I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
p
ute
r
E
ng
in
ee
ring
(
I
J
E
CE
)
Vo
l.
8
,
No
.
4
,
A
u
g
u
s
t
201
8
,
p
p
.
2
1
1
5
~2
1
2
5
I
SS
N:
2
0
8
8
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v8
i
4
.
p
p
2
1
1
5
-
2125
2115
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
co
r
e
.
co
m/
jo
u
r
n
a
ls
/in
d
ex
.
p
h
p
/
I
JE
C
E
A Neural
Net
w
o
rk
Appro
a
ch
to Id
entif
y
H
y
perspe
ct
ra
l
I
m
a
g
e Cont
ent
P
utt
a
s
w
a
m
y
M
.
R
.
1
,
B
a
la
m
u
rug
a
n P
2
1
Re
se
a
rc
h
a
n
d
De
v
e
lo
p
m
e
n
t
Ce
n
ter,
Bh
a
ra
th
iar
Un
iv
e
rsity
,
In
d
ia
2
De
p
a
rt
m
e
n
t
o
f
Co
m
p
u
ter S
c
ien
c
e
,
G
o
v
e
rn
m
e
n
t
A
rts
Co
ll
e
g
e
,
In
d
i
a
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
No
v
26
,
2
0
1
7
R
ev
i
s
ed
Feb
16
,
2
0
1
8
A
cc
ep
ted
Mar
30
,
2
0
1
8
A
H
y
p
e
rsp
e
c
tral
is
th
e
i
m
a
g
in
g
t
e
c
h
n
iq
u
e
th
a
t
c
o
n
tain
s
v
e
ry
lar
g
e
d
im
e
n
sio
n
d
a
ta
w
it
h
th
e
h
u
n
d
re
d
s
o
f
c
h
a
n
n
e
ls.
M
e
a
n
w
h
il
e
,
th
e
Hy
p
e
rsp
e
c
tr
a
l
Im
a
g
e
s
(HISs)
d
e
li
v
e
rs
th
e
c
o
m
p
l
e
te
k
n
o
w
led
g
e
o
f
i
m
a
g
in
g
;
th
e
re
f
o
re
a
p
p
ly
in
g
a
c
las
si
f
ica
ti
o
n
a
lg
o
rit
h
m
is
v
e
r
y
i
m
p
o
rtan
t
to
o
l
f
o
r
p
ra
c
ti
c
a
l
u
se
s.
T
h
e
HSIs
a
re
a
l
w
a
y
s
h
a
v
in
g
a
larg
e
n
u
m
b
e
r
o
f
c
o
rre
late
d
a
n
d
re
d
u
n
d
a
n
t
f
e
a
t
u
re
,
w
h
ich
c
a
u
se
s
th
e
d
e
c
re
m
e
n
t
in
th
e
c
las
sif
ica
ti
o
n
a
c
c
u
ra
c
y
;
m
o
r
e
o
v
e
r,
th
e
f
e
a
tu
re
s
re
d
u
n
d
a
n
c
y
c
o
m
e
u
p
w
it
h
so
m
e
e
x
tra
b
u
rd
e
n
o
f
c
o
m
p
u
tatio
n
th
a
t
w
it
h
o
u
t
a
d
d
in
g
a
n
y
b
e
n
e
f
icia
l
in
f
o
r
m
a
ti
o
n
t
o
t
h
e
c
las
sif
ic
a
ti
o
n
a
c
c
u
ra
c
y
.
In
th
is
stu
d
y
,
a
n
u
n
su
p
e
rv
ise
d
b
a
se
d
Ba
n
d
S
e
lec
ti
o
n
A
lg
o
rit
h
m
(BS
A
)
is
c
o
n
sid
e
re
d
w
it
h
th
e
L
in
e
a
r
P
r
o
jec
ti
o
n
(L
P
)
th
a
t
d
e
p
e
n
d
s
u
p
o
n
th
e
m
e
tri
c
-
b
a
n
d
sim
il
a
rit
ies
.
Af
ter
w
a
rd
s
M
o
n
o
g
e
n
e
ti
c
Bin
a
ry
F
e
a
tu
re
(M
BF
)
h
a
s
c
o
n
sid
e
r
t
o
p
e
rf
o
r
m
th
e
„tex
tu
re
a
n
a
l
y
sis‟
o
f
t
h
e
HSI,
w
h
e
re
th
re
e
o
p
e
ra
ti
o
n
a
l
c
o
m
p
o
n
e
n
t
re
p
re
se
n
ts
th
e
m
o
n
o
g
e
n
e
ti
c
sig
n
a
l
su
c
h
a
s;
p
h
a
se
,
a
m
p
li
tu
d
e
a
n
d
o
rien
tatio
n
.
In
p
o
st
p
ro
c
e
ss
in
g
c
las
sif
ica
ti
o
n
sta
g
e
,
f
e
a
tu
re
-
m
a
p
p
in
g
f
u
n
c
ti
o
n
c
a
n
p
ro
v
i
d
e
im
p
o
rtan
t
in
f
o
rm
a
ti
o
n
,
w
h
ich
h
e
lp
to
a
d
o
p
t
th
e
Ke
rn
e
l
b
a
se
d
Ne
u
r
a
l
Ne
tw
o
rk
(KN
N)
to
o
p
ti
m
ize
th
e
g
e
n
e
ra
li
z
a
ti
o
n
a
b
il
it
y
.
Ho
w
e
v
e
r,
a
n
a
lt
e
rn
a
ti
v
e
m
e
th
o
d
o
f
m
u
l
ti
c
las
s
a
p
p
li
c
a
ti
o
n
c
a
n
b
e
a
d
o
p
t
th
r
o
u
g
h
KN
N,
if
we
c
o
n
sid
e
r
th
e
m
u
lt
i
-
o
u
t
p
u
t
n
o
d
e
s i
n
ste
a
d
o
f
tak
in
g
sin
g
le
-
o
u
tp
u
t
n
o
d
e
.
K
ey
w
o
r
d
:
B
an
d
s
elec
tio
n
a
l
g
o
r
ith
m
(
B
SA
)
H
y
p
er
s
p
ec
t
r
al
i
m
a
g
e
(
HSI
)
L
i
n
ea
r
p
r
o
j
ec
tio
n
(
L
P
)
Mo
n
o
g
en
e
tic
b
in
ar
y
f
ea
t
u
r
e
(
MB
F)
Neu
r
al
n
et
w
o
r
k
(
NN)
Co
p
y
rig
h
t
©
2
0
1
8
In
stit
u
te o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
Al
l
rig
h
ts
re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
P
u
ttas
w
a
m
y
M
.
R
.
,
R
esear
ch
a
n
d
Dev
elo
p
m
en
t
C
en
ter
,
B
h
ar
ath
iar
Un
iv
er
s
it
y
,
C
o
i
m
b
ato
r
e,
T
am
il Na
d
u
6
4
1
0
4
6
,
I
n
d
ia
.
