<s>
Hyperspectral	B-Application
imaging	I-Application
collects	O
and	O
processes	O
information	O
from	O
across	O
the	O
electromagnetic	O
spectrum	O
.	O
</s>
<s>
The	O
goal	O
of	O
hyperspectral	B-Application
imaging	I-Application
is	O
to	O
obtain	O
the	O
spectrum	O
for	O
each	O
pixel	B-Algorithm
in	O
the	O
image	O
of	O
a	O
scene	O
,	O
with	O
the	O
purpose	O
of	O
finding	O
objects	O
,	O
identifying	O
materials	O
,	O
or	O
detecting	O
processes	O
.	O
</s>
<s>
There	O
are	O
push	O
broom	O
scanners	O
and	O
the	O
related	O
whisk	B-Algorithm
broom	I-Algorithm
scanners	I-Algorithm
(	O
spatial	O
scanning	O
)	O
,	O
which	O
read	O
images	O
over	O
time	O
,	O
band	O
sequential	O
scanners	O
(	O
spectral	O
scanning	O
)	O
,	O
which	O
acquire	O
images	O
of	O
an	O
area	O
at	O
different	O
wavelengths	O
,	O
and	O
snapshot	O
hyperspectral	B-Application
imaging	I-Application
,	O
which	O
uses	O
a	O
staring	B-Algorithm
array	I-Algorithm
to	O
generate	O
an	O
image	O
in	O
an	O
instant	O
.	O
</s>
<s>
In	O
hyperspectral	B-Application
imaging	I-Application
,	O
the	O
recorded	O
spectra	O
have	O
fine	O
wavelength	O
resolution	O
and	O
cover	O
a	O
wide	O
range	O
of	O
wavelengths	O
.	O
</s>
<s>
Hyperspectral	B-Application
imaging	I-Application
measures	O
continuous	O
spectral	O
bands	O
,	O
as	O
opposed	O
to	O
multiband	O
imaging	O
which	O
measures	O
spaced	O
spectral	O
bands	O
.	O
</s>
<s>
Engineers	O
build	O
hyperspectral	B-Application
sensors	O
and	O
processing	O
systems	O
for	O
applications	O
in	O
astronomy	O
,	O
agriculture	O
,	O
molecular	O
biology	O
,	O
biomedical	O
imaging	O
,	O
geosciences	O
,	O
physics	O
,	O
and	O
surveillance	O
.	O
</s>
<s>
Hyperspectral	B-Application
sensors	O
look	O
at	O
objects	O
using	O
a	O
vast	O
portion	O
of	O
the	O
electromagnetic	O
spectrum	O
.	O
</s>
<s>
Figuratively	O
speaking	O
,	O
hyperspectral	B-Application
sensors	O
collect	O
information	O
as	O
a	O
set	O
of	O
'	O
images	O
 '	O
.	O
</s>
<s>
These	O
'	O
images	O
 '	O
are	O
combined	O
to	O
form	O
a	O
three-dimensional	O
(	O
x	O
,	O
y	O
,	O
λ	O
)	O
hyperspectral	B-Application
data	B-Algorithm
cube	I-Algorithm
for	O
processing	O
and	O
analysis	O
,	O
where	O
x	O
and	O
y	O
represent	O
two	O
spatial	O
dimensions	O
of	O
the	O
scene	O
,	O
and	O
λ	O
represents	O
the	O
spectral	O
dimension	O
(	O
comprising	O
a	O
range	O
of	O
wavelengths	O
)	O
.	O
</s>
<s>
Technically	O
speaking	O
,	O
there	O
are	O
four	O
ways	O
for	O
sensors	O
to	O
sample	O
the	O
hyperspectral	B-Application
cube	O
:	O
Spatial	O
scanning	O
,	O
spectral	O
scanning	O
,	O
snapshot	O
imaging	O
,	O
and	O
spatio-spectral	O
scanning	O
.	O
</s>
<s>
Hyperspectral	B-Application
cubes	O
are	O
generated	O
from	O
airborne	O
sensors	O
like	O
NASA	O
's	O
Airborne	O
Visible/Infrared	O
Imaging	B-Algorithm
Spectrometer	I-Algorithm
(	O
AVIRIS	O
)	O
,	O
or	O
from	O
satellites	O
like	O
NASA	O
's	O
EO-1	O
with	O
its	O
hyperspectral	B-Application
