<s>
Source	O
separation	O
,	O
blind	B-Application
signal	I-Application
separation	I-Application
(	O
BSS	O
)	O
or	O
blind	B-Application
source	I-Application
separation	I-Application
,	O
is	O
the	O
separation	O
of	O
a	O
set	O
of	O
source	O
signals	O
from	O
a	O
set	O
of	O
mixed	O
signals	O
,	O
without	O
the	O
aid	O
of	O
information	O
(	O
or	O
with	O
very	O
little	O
information	O
)	O
about	O
the	O
source	O
signals	O
or	O
the	O
mixing	O
process	O
.	O
</s>
<s>
It	O
is	O
most	O
commonly	O
applied	O
in	O
digital	B-General_Concept
signal	I-General_Concept
processing	I-General_Concept
and	O
involves	O
the	O
analysis	O
of	O
mixtures	O
of	O
signals	O
;	O
the	O
objective	O
is	O
to	O
recover	O
the	O
original	O
component	O
signals	O
from	O
a	O
mixture	O
signal	O
.	O
</s>
<s>
The	O
human	O
brain	O
can	O
handle	O
this	O
sort	O
of	O
auditory	O
source	O
separation	O
problem	O
,	O
but	O
it	O
is	O
a	O
difficult	O
problem	O
in	O
digital	B-General_Concept
signal	I-General_Concept
processing	I-General_Concept
.	O
</s>
<s>
However	O
,	O
blind	B-Application
signal	I-Application
separation	I-Application
is	O
now	O
routinely	O
performed	O
on	O
multidimensional	B-Algorithm
data	I-Algorithm
,	O
such	O
as	O
images	B-Algorithm
and	O
tensors	B-Device
,	O
which	O
may	O
involve	O
no	O
time	O
dimension	O
whatsoever	O
.	O
</s>
<s>
Some	O
of	O
the	O
more	O
successful	O
approaches	O
are	O
principal	B-Application
components	I-Application
analysis	I-Application
and	O
independent	B-Algorithm
component	I-Algorithm
analysis	I-Algorithm
,	O
which	O
work	O
well	O
when	O
there	O
are	O
no	O
delays	O
or	O
echoes	O
present	O
;	O
that	O
is	O
,	O
the	O
problem	O
is	O
simplified	O
a	O
great	O
deal	O
.	O
</s>
<s>
The	O
field	O
of	O
computational	O
auditory	B-Application
scene	I-Application
analysis	I-Application
attempts	O
to	O
achieve	O
auditory	O
source	O
separation	O
using	O
an	O
approach	O
that	O
is	O
based	O
on	O
human	O
hearing	O
.	O
</s>
<s>
In	O
human	O
perception	O
this	O
ability	O
is	O
commonly	O
referred	O
to	O
as	O
auditory	B-Application
scene	I-Application
analysis	I-Application
or	O
the	O
cocktail	O
party	O
effect	O
.	O
</s>
<s>
The	O
separated	O
images	B-Algorithm
,	O
were	O
separated	O
using	O
and	O
the	O
using	O
Joint	B-Algorithm
Approximation	I-Algorithm
Diagonalization	I-Algorithm
of	I-Algorithm
Eigen-matrices	I-Algorithm
(	O
JADE	B-Algorithm
)	O
algorithm	O
which	O
is	O
based	O
on	O
independent	B-Algorithm
component	I-Algorithm
analysis	I-Algorithm
,	O
ICA	O
.	O
</s>
<s>
This	O
toolbox	O
method	O
can	O
be	O
used	O
with	O
multi-dimensions	O
but	O
for	O
an	O
easy	O
visual	O
aspect	O
images(2-D )	O
were	O
used	O
.	O
</s>
<s>
One	O
of	O
the	O
practical	O
applications	O
being	O
researched	O
in	O
this	O
area	O
is	O
medical	B-Application
imaging	I-Application
of	O
the	O
brain	O
with	O
magnetoencephalography	B-Algorithm
(	O
MEG	O
)	O
.	O
</s>
<s>
In	O
electroencephalogram	B-Application
(	O
EEG	B-Application
)	O
and	O
magnetoencephalography	B-Algorithm
(	O
MEG	O
)	O
,	O
the	O
interference	O
from	O
muscle	O
activity	O
masks	O
the	O
desired	O
signal	O
from	O
brain	B-Application
activity	I-Application
.	O
</s>
<s>
BSS	O
,	O
however	O
,	O
can	O
be	O
used	O
to	O
separate	O
the	O
two	O
so	O
an	O
accurate	O
representation	O
of	O
brain	B-Application
activity	I-Application
may	O
be	O
achieved	O
.	O
</s>
<s>
Blind	B-Application
source	I-Application
separation	I-Application
separates	O
the	O
set	O
of	O
mixed	O
signals	O
,	O
,	O
through	O
the	O
determination	O
of	O
an	O
'	O
unmixing	O
 '	O
matrix	O
,	O
,	O
to	O
'	O
recover	O
 '	O
an	O
approximation	O
of	O
the	O
original	O
signals	O
,	O
.	O
</s>
<s>
Since	O
the	O
chief	O
difficulty	O
of	O
the	O
problem	O
is	O
its	O
underdetermination	O
,	O
methods	O
for	O
blind	B-Application
source	I-Application
separation	I-Application
generally	O
seek	O
to	O
narrow	O
the	O
set	O
of	O
possible	O
solutions	O
in	O
a	O
way	O
that	O
is	O
unlikely	O
to	O
exclude	O
the	O
desired	O
solution	O
.	O
</s>
<s>
In	O
one	O
approach	O
,	O
exemplified	O
by	O
principal	B-Application
and	O
independent	B-Algorithm
component	I-Algorithm
analysis	I-Algorithm
,	O
one	O
seeks	O
source	O
signals	O
that	O
are	O
minimally	O
correlated	O
or	O
maximally	O
independent	B-Algorithm
in	O
a	O
probabilistic	O
or	O
information-theoretic	O
sense	O
.	O
</s>
<s>
A	O
common	O
theme	O
in	O
the	O
second	O
approach	O
is	O
to	O
impose	O
some	O
kind	O
of	O
low-complexity	O
constraint	O
on	O
the	O
signal	O
,	O
such	O
as	O
sparsity	B-Algorithm
in	O
some	O
basis	O
for	O
the	O
signal	O
space	O
.	O
</s>
<s>
There	O
are	O
different	O
methods	O
of	O
blind	B-Application
signal	I-Application
separation	I-Application
:	O
</s>