E
m
ail:
m
r
p
.
g
o
w
d
a
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
HSI
co
n
s
i
s
t
o
f
v
er
y
lar
g
e
d
i
m
e
n
s
io
n
d
ata
w
it
h
th
e
h
u
n
d
r
ed
s
o
f
ch
a
n
n
els
th
at
r
an
g
i
n
g
f
r
o
m
t
h
e
s
h
o
r
t
in
f
r
ar
ed
w
a
v
e
to
th
e
v
i
s
ib
le
r
eg
io
n
at
„
elec
tr
o
-
m
a
g
n
etic
(
E
M)
s
p
ec
tr
u
m
‟
[
1
]
.
A
h
y
p
er
s
p
ec
tr
al
is
th
e
i
m
a
g
in
g
tec
h
n
iq
u
e,
w
h
ic
h
a
cq
u
ir
es
t
h
e
o
b
j
ec
ts
in
f
o
r
m
at
io
n
b
ased
u
p
o
n
th
eir
E
M
-
s
p
ec
tr
u
m
w
i
th
th
e
w
a
v
ele
n
g
t
h
o
f
4
0
0
n
m
to
2
5
0
0
n
m
.
Me
a
n
w
h
ile,
t
h
e
HI
S
d
eli
v
er
s
th
e
co
m
p
lete
k
n
o
w
led
g
e
o
f
i
m
ag
in
g
,
t
h
u
s
t
h
e
s
ev
er
al
o
f
ap
p
licatio
n
li
k
e
a
s
;
m
ater
ia
l
id
en
tific
atio
n
,
tar
g
et
d
etec
tio
n
a
n
d
o
b
j
ec
t
d
is
co
v
er
in
g
h
as
r
ep
o
r
ted
in
[
2
]
.
I
n
f
o
r
m
atio
n
e
x
tr
ac
tio
n
is
v
er
y
s
i
g
n
i
f
ica
n
t
p
r
o
ce
s
s
i
n
HSI
;
t
h
er
ef
o
r
e
ap
p
l
y
i
n
g
a
cla
s
s
if
ica
tio
n
a
lg
o
r
it
h
m
is
v
er
y
i
m
p
o
r
tan
t to
o
l f
o
r
p
r
ac
tical
u
s
e
s
.
Ho
w
e
v
er
,
th
e
HSI
ar
e
al
w
a
y
s
h
av
i
n
g
a
lar
g
e
n
u
m
b
er
o
f
co
r
r
elate
d
an
d
r
ed
u
n
d
an
t
f
ea
t
u
r
e,
w
h
ic
h
ca
u
s
e
s
t
h
e
d
ec
r
e
m
e
n
t
i
n
th
e
class
i
f
icatio
n
ac
c
u
r
ac
y
[
3
]
;
m
o
r
eo
v
er
,
th
e
f
ea
tu
r
e
s
r
ed
u
n
d
an
c
y
co
m
e
u
p
w
i
th
s
o
m
e
e
x
tr
a
b
u
r
d
en
o
f
co
m
p
u
tatio
n
th
a
t
w
it
h
o
u
t
ad
d
in
g
an
y
b
e
n
ef
icial
i
n
f
o
r
m
atio
n
t
o
th
e
clas
s
i
f
icatio
n
ac
cu
r
ac
y
.
He
n
ce
,
HSI
d
ata
p
r
o
ce
s
s
in
g
(
t
h
at
co
n
tai
n
h
i
g
h
v
o
lu
m
e
d
ata)
b
ec
o
m
e
s
o
m
e
w
h
at
d
if
f
ic
u
lt,
e
s
p
ec
iall
y
w
it
h
t
h
e
s
u
p
er
v
is
ed
lear
n
in
g
m
et
h
o
d
b
ec
au
s
e
th
e
ac
c
u
r
ac
y
o
f
class
i
f
icatio
n
d
ec
r
ea
s
es
w
it
h
s
p
ec
i
f
ic
s
et
o
f
tr
ain
i
n
g
as
i
n
cr
ea
s
i
n
g
f
ea
t
u
r
es
n
u
m
b
er
,
th
i
s
ca
lled
as
„
Hu
g
h
e
s
Occ
u
r
r
en
ce
‟
.
I
n
o
r
d
er
to
ac
h
iev
e
h
i
g
h
er
class
i
f
icatio
n
ac
c
u
r
ac
y
,
d
i
m
en
s
io
n
alit
y
r
ed
u
c
tio
n
ap
p
r
o
ac
h
b
ec
o
m
i
n
g
v
er
y
b
e
n
ef
icial
a
n
d
it
also
r
ed
u
ce
s
t
h
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
8
,
No
.
4
,
A
u
g
u
s
t 2
0
1
8
:
2
1
1
5
–
2
1
2
5
2116
r
eq
u
ir
e
m
en
t
o
f
d
ata
s
to
r
ag
e
a
n
d
co
m
p
u
tatio
n
al
ti
m
e
[
4
]
.
I
n
[
5
]
,
th
e
y
co
n
clu
d
ed
t
h
at
r
ed
u
cin
g
i
n
n
u
m
b
er
o
f
f
ea
t
u
r
es ten
d
s
to
ac
h
iev
e
b
ette
r
class
if
icatio
n
ac
c
u
r
ac
y
.
I
n
d
i
m
en
s
io
n
al
it
y
r
ed
u
ctio
n
,
th
e
m
o
s
t
co
m
m
o
n
l
y
ap
p
lied
tech
n
iq
u
e
is
s
u
b
j
ec
ted
to
th
is
p
ap
er
is
f
ea
t
u
r
e
s
elec
tio
n
(
FS
)
,
w
h
ich
ch
o
o
s
es
a
n
i
m
p
o
r
ta
n
t
s
u
b
-
s
et
f
ea
t
u
r
e
f
r
o
m
th
e
m
ai
n
f
ea
t
u
r
e
s
et
an
d
d
is
p
o
s
e
t
h
e
r
e
m
ain
in
g
f
ea
tu
r
e
s
.
A
n
o
t
h
er
ap
p
r
o
ac
h
is
f
ea
t
u
r
e
ex
tr
ac
tio
n
(
FE)
,
w
h
ic
h
g
en
er
all
y
e
x
tr
ac
ts
th
e
i
m
p
o
r
tan
t
p
r
o
p
er
ties
f
r
o
m
a
f
ea
t
u
r
e
s
et
an
d
th
e
n
tr
a
n
s
f
o
r
m
s
t
h
e
„
m
ain
d
ata‟
to
cr
ea
te
t
h
at
m
o
r
e
s
ep
ar
ab
le.