instrument	O
Hyperion	O
.	O
</s>
<s>
If	O
the	O
scanner	O
detects	O
a	O
large	O
number	O
of	O
fairly	O
narrow	O
frequency	O
bands	O
,	O
it	O
is	O
possible	O
to	O
identify	O
objects	O
even	O
if	O
they	O
are	O
only	O
captured	O
in	O
a	O
handful	O
of	O
pixels	B-Algorithm
.	O
</s>
<s>
If	O
the	O
pixels	B-Algorithm
are	O
too	O
large	O
,	O
then	O
multiple	O
objects	O
are	O
captured	O
in	O
the	O
same	O
pixel	B-Algorithm
and	O
become	O
difficult	O
to	O
identify	O
.	O
</s>
<s>
If	O
the	O
pixels	B-Algorithm
are	O
too	O
small	O
,	O
then	O
the	O
intensity	O
captured	O
by	O
each	O
sensor	O
cell	O
is	O
low	O
,	O
and	O
the	O
decreased	O
signal-to-noise	O
ratio	O
reduces	O
the	O
reliability	O
of	O
measured	O
features	O
.	O
</s>
<s>
The	O
acquisition	O
and	O
processing	O
of	O
hyperspectral	B-Application
images	I-Application
is	O
also	O
referred	O
to	O
as	O
imaging	O
spectroscopy	O
or	O
,	O
with	O
reference	O
to	O
the	O
hyperspectral	B-Application
cube	O
,	O
as	O
3D	O
spectroscopy	O
.	O
</s>
<s>
There	O
are	O
four	O
basic	O
techniques	O
for	O
acquiring	O
the	O
three-dimensional	O
(	O
x	O
,	O
y	O
,	O
λ	O
)	O
dataset	O
of	O
a	O
hyperspectral	B-Application
cube	O
.	O
</s>
<s>
Hyperspectral	B-Application
imaging	I-Application
(	O
HSI	O
)	O
devices	O
for	O
spatial	O
scanning	O
obtain	O
slit	O
spectra	O
by	O
projecting	O
a	O
strip	O
of	O
the	O
scene	O
onto	O
a	O
slit	O
and	O
dispersing	O
the	O
slit	O
image	O
with	O
a	O
prism	O
or	O
a	O
grating	O
.	O
</s>
<s>
With	O
these	O
line-scan	B-Algorithm
cameras	I-Algorithm
,	O
the	O
spatial	O
dimension	O
is	O
collected	O
through	O
platform	O
movement	O
or	O
scanning	O
.	O
</s>
<s>
Nonetheless	O
,	O
line-scan	O
systems	O
are	O
particularly	O
common	O
in	O
remote	B-Application
sensing	I-Application
,	O
where	O
it	O
is	O
sensible	O
to	O
use	O
mobile	O
platforms	O
.	O
</s>
<s>
A	O
special	O
case	O
of	O
line	O
scanning	O
is	O
point	O
scanning	O
(	O
with	O
a	O
whisk	B-Algorithm
broom	I-Algorithm
scanner	I-Algorithm
)	O
,	O
where	O
a	O
point-like	O
aperture	O
is	O
used	O
instead	O
of	O
a	O
slit	O
,	O
and	O
the	O
sensor	O
is	O
essentially	O
one-dimensional	O
instead	O
of	O
2-D	O
.	O
</s>
<s>
The	O
spatial	O
features	O
on	O
each	O
of	O
the	O
images	O
may	O
be	O
used	O
to	O
realign	O
the	O
pixels	B-Algorithm
.	O
</s>
<s>
HSI	O
devices	O
for	O
non-scanning	O
yield	O
the	O
full	O
datacube	B-Algorithm
at	O
once	O
,	O
without	O
any	O
scanning	O
.	O
</s>
<s>
Figuratively	O
speaking	O
,	O
a	O
single	O
snapshot	O
represents	O
a	O
perspective	O
projection	O
of	O
the	O
datacube	B-Algorithm
,	O
from	O
which	O
its	O
three-dimensional	O
structure	O
can	O
be	O
reconstructed	O
.	O
</s>
<s>
The	O
most	O
prominent	O
benefits	O
of	O
these	O
snapshot	O
hyperspectral	B-Application
imaging	I-Application
systems	O
are	O
the	O
snapshot	O
advantage	O
(	O
higher	O
light	O
throughput	O
)	O
and	O
shorter	O
acquisition	O
time	O
.	O
</s>
<s>
A	O
