<s>
Microscope	B-Algorithm
image	I-Algorithm
processing	I-Algorithm
is	O
a	O
broad	O
term	O
that	O
covers	O
the	O
use	O
of	O
digital	B-Algorithm
image	I-Algorithm
processing	I-Algorithm
techniques	O
to	O
process	O
,	O
analyze	O
and	O
present	O
images	O
obtained	O
from	O
a	O
microscope	O
.	O
</s>
<s>
A	O
number	O
of	O
manufacturers	O
of	O
microscopes	O
now	O
specifically	O
design	O
in	O
features	O
that	O
allow	O
the	O
microscopes	O
to	O
interface	O
to	O
an	O
image	B-Algorithm
processing	I-Algorithm
system	O
.	O
</s>
<s>
While	O
this	O
required	O
the	O
use	O
of	O
a	O
frame	B-Device
grabber	I-Device
to	O
digitize	O
the	O
images	O
,	O
video	O
cameras	O
provided	O
images	O
at	O
full	O
video	O
frame	O
rate	O
(	O
25-30	O
frames	O
per	O
second	O
)	O
allowing	O
live	O
video	O
recording	O
and	O
processing	O
.	O
</s>
<s>
Today	O
,	O
acquisition	O
is	O
usually	O
done	O
using	O
a	O
CCD	B-Algorithm
camera	I-Algorithm
mounted	O
in	O
the	O
optical	O
path	O
of	O
the	O
microscope	O
.	O
</s>
<s>
Very	O
often	O
,	O
very	O
high	B-Algorithm
resolution	I-Algorithm
cameras	O
are	O
employed	O
to	O
gain	O
as	O
much	O
direct	O
information	O
as	O
possible	O
.	O
</s>
<s>
Often	O
digital	O
cameras	O
used	O
for	O
this	O
application	O
provide	O
pixel	B-Algorithm
intensity	O
data	O
to	O
a	O
resolution	O
of	O
12-16	O
bits	O
,	O
much	O
higher	O
than	O
is	O
used	O
in	O
consumer	O
imaging	O
products	O
.	O
</s>
<s>
Combined	O
with	O
the	O
high	O
image	B-Algorithm
resolution	I-Algorithm
,	O
this	O
approach	O
can	O
generate	O
vast	O
quantities	O
of	O
raw	O
data	O
,	O
which	O
can	O
be	O
a	O
challenge	O
to	O
deal	O
with	O
,	O
even	O
with	O
a	O
modern	O
computer	O
system	O
.	O
</s>
<s>
It	O
should	O
be	O
observed	O
that	O
while	O
current	O
CCD	B-Algorithm
detectors	O
allow	O
very	O
high	O
image	B-Algorithm
resolution	I-Algorithm
,	O
often	O
this	O
involves	O
a	O
trade-off	O
because	O
,	O
for	O
a	O
given	O
chip	O
size	O
,	O
as	O
the	O
pixel	B-Algorithm
count	O
increases	O
,	O
the	O
pixel	B-Algorithm
size	O
decreases	O
.	O
</s>
<s>
As	O
the	O
pixels	B-Algorithm
get	O
smaller	O
,	O
their	O
well	O
depth	O
decreases	O
,	O
reducing	O
the	O
number	O
of	O
electrons	O
that	O
can	O
be	O
stored	O
.	O
</s>
<s>
Because	O
microscope	O
images	O
have	O
an	O
intrinsic	O
limiting	O
resolution	O
,	O
it	O
often	O
makes	O
little	O
sense	O
to	O
use	O
a	O
noisy	O
,	O
high	B-Algorithm
resolution	I-Algorithm
detector	O
for	O
image	O
acquisition	O
.	O
</s>
<s>
A	O
more	O
modest	O
detector	O
,	O
with	O
larger	O
pixels	B-Algorithm
,	O
can	O
often	O
produce	O
much	O
higher	O
quality	O
images	O
because	O
of	O
reduced	O
noise	O
.	O
</s>
<s>
Image	B-Algorithm
processing	I-Algorithm
for	O
microscopy	O
application	O
begins	O
with	O
fundamental	O
techniques	O
intended	O
to	O
most	O
accurately	O
reproduce	O
the	O
information	O
contained	O
in	O
the	O
microscopic	O
sample	O
.	O
</s>
<s>
Another	O
class	O
of	O
common	O
2D	O
operations	O
called	O
image	O
convolution	B-Language
are	O
often	O
used	O
to	O
reduce	O
or	O
enhance	O
image	O
details	O
.	O
</s>
<s>
Such	O
"	O
blurring	O
"	O
and	O
"	O
sharpening	O
"	O
algorithms	O
in	O
most	O
programs	O
work	O
by	O
altering	O
a	O
pixel	B-Algorithm
's	O
value	O
based	O
on	O
a	O
weighted	O
sum	O
of	O
that	O
and	O
the	O
surrounding	O
pixels	B-Algorithm
(	O
a	O
more	O
detailed	O
description	O
of	O
kernel	O
based	O
convolution	B-Language
deserves	O
an	O
entry	O
for	O
itself	O
)	O
or	O
by	O
altering	O
the	O
frequency	O
domain	O
function	O
of	O
the	O
image	O
using	O
Fourier	B-Algorithm
Transform	I-Algorithm
.	O
</s>
<s>
Most	O
image	B-Algorithm
processing	I-Algorithm
techniques	O
are	O
performed	O
in	O
the	O
Frequency	O
domain	O
.	O
</s>
<s>
This	O
process	O
is	O
called	O
deconvolution	B-Algorithm
,	O
and	O
a	O
variety	O
of	O
algorithms	O
have	O
been	O
developed	O
,	O
some	O
of	O
great	O
mathematical	O
complexity	O
.	O
</s>
<s>
A	O
common	O
approach	O
is	O
to	O
create	O
an	O
image	O
mask	O
which	O
only	O
includes	O
pixels	B-Algorithm
that	O
match	O
certain	O
criteria	O
,	O
then	O
perform	O
simpler	O
scanning	O
operations	O
on	O
the	O
resulting	O
mask	O
.	O
</s>