C
o
n
s
id
er
in
g
b
o
th
th
e
f
ea
t
u
r
e
ap
p
r
o
ac
h
,
a
f
ea
t
u
r
e
s
elec
tio
n
i
s
m
o
s
t
s
u
it
ab
le
f
o
r
th
e
„
r
ea
l
d
ataset‟
w
i
t
h
r
esp
ec
t
to
f
ea
tu
r
e
ex
tr
ac
tio
n
.
Ho
w
ev
er
,
th
e
HSI
p
r
o
v
id
es
a
co
m
p
r
eh
en
s
i
v
e
s
ep
ar
atio
n
o
f
an
alo
g
o
u
s
s
u
r
f
ac
e
m
ater
ials
,
t
h
is
k
in
d
o
f
s
p
ec
tr
al
f
ea
t
u
r
e
ar
e
s
p
e
ciall
y
co
r
r
elate
w
i
t
h
t
h
e
a
d
j
ac
en
t
b
an
d
s
,
h
e
n
ce
it
p
r
o
v
id
e
a
r
ed
u
n
d
an
t
in
f
o
r
m
atio
n
[
6
]
.
A
Hu
g
h
es
ef
f
ec
t
[
7
]
,
is
r
elate
d
to
lo
w
g
e
n
er
aliza
tio
n
ab
ilit
y
o
f
th
e
„
cla
s
s
i
f
ier
‟
,
w
h
ic
h
f
r
eq
u
en
tl
y
h
a
s
en
co
u
n
ter
ed
in
s
e
v
er
al
„
p
atter
n
r
ec
o
g
n
itio
n
‟
ap
p
licatio
n
s
s
u
c
h
a
s
;
o
b
j
e
ct
r
ec
o
g
n
iti
o
n
,
tex
t
ca
teg
o
r
izatio
n
,
co
m
p
u
ter
v
i
s
io
n
a
n
d
g
e
n
e
e
x
p
r
ess
io
n
d
ata
[
8
]
,
[
9
]
.
T
h
er
e
ar
e
s
ev
er
al
f
ea
t
u
r
e
tech
n
iq
u
e
s
s
u
c
h
as
„
MN
F‟
(
m
i
n
i
m
u
m
-
n
o
i
s
e
f
r
ac
tio
n
[
1
0
]
)
,
„
P
C
A‟
(
p
r
in
c
ip
al
co
m
p
o
n
e
n
t
a
n
al
y
s
i
s
[
1
1
]
)
,
„
SP
P
‟
(
s
p
ar
s
it
y
p
r
eser
v
in
g
p
r
o
j
ec
tio
n
[
1
2
]
)
,
„
L
P
P
‟
(
lo
c
al
p
r
eser
v
i
n
g
p
r
o
j
e
ctio
n
[
1
3
]
)
,
„
MSM
E
‟
(
m
u
lti
-
s
tr
u
ct
u
r
e
m
an
if
o
ld
e
m
b
ed
d
in
g
[
1
4
]
)
,
an
d
„
SP
A‟
(
s
p
ar
s
it
y
p
r
eser
v
i
n
g
an
al
y
s
is
[
1
5
]
)
,
etc.
T
h
o
u
g
h
th
e
s
o
m
e
i
m
p
o
r
tan
t
in
f
o
r
m
atio
n
at
f
ea
t
u
r
e
ap
p
r
o
ac
h
h
a
s
n
o
t
o
b
tain
ed
p
r
o
p
er
ly
,
th
er
e
f
o
r
e
it
ca
u
s
es
t
h
e
p
er
f
o
r
m
an
ce
d
eg
r
ad
atio
n
i
n
H
SI
class
i
f
icatio
n
.
T
h
er
ef
o
r
e,
it
i
s
i
m
p
er
io
u
s
to
d
e
v
elo
p
an
e
f
f
i
cien
t
a
n
d
n
e
w
f
ea
t
u
r
e
s
elec
tio
n
tec
h
n
iq
u
e,
w
h
ic
h
in
te
g
r
ates t
h
e
s
p
ec
tr
al
b
an
d
s
el
ec
tio
n
an
d
clas
s
i
f
icatio
n
to
r
e
m
o
tel
y
s
en
s
e
th
e
H
SI.
T
h
e
d
im
e
n
s
io
n
alit
y
ca
n
b
e
ac
h
iev
e
th
r
o
u
g
h
„
b
an
d
s
elec
tio
n
ap
p
r
o
ac
h
‟
th
a
t
is
B
S
A;
t
h
er
e
ar
e
t
w
o
t
y
p
es
o
f
B
S
A
(
i.e
.
,
s
u
p
er
v
is
ed
an
d
u
n
s
u
p
er
v
i
s
ed
B
SA
)
.
Her
e,
w
e
ar
e
p
er
f
o
r
m
i
n
g
u
n
s
u
p
er
v
i
s
ed
B
SA
,
w
h
ic
h
is
u
s
ed
to
g
et
th
e
s
u
p
er
lativ
e
i
n
f
o
r
m
at
io
n
b
an
d
s
w
it
h
o
u
t
h
a
v
i
n
g
an
y
i
n
f
o
r
m
atio
n
o
f
o
b
j
ec
tiv
es.
I
n
th
i
s
s
tu
d
y
,
a
n
u
n
s
u
p
er
v
is
ed
b
ased
B
SA
is
co
n
s
id
er
ed
w
it
h
th
e
L
P
th
at
d
ep
en
d
s
u
p
o
n
th
e
m
e
tr
ic
-
b
an
d
s
i
m
ilar
itie
s
.
Af
ter
w
ar
d
s
MB
F
h
a
s
co
n
s
id
er
to
p
er
f
o
r
m
th
e
„
te
x
t
u
r
e
an
al
y
s
is
‟
o
f
th
e
H
SI,
w
h
er
e
th
r
ee
o
p
er
atio
n
al
co
m
p
o
n
e
n
t
r
ep
r
ese
n
ts
th
e
m
o
n
o
g
e
n
etic
s
ig
n
al
s
u
c
h
a
s
;
p
h
a
s
e,
a
m
p
lit
u
d
e
a
n
d
o
r
ien
tatio
n
.
I
n
p
o
s
t
p
r
o
ce
s
s
i
n
g
class
i
f
icatio
n
s
tag
e,
Neu
r
al
Net
w
o
r
k
(
NN)
p
er
f
o
r
m
s
th
e
class
i
f
icatio
n
p
r
o
ce
s
s
a
f
ter
th
e
f
ea
t
u
r
e
e
x
tr
ac
tio
n
f
r
o
m
MB
F.
T
h
e
NN
co
m
p
r
is
es
o
n
e
h
id
d
en
la
y
er
a
n
d
o
n
e
o
u
tp
u
t
la
y
er
;
th
e
w
ei
g
h
t
a
s
s
ig
n
m
en
t
is
p
er
f
o
r
m
r
an
d
o
m
l
y
i
n
b
et
w
ee
n
„
I
n
p
u
t
o
f
NN‟
a
n
d
h
id
d
en
la
y
er
.