number	O
of	O
systems	O
have	O
been	O
designed	O
,	O
including	O
computed	B-Algorithm
tomographic	I-Algorithm
imaging	I-Algorithm
spectrometry	I-Algorithm
(	O
CTIS	O
)	O
,	O
fiber-reformatting	O
imaging	O
spectrometry	O
(	O
FRIS	O
)	O
,	O
integral	O
field	O
spectroscopy	O
with	O
lenslet	O
arrays	O
(	O
IFS-L	O
)	O
,	O
multi-aperture	O
integral	O
field	O
spectrometer	O
(	O
Hyperpixel	O
Array	O
)	O
,	O
integral	O
field	O
spectroscopy	O
with	O
image	O
slicing	O
mirrors	O
(	O
IFS-S	O
)	O
,	O
image-replicating	O
imaging	O
spectrometry	O
(	O
IRIS	O
)	O
,	O
filter	O
stack	O
spectral	O
decomposition	O
(	O
FSSD	O
)	O
,	O
coded	O
aperture	O
snapshot	O
spectral	O
imaging	O
(	O
CASSI	O
)	O
,	O
image	O
mapping	O
spectrometry	O
(	O
IMS	O
)	O
,	O
and	O
multispectral	O
Sagnac	O
interferometry	O
(	O
MSI	O
)	O
.	O
</s>
<s>
In	O
an	O
effort	O
to	O
reduce	O
the	O
computational	O
demands	O
and	O
potentially	O
the	O
high	O
cost	O
of	O
non-scanning	O
hyperspectral	B-Application
instrumentation	O
,	O
prototype	O
devices	O
based	O
on	O
Multivariate	O
Optical	O
Computing	O
have	O
been	O
demonstrated	O
.	O
</s>
<s>
Hyperspectral	B-Application
imaging	I-Application
is	O
part	O
of	O
a	O
class	O
of	O
techniques	O
commonly	O
referred	O
to	O
as	O
spectral	O
imaging	O
or	O
spectral	O
analysis	O
.	O
</s>
<s>
The	O
term	O
“	O
hyperspectral	B-Application
imaging	I-Application
”	O
derives	O
from	O
the	O
development	O
of	O
NASA	O
's	O
Airborne	O
Imaging	B-Algorithm
Spectrometer	I-Algorithm
(	O
AIS	O
)	O
and	O
AVIRIS	O
in	O
the	O
mid-1980s	O
.	O
</s>
<s>
Although	O
NASA	O
prefers	O
the	O
earlier	O
term	O
“	O
imaging	O
spectroscopy	O
”	O
over	O
“	O
hyperspectral	B-Application
imaging	I-Application
,	O
”	O
use	O
of	O
the	O
latter	O
term	O
has	O
become	O
more	O
prevalent	O
in	O
scientific	O
and	O
non-scientific	O
language	O
.	O
</s>
<s>
Hyperspectral	B-Application
imaging	I-Application
is	O
related	O
to	O
multispectral	O
imaging	O
.	O
</s>
<s>
Hyperspectral	B-Application
imaging	I-Application
(	O
HSI	O
)	O
uses	O
continuous	O
and	O
contiguous	O
ranges	O
of	O
wavelengths	O
(	O
e.g.	O
</s>
<s>
Hyperspectral	B-Application
deals	O
with	O
imaging	O
narrow	O
spectral	O
bands	O
over	O
a	O
continuous	O
spectral	O
range	O
,	O
producing	O
the	O
spectra	O
of	O
all	O
pixels	B-Algorithm
in	O
the	O
scene	O
.	O
</s>
<s>
A	O
sensor	O
with	O
only	O
20	O
bands	O
can	O
also	O
be	O
hyperspectral	B-Application
when	O
it	O
covers	O
the	O
range	O
from	O
500	O
to	O
700nm	O
with	O
20	O
bands	O
each	O
10nm	O
wide	O
.	O
</s>
<s>
These	O
sensors	O
often	O
have	O
(	O
but	O
not	O
necessarily	O
)	O
a	O
low	O
spatial	O
resolution	O
of	O
several	O
pixels	B-Algorithm
only	O
,	O
a	O
restriction	O
imposed	O
by	O
the	O
high	O
data	O
rate	O
.	O
</s>
<s>
Hyperspectral	B-Application
remote	B-Application
sensing	I-Application
is	O
used	O
in	O
a	O
wide	O
array	O
of	O
applications	O
.	O
</s>
<s>
Although	O
originally	O
