General_Concept
fusion	I-General_Concept
apply	O
.	O
</s>
<s>
In	O
areas	O
such	O
as	O
business	B-General_Concept
intelligence	I-General_Concept
,	O
for	O
example	O
,	O
data	B-General_Concept
integration	I-General_Concept
is	O
used	O
to	O
describe	O
the	O
combining	O
of	O
data	O
,	O
whereas	O
data	B-General_Concept
fusion	I-General_Concept
is	O
integration	O
followed	O
by	O
reduction	O
or	O
replacement	O
.	O
</s>
<s>
Data	B-General_Concept
integration	I-General_Concept
might	O
be	O
viewed	O
as	O
set	O
combination	O
wherein	O
the	O
larger	O
set	O
is	O
retained	O
,	O
whereas	O
fusion	O
is	O
a	O
set	O
reduction	O
technique	O
with	O
improved	O
confidence	O
.	O
</s>
<s>
A	O
Data	B-General_Concept
fusion	I-General_Concept
based	O
approach	O
that	O
utilizes	O
the	O
road	O
side	O
collected	O
acoustic	O
,	O
image	O
and	O
sensor	O
data	O
has	O
been	O
shown	O
to	O
combine	O
the	O
advantages	O
of	O
the	O
different	O
individual	O
methods	O
.	O
</s>
<s>
Using	O
signal	O
processing	O
and	O
data	B-General_Concept
fusion	I-General_Concept
techniques	O
such	O
as	O
feature	O
generation	O
,	O
feasibility	O
study	O
and	O
principal	B-Application
component	I-Application
analysis	I-Application
(	O
PCA	O
)	O
such	O
sensory	O
data	O
will	O
greatly	O
improve	O
the	O
positive	O
rate	O
of	O
classifying	O
the	O
motion	O
and	O
contextual	O
relevant	O
status	O
of	O
the	O
device	O
.	O
</s>
<s>
Gaussian	B-General_Concept
processes	I-General_Concept
are	O
a	O
popular	O
machine	O
learning	O
model	O
.	O
</s>
<s>
If	O
an	O
auto-regressive	B-Algorithm
relationship	O
between	O
the	O
data	O
is	O
assumed	O
,	O
and	O
each	O
data	O
source	O
is	O
assumed	O
to	O
be	O
Gaussian	B-General_Concept
process	I-General_Concept
,	O
this	O
constitutes	O
a	O
non-linear	O
Bayesian	B-General_Concept
regression	I-General_Concept
problem	O
.	O
</s>