F
ea
tu
r
e
m
ap
p
in
g
f
u
n
ctio
n
ca
n
p
r
o
v
id
e
im
p
o
r
tan
t
in
f
o
r
m
atio
n
,
w
h
ic
h
h
elp
to
ad
o
p
t
th
e
k
er
n
el
b
ased
NN
(
KNN)
to
o
p
tim
ize
t
h
e
g
e
n
er
ali
za
tio
n
ab
ilit
y
[
1
6
]
.
Ho
w
e
v
er
,
an
alter
n
ati
v
e
m
eth
o
d
o
f
m
u
lticlas
s
ap
p
licatio
n
c
an
b
e
p
r
o
v
id
e
th
r
o
u
g
h
KNN,
if
w
e
co
n
s
id
er
th
e
m
u
lti
-
o
u
tp
u
t
n
o
d
es
i
n
s
tead
o
f
tak
i
n
g
s
in
g
le
-
o
u
tp
u
t
n
o
d
e.
T
h
e
t
w
o
HSI
d
ata
s
et
n
a
m
el
y
as
Salin
a
s
-
s
ce
n
e
[
1
7
]
an
d
P
av
ia
U
n
i
v
er
s
it
y
[
1
7
]
,
w
e
ar
e
g
o
in
g
to
u
s
e
t
h
r
o
u
g
h
o
u
t t
h
is
p
ap
er
to
v
a
lid
ate
o
u
r
c
lass
if
i
ca
tio
n
r
es
u
lt
s
w
i
t
h
r
esp
ec
t
o
th
er
s
tate
-
of
-
ar
t
m
et
h
o
d
s
.
T
h
is
p
ap
er
is
o
r
g
an
ized
a
s
f
o
llo
w
s
;
Sect
io
n
2
p
r
o
v
id
e
t
h
e
d
etailed
s
u
r
v
e
y
o
f
s
tate
-
of
-
t
h
e
-
ar
t,
Sectio
n
3
g
iv
es
t
h
e
p
r
o
p
o
s
ed
m
e
th
o
d
o
lo
g
ies
k
n
o
w
led
g
e,
Sect
i
o
n
4
p
r
o
v
id
es
th
e
ex
p
er
i
m
e
n
tal
r
es
u
lt a
n
d
a
n
al
y
s
is
o
f
cl
as
s
i
f
icatio
n
.
F
in
al
l
y
,
Se
ctio
n
5
co
n
clu
d
es t
h
is
p
ap
er
.
2.
L
I
T
RA
T
UR
E
SURV
E
Y
An
al
y
s
i
s
o
f
H
SIs
ta
s
k
i
s
d
if
f
icu
lt
b
ec
au
s
e
th
e
d
ata
s
e
ts
w
h
ich
ar
e
h
av
in
g
e
x
tr
e
m
el
y
lar
g
e
d
i
m
en
s
io
n
al
it
y
,
s
o
ch
o
o
s
e
s
o
m
e
es
s
e
n
tial
f
ea
t
u
r
es
it
is
m
o
s
t
s
i
g
n
i
f
ican
t,
an
d
h
elp
f
u
l
f
o
r
lear
n
in
g
.
I
n
o
r
d
er
to
s
elec
t
es
s
e
n
tial
f
ea
t
u
r
es
o
f
h
et
er
o
g
en
eo
u
s
,
in
[
1
8
]
th
e
y
h
a
s
p
r
o
p
o
s
ed
a
m
et
h
o
d
ca
lled
s
p
ar
s
e
f
ea
t
u
r
e
s
elec
tio
n
m
et
h
o
d
w
h
ich
i
s
b
u
ilt
o
n
r
eg
u
lar
ized
r
eg
r
es
s
io
n
m
o
d
el.
Ho
w
ev
er
,
in
t
h
i
s
p
r
esen
tati
o
n
w
i
th
t
h
e
n
o
is
e
in
f
o
r
m
atio
n
ar
e
co
n
ta
m
i
n
ated
b
ec
au
s
e
o
f
i
m
a
g
i
n
g
d
ev
ices.
T
h
at
n
o
is
e
w
ill
a
f
f
ec
ts
t
h
e
p
r
o
ce
s
s
o
f
lear
n
i
n
g
f
o
r
ex
a
m
p
le
h
y
p
er
s
p
ec
tr
al
i
m
a
g
es
o
f
h
i
g
h
d
i
m
e
n
s
io
n
al
d
ata
a
n
al
y
s
i
s
.
R
ed
u
n
d
an
c
y
r
ed
u
ci
n
g
as
w
ell
as
p
r
eser
v
i
n
g
d
ata
th
ese
t
w
o
cr
itical
p
r
o
b
lem
s
,
w
h
ic
h
ar
e
n
ec
es
s
ar
i
l
y
to
b
e
h
an
d
led
:
at
th
e
f
ea
t
u
r
e
s
ele
ctio
n
.
I
n
[
1
9
]
,
b
ased
o
n
a
r
ec
e
n
tl
y
d
esi
g
n
ed
m
e
m
etic
p
r
o
ce
d
u
r
e
th
e
y
p
r
o
p
o
s
ed
a
f
ea
tu
r
e
s
elec
tio
n
tec
h
n
iq
u
e
f
o
r
h
y
p
er
s
p
ec
tr
al
i
m
a
g
e
class
if
ica
tio
n
.
T
h
e
y
d
esig
n
ed
a
s
u
i
tab
le
o
b
j
ec
tiv
e
ta
s
k
i
n
s
id
e
t
h
e
p
r
o
p
o
s
ed
tech
n
iq
u
e,
t
h
at
ab
le
b
e
m
ea
s
u
r
e
t
h
e
e
n
clo
s
ed
r
ed
u
n
d
a
n
c
y
in
f
o
as
w
ell
a
s
es
s
en
tial
i
n
f
o
o
n
t
h
e
s
elec
ted
f
ea
t
u
r
e
s
u
b
s
ets.
I
n
[
2
0
]
,
th
e
y
s
elec
ted
n
o
r
ed
u
n
d
an
t
i
n
f
o
r
m
a
tio
n
p
lu
s
Gab
o
r
f
ea
tu
r
e
s
f
o
r
h
y
p
er
s
p
ec
tr
al
i
m
a
g
e
class
i
f
icati
o
n
an
d
p
r
o
p
o
s
ed
a
Ma
r
k
o
v
b
lan
k
et
b
ased
s
y
m
m
e
tr
ical
u
n
ce
r
tai
n
t
y
ap
p
r
o
ac
h
.
A
f
as
t
f
o
r
w
ar
d
(
FS
)
ap
p
r
o
ac
h
[
2
1
]
,
w
h
ic
h
b
ased
u
p
o
n
Gau
s
s
ian
M
ix
t
u
r
e
(
GM
)
p
r
o
to
t
y
p
e
class
i
f
ier
.