developed	O
for	O
mining	O
and	O
geology	O
(	O
the	O
ability	O
of	O
hyperspectral	B-Application
imaging	I-Application
to	O
identify	O
various	O
minerals	O
makes	O
it	O
ideal	O
for	O
the	O
mining	O
and	O
oil	O
industries	O
,	O
where	O
it	O
can	O
be	O
used	O
to	O
look	O
for	O
ore	O
and	O
oil	O
)	O
,	O
it	O
has	O
now	O
spread	O
into	O
fields	O
as	O
widespread	O
as	O
ecology	O
and	O
surveillance	O
,	O
as	O
well	O
as	O
historical	O
manuscript	O
research	O
,	O
such	O
as	O
the	O
imaging	O
of	O
the	O
Archimedes	O
Palimpsest	O
.	O
</s>
<s>
On	O
a	O
smaller	O
scale	O
,	O
NIR	O
hyperspectral	B-Application
imaging	I-Application
can	O
be	O
used	O
to	O
rapidly	O
monitor	O
the	O
application	O
of	O
pesticides	O
to	O
individual	O
seeds	O
for	O
quality	O
control	O
of	O
the	O
optimum	O
dose	O
and	O
homogeneous	O
coverage	O
.	O
</s>
<s>
Although	O
the	O
cost	O
of	O
acquiring	O
hyperspectral	B-Application
images	I-Application
is	O
typically	O
high	O
,	O
for	O
specific	O
crops	O
and	O
in	O
specific	O
climates	O
,	O
hyperspectral	B-Application
remote	B-Application
sensing	I-Application
use	O
is	O
increasing	O
for	O
monitoring	O
the	O
development	O
and	O
health	O
of	O
crops	O
.	O
</s>
<s>
In	O
Australia	O
,	O
work	O
is	O
under	O
way	O
to	O
use	O
imaging	B-Algorithm
spectrometers	I-Algorithm
to	O
detect	O
grape	O
variety	O
and	O
develop	O
an	O
early	O
warning	O
system	O
for	O
disease	O
outbreaks	O
.	O
</s>
<s>
Furthermore	O
,	O
work	O
is	O
underway	O
to	O
use	O
hyperspectral	B-Application
data	O
to	O
detect	O
the	O
chemical	O
composition	O
of	O
plants	O
,	O
which	O
can	O
be	O
used	O
to	O
detect	O
the	O
nutrient	O
and	O
water	O
status	O
of	O
wheat	O
in	O
irrigated	O
systems	O
.	O
</s>
<s>
On	O
a	O
smaller	O
scale	O
,	O
NIR	O
hyperspectral	B-Application
imaging	I-Application
can	O
be	O
used	O
to	O
rapidly	O
monitor	O
the	O
application	O
of	O
pesticides	O
to	O
individual	O
seeds	O
for	O
quality	O
control	O
of	O
the	O
optimum	O
dose	O
and	O
homogeneous	O
coverage	O
.	O
</s>
<s>
In	O
2004	O
,	O
the	O
first	O
study	O
relating	O
this	O
problem	O
with	O
hyperspectral	B-Application
imaging	I-Application
was	O
published	O
.	O
</s>
<s>
Hyperspectral	B-Application
libraries	O
that	O
are	O
representative	O
of	O
the	O
diversity	O
of	O
ingredients	O
usually	O
present	O
in	O
the	O
preparation	O
of	O
compound	O
feeds	O
were	O
constructed	O
.	O
</s>
<s>
These	O
libraries	O
can	O
be	O
used	O
together	O
with	O
chemometric	O
tools	O
to	O
investigate	O
the	O
limit	O
of	O
detection	O
,	O
specificity	O
and	O
reproducibility	O
of	O
the	O
NIR	O
hyperspectral	B-Application
imaging	I-Application
method	O
for	O
the	O
detection	O
and	O
quantification	O
of	O
animal	O
ingredients	O
in	O
feed	O
.	O
</s>
<s>
Hyperspectral	B-Application
imaging	I-Application
can	O
provide	O
information	O
about	O
the	O
chemical	O
constituents	O
of	O
materials	O
which	O
makes	O
it	O
useful	O
for	O