T
h
e
GM
-
clas
s
i
f
ier
h
a
s
u
s
ed
f
o
r
class
i
f
y
i
n
g
th
e
h
y
p
er
s
p
ec
tr
al
i
m
a
g
es,
t
h
i
s
ap
p
r
o
ac
h
ch
o
o
s
e
th
e
s
p
ec
tr
al
f
ea
t
u
r
e
to
in
cr
ea
s
e
th
e
p
r
e
d
ictio
n
r
ate
o
f
cla
s
s
i
f
icat
io
n
.
T
h
is
ca
n
e
x
ec
u
te
th
r
o
u
g
h
„
k
-
f
o
ld
‟
cr
o
s
s
v
alid
atio
n
i
n
o
r
d
er
to
ac
h
iev
e
ef
f
icie
n
t
i
m
p
le
m
en
ta
tio
n
a
n
d
f
a
s
t
co
m
p
u
tin
g
ti
m
e.
I
n
i
tiall
y
,
th
e
GM
ca
n
u
p
g
r
ad
ed
w
it
h
co
m
p
u
ted
class
i
f
icat
io
n
r
ate,
in
s
tead
o
f
r
e
-
e
s
ti
m
ate
o
v
er
all
m
o
d
el.
Fi
n
all
y
,
GM
m
ar
g
in
a
li
za
tio
n
ca
n
b
e
s
u
b
-
m
o
d
el
f
r
o
m
th
e
f
u
ll
lear
n
in
g
m
o
d
el
o
f
s
p
ec
tr
al
f
ea
t
u
r
es.
T
h
e
in
v
es
tig
a
tio
n
o
f
th
e
u
n
s
u
p
er
v
is
ed
-
B
S
A
[
2
2
]
f
o
r
th
e
s
p
ec
tr
al
FS
a
n
d
th
e
cla
s
s
i
f
icatio
n
p
r
o
ce
s
s
b
ased
u
p
o
n
S
VM
(
Su
p
p
o
r
t
Vec
to
r
Ma
ch
in
e)
[
2
3
]
,
ar
e
m
o
s
tl
y
u
s
ed
in
m
o
r
p
h
o
lo
g
ical
p
r
o
f
il
es
(
„
MP
‟
)
.
T
h
e
d
if
f
icu
lt
y
at
h
ig
h
d
i
m
e
n
s
io
n
ali
t
y
ca
n
m
i
n
i
m
ize
b
y
r
es
u
lti
n
g
f
ea
t
u
r
es a
n
d
it is
m
o
s
t
u
s
u
al
ac
tiv
it
y
w
h
er
e
MP
h
as e
x
tr
ac
ted
f
r
o
m
P
C
A
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
l
ec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
A
N
eu
r
a
l Netw
o
r
k
A
p
p
r
o
a
ch
t
o
I
d
en
tify H
yp
ers
p
ec
tr
a
l I
ma
g
e
C
o
n
ten
t (
P
u
tta
s
w
a
my
M
.
R
.
)
2117
T
h
e
m
ai
n
ch
alle
n
g
e
o
f
clas
s
i
f
icatio
n
p
r
o
b
lem
s
i
s
s
elec
tio
n
o
f
a
s
u
b
s
et,
w
h
ic
h
h
as
e
f
f
icien
t
an
d
m
i
n
i
m
u
m
f
r
o
m
th
e
e
n
o
r
m
o
u
s
a
m
o
u
n
t
o
f
f
ea
tu
r
es.
T
h
e
en
ti
r
e
f
ea
tu
r
e
s
elec
tio
n
m
e
th
o
d
s
co
n
s
is
t
o
f
a
s
ea
r
ch
alg
o
r
ith
m
u
s
ed
f
o
r
s
elec
tio
n
o
f
h
ig
h
t
i
m
e
co
n
s
u
m
i
n
g
f
i
n
est
ca
n
d
id
ate
f
r
o
m
t
h
e
p
o
s
s
ib
le
an
s
w
er
s
.
I
n
[
2
4
]
,
p
r
o
p
o
s
ed
d
u
al
f
ea
t
u
r
e
s
elec
tio
n
ap
p
r
o
ac
h
es
w
it
h
o
u
t
r
eq
u
ir
e
m
en
t
o
f
s
ea
r
c
h
al
g
o
r
ith
m
.
W
it
h
u
s
i
n
g
m
ea
n
v
al
u
e
s
p
lu
s
s
i
m
p
le
ca
lc
u
lat
io
n
o
f
s
ta
n
d
ar
d
d
ev
iatio
n
e
f
f
ec
tiv
e
f
ea
t
u
r
es
o
f
s
u
b
s
et
ar
e
s
elec
ted
f
r
o
m
th
e
m
et
h
o
d
.
I
n
[
2
5
]
,
[
2
6
]
,
th
e
y
e
x
p
r
ess
ed
t
w
o
class
cla
s
s
i
f
icatio
n
is
s
u
e,
w
h
e
r
e
m
o
d
if
ied
(
KNN
)
k
n
ea
r
es
t
n
eig
h
b
o
r
s
clas
s
i
f
ier
m
et
h
o
d
h
as
b
ee
n
ap
p
lied
to
cr
ea
te
v
alid
it
y
r
ati
n
g
,
m
a
x
i
m
al
co
h
er
en
ce
,
p
lu
s
ca
te
g
o
r
ize
t
h
e
test
s
a
m
p
le
s
b
y
k
-
f
o
r
d
v
alid
atio
n
.
I
n
[
2
7
]
,
th
e
y
e
m
p
h
asi
s
o
n
t
h
e
HI
S
class
i
f
icatio
n
d
ata,
w
h
ich
ca
p
tu
r
ed
f
r
o
m
t
h
e
E
n
v
ir
o
n
m
e
n
tal
P
r
o
g
r
a
m
(
E
n
P
)
.
Mo
r
eo
v
er
,
th
ey
co
n
s
id
er
ed
th
e
d
ataset
th
at
p
r
ese
n
ted
with
E
n
P
co
n
tes
t,
a
s
tan
d
ar
d
h
as
r
ec
en
t
l
y
i
n
itial
i
ze
d
w
it
h
s
ev
er
al
cla
s
s
i
f
y
in
g
o
b
j
ec
t
an
d
th
e
lan
d
u
s
e
cla
s
s
es
d
ep
en
d
s
u
p
o
n
h
y
p
er
s
p
ec
tr
al
d
ata.
I
n
t
h
i
s
[
2
8
]
,
th
e
y
h
a
s
co
n
s
id
er
ed
„
R
ad
i
al
B
asis
F
u
n
ctio
n
‟
(
R
B
F),
w
h
er
e
f
u
l
l
b
an
d
w
id
th
-
R
B
F
k
er
n
el
h
as
u
s
ed
as
f
ea
t
u
r
e
w
eig
h
t
s
,
w
h
e
n
ev
er
t
h
e
f
ea
t
u
r
e
v
al
u
es
b
ec
o
m
e
r
escaled
th
r
o
u
g
h
t
h
e
z
-
s
co
r
e
s
v
alu
e.
T
h
e
m
et
h
o
d
o
lo
g
y
o
f
HI
S
al
lo
w
s
r
ec
o
g
n
izi
n
g
m
a
ter
ials
t
h
r
o
u
g
h
u
s
i
n
g
p
h
o
to
-
t
h
er
m
a
l
„
in
f
r
ar
ed
s
p
ec
tr
o
s
co
p
y
‟
[
29
]
-
[
3
0
]
.
I
n
o
r
d
er
to
o
b
tain
th
e
in
f
r
ar
ed
s
p
ec
t
r
o
s
co
p
y
,
a
n
i
n
f
r
ar
ed
ca
m
er
a
c
ap
tu
r
e
th
e
h
u
n
d
r
ed
s
o
f
i
m
a
g
es
o
f
a
n
o
b
j
ec
t
at
d
i
f
f
er
en
t
ch
a
n
n
els
w
a
v
elen
g
t
h
s
i
m
u
ltan
eo
u
s
l
y
w
h
ile
a
„
QC
L
‟
(
q
u
an
t
u
m
ca
s
ca
d
e
laser
)
ca
u
s
e
s
t
h
e
s
u
b
j
ec
ted
m
ater
ial
to
b
e
b
r
ig
h
ten
ed
.
T
h
e
m
ai
n
p
r
o
b
le
m
s
t
h
at
e
f
f
ec
tin
g
t
h
e
cla
s
s
i
f
icatio
n
ar
r
iv
es
f
r
o
m
t
h
e
d
is
p
ar
ag
i
n
g
r
atio
o
f
h
ig
h
d
i
m
en
s
io
n
HI
S
a
n
d
th
e
s
m
all
s
ize
„
tr
ai
n
i
n
g
d
at
a‟
[
3
1
]
.
T
h
ese
k
in
d
o
f
p
r
o
b
lem
s
ca
u
s
e
s
ad
v
an
ce
m
en
t
in
m
ac
h
in
e
lear
n
in
g
a
p
p
r
o
ac
h
es
s
u
ch
a
s
;
m
u
ltip
le
lear
n
in
g
s
y
s
te
m
s
,
tr
an
s
d
u
ct
iv
e
lear
n
i
n
g
,
s
e
m
i
-
s
u
p
er
v
is
ed
l
ea
r
n
i
n
g
a
n
d
ac
tiv
e
l
ea
r
n
in
g
[
3
2
]
-
[
3
4
]
an
d
s
e
v
er
al
m
et
h
o
d
s
o
f
lear
n
i
n
g
d
ep
en
d
u
p
o
n
t
h
e
u
n
lab
eled
s
a
m
p
les
[
3
5
]
,
[
3
6
]
.
T
o
o
v
er
co
m
e
f
r
o
m
s
u
ch
k
i
n
d
o
f
is
s
u
e
s
,
s
o
m
e
s
tr
ate
g
y
h
a
s
co
n
s
id
er
ed
en
h
a
n
ci
n
g
n
e
w
s
i
m
ilar
it
y
m
ea
s
u
r
e
m
en
t
f
o
r
t
h
e
b
an
d
r
ed
u
ctio
n
an
d
d
i
m
e
n
s
io
n
r
ed
u
ctio
n
o
f
HI
S
[
3
7
]
,
[
3
8
]
.
Ho
w
e
v
er
,
t
h
e
h
ig
h
co
r
r
elatio
n
b
et
w
ee
n
th
e
b
an
d
s
h
a
s
m
is
id
en
ti
f
y
f
o
r
d
esig
n
i
n
g
n
e
w
tech
n
iq
u
es
to
r
ed
u
ce
th
e
d
ata
d
i
m
en
s
io
n
al
it
y
,
w
h
ic
h
i
n
cl
u
d
in
g
s
o
m
e
o
f
t
h
e
m
eth
o
d
s
t
h
at
h
as
g
r
ea
t
ac
ce
p
t
a
n
ce
s
u
c
h
as;
P
C
A
[
3
9
]
,
„
lin
ea
r
d
is
cr
i
m
i
n
an
t
a
n
al
y
s
is
‟
(
L
D
A
)
[
4
0
]
o
r
MN
F
[
4
1
]
.
I
t
also
h
as
b
ee
n
in
v
esti
g
ated
in
p
r
ec
ed
in
g
w
o
r
k
s
th
at
HI
S
cla
s
s
i
f
icatio
n
af
ter
th
e
p
r
ep
r
o
ce
s
s
in
g
p
r
o
ce
s
s
d
i
m
en
s
io
n
ali
t
y
r
ed
u
ctio
n
„
o
r
‟
b
an
d
s
elec
tio
n
ar
e
u
s
u
al
l
y
o
u
tp
er
f
o
r
m
s
b
et
ter
clas
s
i
f
icatio
n
w
i
th
r
esp
ec
t
to
tr
ad
itio
n
al
HSI
d
ata
class
i
f
icatio
n
[
4
2
]
,
[
4
3
]
,
w
h
ich
i
n
o
r
d
er
to
r
e
d
u
ce
th
e
co
m
p
u
tatio
n
al
co
m
p
lex
i
t
y
.
3.
P
RO
P
O
SE
D
H
YP
E
R
SPEC
T
RA
L
CL
A
SS
I
F
I
CA
T
I
O
N
AL
G
O
RI
T
H
M
T
h
e
p
r
esen
c
e
o
f
h
ig
h
s
p
ec
t
r
al
co
r
r
el
ati
o
n
in
th
e
HS
I
ca
u
s
es
d
if
f
icu
l
ty
in
f
ea
tu
r
e
s
el
ec
ti
o
n
p
r
o
ce
s
s
th
a
t
is
w
h
y
it‟
s
n
ec
ess
ar
y
t
o
p
e
r
f
o
r
m
d
im
en
s
io
n
al
r
ed
u
cti
o
n
a
t
s
p
ec
t
r
al
f
ea
tu
r
e
in
o
r
d
e
r
t
o
o
b
t
ai
n
a
u
s
ef
u
l
„
c
o
m
p
ac
t
s
et‟
s
p
ec
tr
al
f
e
atu
r
e.
T
h
e
d
im
en
s
io
n
ali
ty
ca
n
b
e
ac
h
iev
e
th
r
o
u
g
h
„
b
an
d
s
ele
cti
o
n
ap
p
r
o
ac
h
‟
th
at
is
B
SA
;
th
e
r
e
ar
e
tw
o
t
y
p
es
o
f
B
SA
(
i.
e.