waste	O
sorting	O
and	O
recycling	O
.	O
</s>
<s>
HSI	O
cameras	O
can	O
be	O
integrated	O
with	O
machine	B-General_Concept
vision	I-General_Concept
systems	O
and	O
,	O
via	O
simplifying	O
platforms	O
,	O
allow	O
end-customers	O
to	O
create	O
new	O
waste	O
sorting	O
applications	O
and	O
other	O
sorting/identification	O
applications	O
.	O
</s>
<s>
A	O
system	O
of	O
machine	O
learning	O
and	O
hyperspectral	B-Application
camera	I-Application
can	O
distinguish	O
between	O
12	O
different	O
types	O
of	O
plastics	O
such	O
as	O
PET	O
and	O
PP	O
for	O
automated	O
separation	O
of	O
waste	O
of	O
,	O
as	O
of	O
2020	O
,	O
highly	O
unstandardized	O
plastics	O
products	O
and	O
packaging	B-Algorithm
.	O
</s>
<s>
and	O
Optina	O
Diagnostics	O
to	O
test	O
the	O
use	O
of	O
hyperspectral	B-Application
photography	I-Application
in	O
the	O
diagnosis	O
of	O
retinopathy	O
and	O
macular	O
edema	O
before	O
damage	O
to	O
the	O
eye	O
occurs	O
.	O
</s>
<s>
The	O
metabolic	O
hyperspectral	B-Application
camera	I-Application
will	O
detect	O
a	O
drop	O
in	O
oxygen	O
consumption	O
in	O
the	O
retina	O
,	O
which	O
indicates	O
potential	O
disease	O
.	O
</s>
<s>
In	O
the	O
food	O
processing	O
industry	O
,	O
hyperspectral	B-Application
imaging	I-Application
,	O
combined	O
with	O
intelligent	O
software	O
,	O
enables	O
digital	O
sorters	O
(	O
also	O
called	O
optical	B-General_Concept
sorters	I-General_Concept
)	O
to	O
identify	O
and	O
remove	O
defects	O
and	O
foreign	O
material	O
(	O
FM	O
)	O
that	O
are	O
invisible	O
to	O
traditional	O
camera	O
and	O
laser	O
sorters	O
.	O
</s>
<s>
Adopting	O
hyperspectral	B-Application
imaging	I-Application
on	O
digital	O
sorters	O
achieves	O
non-destructive	O
,	O
100	O
percent	O
inspection	O
in-line	O
at	O
full	O
production	O
volumes	O
.	O
</s>
<s>
The	O
sorter	O
’s	O
software	O
compares	O
the	O
hyperspectral	B-Application
images	I-Application
collected	O
to	O
user-defined	O
accept/reject	O
thresholds	O
,	O
and	O
the	O
ejection	O
system	O
automatically	O
removes	O
defects	O
and	O
foreign	O
material	O
.	O
</s>
<s>
The	O
recent	O
commercial	O
adoption	O
of	O
hyperspectral	B-Application
sensor-based	O
food	O
sorters	O
is	O
most	O
advanced	O
in	O
the	O
nut	O
industry	O
where	O
installed	O
systems	O
maximize	O
the	O
removal	O
of	O
stones	O
,	O
shells	O
and	O
other	O
foreign	O
material	O
(	O
FM	O
)	O
and	O
extraneous	O
vegetable	O
matter	O
(	O
EVM	O
)	O
from	O
walnuts	O
,	O
pecans	O
,	O
almonds	O
,	O
pistachios	O
,	O
peanuts	O
and	O
other	O
nuts	O
.	O
</s>
<s>
Commercial	O
adoption	O
of	O
hyperspectral	B-Application
sorters	O
is	O
also	O
advancing	O
at	O
a	O
fast	O
pace	O
in	O
the	O
potato	O
processing	O
industry	O
where	O
the	O
technology	O
promises	O
to	O
solve	O
a	O
number	O
of	O
outstanding	O
product	O
quality	O
problems	O
.	O
</s>
<s>
Work	O
is	O
underway	O
to	O
use	O
hyperspectral	B-Application
imaging	I-Application
to	O
detect	O
“	O
sugar	O