,
s
u
p
e
r
v
is
e
d
an
d
u
n
s
u
p
e
r
v
is
e
d
B
S
A
)
.
He
r
e
,
w
e
ar
e
p
e
r
f
o
r
m
in
g
u
n
s
u
p
er
v
is
e
d
B
SA
,
w
h
ich
is
u
s
ed
to
g
et
th
e
s
u
p
e
r
lativ
e
in
f
o
r
m
atio
n
b
an
d
s
w
ith
o
u
t
h
av
in
g
an
y
in
f
o
r
m
atio
n
o
f
o
b
jec
tiv
es
.
I
n
th
is
s
tu
d
y
,
an
u
n
s
u
p
er
v
is
e
d
b
ase
d
B
SA
is
co
n
s
i
d
e
r
e
d
w
ith
th
e
li
n
ea
r
p
r
o
ject
io
n
th
a
t
d
e
p
en
d
s
u
p
o
n
th
e
m
etr
i
c
-
b
a
n
d
s
im
ilar
iti
es.
T
h
is
m
eth
o
d
o
l
o
g
y
d
o
es
n
o
t
r
e
q
u
ir
e
an
y
in
f
o
r
m
ati
o
n
r
eg
a
r
d
in
g
p
r
i
o
r
-
class
t
o
ca
t
eg
o
r
iz
e
th
e
l
ev
el
o
f
d
at
a
an
a
ly
s
is
.
H
e
r
e
,
w
e
s
ele
ct
t
h
e
tw
o
d
if
f
e
r
en
t
b
an
d
s
(
i.
e.
,
)
f
r
o
m
L
P
-
B
SA
[
4
4
]
an
d
th
e
s
u
b
s
et
o
f
b
an
d
s
is
d
ef
in
e
d
as;
*
+
(
1
)
T
h
e
n
ew
s
ele
cte
d
s
u
b
s
et
o
f
b
a
n
d
is
(
)
(
)
(
2
)
w
h
er
e,
is
a
th
i
r
d
b
an
d
an
d
a
b
o
v
e
E
q
u
at
io
n
(
2
)
h
as
r
e
p
e
at
ed
u
n
til
th
i
r
d
b
an
d
is
m
o
r
e
th
an
t
h
e
s
u
b
s
et
o
f
f
i
r
s
t
t
w
o
b
an
d
s
,
w
h
en
th
is
cr
it
e
r
i
a
h
as sa
tis
f
y
th
en
o
n
ly
th
e
E
q
u
at
i
o
n
(
2
)
is
u
p
d
at
e
d
.
(
)
(
)
(
3
)
Her
e,
d
en
o
t
es
th
e
p
r
e
d
i
cti
o
n
b
a
n
d
o
f
L
P
b
y
s
el
ec
tin
g
f
i
r
s
t
tw
o
b
an
d
s
(
i.
e.
,
)
an
d
th
e
e
r
r
o
r
in
L
P
ca
n
b
e
m
in
i
m
alize
th
r
o
u
g
h
s
ele
ctin
g
p
ar
am
eter
s
,
‖
‖
(
4
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
8
,
No
.
4
,
A
u
g
u
s
t 2
0
1
8
:
2
1
1
5
–
2
1
2
5
2118
I
n
o
r
d
er
t
o
ca
lcu
la
te
th
e
m
in
i
m
al
s
q
u
a
r
e
s
o
lu
ti
o
n
,
(
)
(
5
)
Fu
r
th
e
r
m
o
r
e
,
(
(
)
⁄
)
(
6
)
W
h
e
r
e
,
d
en
o
tes
th
e
„
in
p
u
t
d
at
a
‟
in
m
atr
ix
f
o
r
m
at
(
i
.
e
.
,
m
atr
ix
f
o
r
m
)
.
He
r
e
,
f
i
r
s
t
co
lu
m
n
co
n
tain
s
in
g
le
v
alu
e,
s
ec
o
n
d
co
lu
m
n
co
n
ta
in
s
a
ll
p
ix
els
v
alu
es
o
f
b
an
d
an
d
,
th
ir
d
c
o
lu
m
n
c
o
n
tain
s
all
p
ix
els
v
alu
es
o
f
b
an
d
.
A
d
d
iti
o
n
a
lly
,
th
e
r
e
p
r
es
en
ts
a
v
e
ct
o
r
m
atr
ix
th
at
in
c
lu
d
e
th
e
al
l
p
ix
el
v
alu
es
o
f
,
th
e
b
an
d
s
an
d
ar
e
c
o
m
p
ar
ed
w
ith
b
a
n
d
,
if
th
e
v
alu
e
o
f
th
i
r
d
b
an
d
f
o
u
n
d
m
o
r
e
th
an
th
e
s
u
b
s
et
o
f
b
an
d
(
)
is
u
p
d
ate
d
.
T
h
e
M
B
F
[
4
5
]
h
as
c
o
n
s
id
er
e
d
p
e
r
f
o
r
m
in
g
th
e
„
tex
tu
r
e
an
aly
s
es
o
f
th
e
HS
I
,
w
h
er
e
th
r
e
e
o
p
e
r
a
ti
o
n
al
co
m
p
o
n
en
ts
r
ep
r
es
en
ts
th
e
m
o
n
o
g
en
eti
c
s
ig
n
al
s
u
ch
as;
p
h
ase,
am
p
litu
d
e
an
d
o
r
ien
t
ati
o
n
.
T
h
e
o
p
e
r
a
to
r
s
o
f
MB
F
ca
n
b
e
g
iv
en
as „
M
B
F
-
P‟
(
M
B
F
-
P
h
ase‟
)
,
„
M
B
F
-
A
‟
(
M
B
F
-
Am
p
litu
d
e
)
an
d
„
MBF
-
O‟
(
M
B
F
-
Or
ien
t
ati
o
n
)
.
T
h
is
c
an
b
e
c
o
m
p
u
ted
th
r
o
u
g
h
[
4
5
]
,
w
h
er
e
p
a
r
am
ete
r
s
o
f
l
o
c
al
v
a
r
i
ab
le
an
d
th
e
im
ag
e
in
te
n
s
ity
ar
e
in
teg
r
at
e
d
to
p
r
o
v
i
d
e
in
d
iv
id
u
a
l
o
p
e
r
a
to
r
o
f
m
o
n
o
g
en
et
ic
s
ig
n
al
.
-
(
)
[
(
)
(
)
(
)
]
(
7
)
-
(
)
[
(
)
(
)
(
)
]
(
8
)
-
(
)
[
(
)
(
)
(
)
]
(
9
)
T
h
e
M
B
F
-
X
r
ep
r
es
en
ts
th
e
p
atte
r
n
n
u
m
b
er
o
f
M
B
F
(
-
*
+
)
,
h
er
e,
is
p
ix
el
o
f
m
o
n
o
g
en
e
tic
s
ig
n
al
th
at
ca
n
b
e
en
c
o
d
e
d
w
i
th
2
-
b
its
(
(
)
(
)
)
.