ends	O
,	O
”	O
“	O
hollow	O
heart	O
”	O
and	O
“	O
common	O
scab	O
,	O
”	O
conditions	O
that	O
plague	O
potato	O
processors	O
.	O
</s>
<s>
Geological	O
samples	O
,	O
such	O
as	O
drill	O
cores	O
,	O
can	O
be	O
rapidly	O
mapped	O
for	O
nearly	O
all	O
minerals	O
of	O
commercial	O
interest	O
with	O
hyperspectral	B-Application
imaging	I-Application
.	O
</s>
<s>
Hyperspectral	B-Application
remote	B-Application
sensing	I-Application
of	O
minerals	O
is	O
well	O
developed	O
.	O
</s>
<s>
Hyperspectral	B-Application
surveillance	O
is	O
the	O
implementation	O
of	O
hyperspectral	B-Application
scanning	O
technology	O
for	O
surveillance	O
purposes	O
.	O
</s>
<s>
Hyperspectral	B-Application
imaging	I-Application
is	O
particularly	O
useful	O
in	O
military	O
surveillance	O
because	O
of	O
countermeasures	O
that	O
military	O
entities	O
now	O
take	O
to	O
avoid	O
airborne	O
surveillance	O
.	O
</s>
<s>
The	O
idea	O
that	O
drives	O
hyperspectral	B-Application
surveillance	O
is	O
that	O
hyperspectral	B-Application
scanning	O
draws	O
information	O
from	O
such	O
a	O
large	O
portion	O
of	O
the	O
light	O
spectrum	O
that	O
any	O
given	O
object	O
should	O
have	O
a	O
unique	O
spectral	O
signature	O
in	O
at	O
least	O
a	O
few	O
of	O
the	O
many	O
bands	O
that	O
are	O
scanned	O
.	O
</s>
<s>
Hyperspectral	B-Application
imaging	I-Application
has	O
also	O
shown	O
potential	O
to	O
be	O
used	O
in	O
facial	O
recognition	O
purposes	O
.	O
</s>
<s>
Facial	O
recognition	O
algorithms	O
using	O
hyperspectral	B-Application
imaging	I-Application
have	O
been	O
shown	O
to	O
perform	O
better	O
than	O
algorithms	O
using	O
traditional	O
imaging	O
.	O
</s>
<s>
Traditionally	O
,	O
commercially	O
available	O
thermal	O
infrared	O
hyperspectral	B-Application
imaging	I-Application
systems	O
have	O
needed	O
liquid	O
nitrogen	O
or	O
helium	O
cooling	O
,	O
which	O
has	O
made	O
them	O
impractical	O
for	O
most	O
surveillance	O
applications	O
.	O
</s>
<s>
In	O
2010	O
,	O
Specim	O
introduced	O
a	O
thermal	O
infrared	O
hyperspectral	B-Application
camera	I-Application
that	O
can	O
be	O
used	O
for	O
outdoor	O
surveillance	O
and	O
UAV	O
applications	O
without	O
an	O
external	O
light	O
source	O
such	O
as	O
the	O
sun	O
or	O
the	O
moon	O
.	O
</s>
<s>
In	O
astronomy	O
,	O
hyperspectral	B-Application
imaging	I-Application
is	O
used	O
to	O
determine	O
a	O
spatially-resolved	O
spectral	O
image	O
.	O
</s>
<s>
Since	O
a	O
spectrum	O
is	O
an	O
important	O
diagnostic	O
,	O
having	O
a	O
spectrum	O
for	O
each	O
pixel	B-Algorithm
allows	O
more	O
science	O
cases	O
to	O
be	O
addressed	O
.	O
</s>
<s>
In	O
astronomy	O
,	O
this	O
technique	O
is	O
commonly	O
referred	O
to	O
as	O
integral	O
field	O
spectroscopy	O
,	O
and	O
examples	O
of	O
this	O
technique	O
include	O
FLAMES	O
and	O
SINFONI	O
on	O
the	O
Very	O
Large	O
Telescope	O
,	O
but	O
also	O
the	O
Advanced	O
CCD	O
Imaging	B-Algorithm
Spectrometer	I-Algorithm
on	O
Chandra	O
X-ray	O
Observatory	O
uses	O
this	O
technique	O
.	O
</s>
<s>
These	O
threats	O
are	O
mostly	O
invisible	O
but	O
detectable	O
by	O
hyperspectral	B-Application
imaging	I-Application
technology	O
.	O
</s>
<s>
Recent	O
research	O
indicates	O
that	O
hyperspectral	B-Application
imaging	I-Application
may	O
be	O
useful	O
to	O
detect	O
the	O
development	O
of	O
cracks	O
in	O
pavements	O
which	O
are	O
hard	O
to	O
detect	O
from	O
images	O
taken	O
with	O
visible	O
spectrum	O
cameras	O
.	O
</s>
<s>
Hyperspectral	B-Application
imaging	I-Application
has	O
also	O
been	O
used	O
to	O
detect	O
cancer	O
,	O
identify	O
nerves	O
and	O
analyze	O
bruises	O
.	O
</s>
<s>
In	O
February	O
2019	O
,	O
an	O
organization	O
founded	O
by	O
the	O
world	O
's	O
major	O
space	O
industries	O
,	O
the	O
Consultative	O
Committee	O
for	O
Space	O
Data	O
Standards	O
(	O
)	O
,	O
approved	O
a	O
standard	O
for	O
both	O
lossless	O
and	O
near-lossless	O
compression	O
of	O
multispectral	O
and	O
hyperspectral	B-Application
images	I-Application
(	O
)	O
.	O
</s>
<s>
The	O
primary	O
advantage	O
to	O
hyperspectral	B-Application
imaging	I-Application
is	O
that	O
,	O
because	O
an	O
entire	O
spectrum	O
is	O
acquired	O
at	O
each	O
point	O
,	O
the	O
operator	O
needs	O
no	O
prior	O
knowledge	O
of	O
the	O
sample	O
,	O
and	O
postprocessing	O
allows	O
all	O
available	O
information	O
from	O
the	O
dataset	O
to	O
be	O
mined	O
.	O
</s>
<s>
Hyperspectral	B-Application
imaging	I-Application
can	O
also	O
take	O
advantage	O
of	O
the	O
spatial	O
relationships	O
among	O
the	O
different	O
spectra	O
in	O
a	O
neighbourhood	O
,	O
allowing	O
more	O
elaborate	O
spectral-spatial	O
models	O
for	O
a	O
more	O
accurate	O
segmentation	B-Algorithm
and	O
classification	O
of	O
the	O
image	O
.	O
</s>
<s>
Fast	O
computers	O
,	O
sensitive	O
detectors	O
,	O
and	O
large	O
data	O
storage	O
capacities	O
are	O
needed	O
for	O
analyzing	O
hyperspectral	B-Application
data	O
.	O
</s>
<s>
Significant	O
data	O
storage	O
capacity	O
is	O
necessary	O
since	O
uncompressed	O
hyperspectral	B-Application
cubes	O
are	O
large	O
,	O
multidimensional	O
datasets	O
,	O
potentially	O
exceeding	O
hundreds	O
of	O
megabytes	O
.	O
</s>
<s>
All	O
of	O
these	O
factors	O
greatly	O
increase	O
the	O
cost	O
of	O
acquiring	O
and	O
processing	O
hyperspectral	B-Application
data	O
.	O
</s>
<s>
Also	O
,	O
one	O
of	O
the	O
hurdles	O
researchers	O
have	O
had	O
to	O
face	O
is	O
finding	O
ways	O
to	O
program	O
hyperspectral	B-Application
satellites	O
to	O
sort	O
through	O
data	O
on	O
their	O
own	O
and	O
transmit	O
only	O
the	O
most	O
important	O
images	O
,	O
as	O
both	O
transmission	O
and	O
storage	O
of	O
that	O
much	O
data	O
could	O
prove	O
difficult	O
and	O
costly	O
.	O
</s>
<s>
As	O
a	O
relatively	O
new	O
analytical	O
technique	O
,	O
the	O
full	O
potential	O
of	O
hyperspectral	B-Application
imaging	I-Application
has	O
not	O
yet	O
been	O
realized	O
.	O
</s>