E
ac
h
o
f
th
e
M
B
F
o
p
e
r
a
to
r
f
e
atu
r
e
c
an
b
e
u
s
e
d
as
in
d
iv
i
d
u
ally
f
o
r
th
e
class
if
ic
ati
o
n
p
r
o
c
ess
,
o
th
er
w
is
e
th
e
co
m
b
in
at
io
n
o
f
th
is
ca
n
b
e
u
s
e
d
f
o
r
f
u
r
th
er
class
if
i
ca
t
io
n
p
r
o
c
ess
.
I
n
o
r
d
e
r
to
p
r
o
v
i
d
e
th
e
o
p
tim
ize
d
H
SI,
th
e
s
in
g
l
e
-
v
ec
t
o
r
h
is
t
o
g
r
a
m
f
o
r
th
e
MB
F
-
P,
MB
F
-
A
an
d
,
M
B
F
-
O
is
;
(
)
{
*
+
(
1
0
)
w
h
er
e,
(
)
r
e
p
r
esen
ts
th
e
(
)
h
is
t
o
g
r
am
v
ec
t
o
r
r
ep
r
esen
ts
is
f
ea
t
u
r
e
m
ap
o
n
th
e
s
c
al
e
an
d
d
en
o
tes
a
lev
e
l
o
f
s
u
b
-
r
eg
i
o
n
.
N
eu
r
a
l
N
etw
o
r
k
(
NN
)
p
e
r
f
o
r
m
s
th
e
class
if
i
ca
ti
o
n
p
r
o
c
e
s
s
af
te
r
th
e
f
ea
tu
r
e
ex
tr
ac
ti
o
n
f
r
o
m
MB
F.
T
h
e
N
N
co
m
p
r
is
es
o
n
e
h
i
d
d
en
lay
er
an
d
o
n
e
o
u
t
p
u
t
lay
er
;
th
e
w
eig
h
t
ass
ig
n
m
en
t
is
p
e
r
f
o
r
m
r
an
d
o
m
l
y
in
b
etw
ee
n
„
I
n
p
u
t
o
f
NN‟
an
d
h
i
d
d
en
lay
er
.
W
eig
h
ts
o
f
th
e
lin
ea
r
o
u
t
p
u
t
lay
er
ca
n
b
e
d
et
e
r
m
in
ed
th
r
o
u
g
h
L
R
A
(
L
in
ea
r
R
eg
r
ess
i
o
n
A
n
aly
s
is
)
[
4
6
]
,
w
h
ich
w
ill
r
ed
u
c
e
th
e
c
o
m
p
u
tatio
n
t
im
e/co
s
t
.
Featu
r
e
m
ap
p
in
g
f
u
n
ctio
n
ca
n
p
r
o
v
i
d
e
im
p
o
r
t
an
t in
f
o
r
m
atio
n
,
w
h
ich
h
elp
t
o
ad
o
p
t th
e
k
e
r
n
e
l b
as
ed
N
N
(
K
NN)
to
o
p
t
im
ize
th
e
g
en
e
r
a
liz
ati
o
n
ab
i
lity
[
1
6
]
.
A
n
alter
n
at
iv
e
m
eth
o
d
o
f
m
u
l
ticl
a
s
s
a
p
p
lic
ati
o
n
c
an
b
e
p
r
o
v
id
e
th
r
o
u
g
h
KNN,
if
w
e
co
n
s
id
e
r
th
e
m
u
lti
-
o
u
tp
u
t
n
o
d
es
in
s
te
ad
o
f
t
ak
in
g
s
in
g
le
-
o
u
tp
u
t
n
o
d
e
.
T
h
e
c
lass
if
i
er
o
f
-
class
h
av
e
-
n
u
m
b
er
o
f
o
u
t
p
u
t
n
o
d
es,
if
th
e
o
r
ig
in
al
l
ab
el
o
f
c
lass
is
,
th
en
th
e
o
u
t
p
u
t
v
ec
t
o
r
w
ith
-
o
u
tp
u
t n
o
d
es
is
;
,
-
(
1
1
)
I
f
w
e
co
n
s
i
d
e
r
o
n
ly
e
lem
en
t o
f
o
u
t
p
u
t
v
ec
to
r
(
)
is
o
n
e
,
th
en
o
u
tp
u
t v
ec
t
o
r
is
;
[
]
(
1
2
)
w
h
er
e,
r
es
ts
o
f
th
e
co
m
p
o
n
en
t
s
ar
e
a
d
j
u
s
t
in
g
to
ze
r
o
an
d
th
e
m
u
lti
-
o
u
tp
u
t
n
o
d
es
f
o
r
th
e
cl
ass
if
ica
ti
o
n
p
r
o
c
ess
o
f
NN
c
an
b
e
p
r
o
v
i
d
in
g
as;
‖
‖
∑
‖
‖
(
1
3
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
l
ec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
A
N
eu
r
a
l Netw
o
r
k
A
p
p
r
o
a
ch
t
o
I
d
en
tify H
yp
ers
p
ec
tr
a
l I
ma
g
e
C
o
n
ten
t (
P
u
tta
s
w
a
my
M
.
R
.
)
2119
w
h
er
e
u
s
er
-
s
p
ec
if
i
ed
p
a
r
am
et
er
(
H)
p
r
o
v
i
d
es
an
a
d
j
u
s
tm
en
t
d
is
tan
c
e
b
etw
ee
n
th
e
tr
ai
n
in
g
er
r
o
r
an
d
th
e
s
ep
ar
atin
g
m
ar
g
in
.
T
h
e
o
u
tp
u
t
„
w
eig
h
t
v
ec
t
o
r
‟
b
etw
ee
n
o
u
tp
u
t
l
ay
er
an
d
h
i
d
d
en
lay
er
is
d
en
o
t
e
as
.
H
id
d
en
lay
er
o
f
o
u
t
p
u
t th
at
c
o
r
r
es
p
o
n
d
s
t
o
in
p
u
t
s
am
p
le
(
)
is
(
)
,
(
)
,
w
h
er
e
(
1
4
)
T
h
e
„
t
r
a
in
in
g
e
r
r
o
r
v
ec
t
o
r
‟
(
)
o
f
-
o
u
tp
u
t n
o
d
es
w
.
r
.
t
th
e
(
)
tr
ain
in
g
s
am
p
le
is
,
[
]
(
1
5
)
„
Kar
u
s
h
–
Ku
h
n
–
T
u
ck
e
r
‟
a
p
p
r
o
ac
h
(
KK
T
)
[
4
7
]
is
c
o
n
s
i
d
e
r
e
d
f
o
r
t
r
ain
in
g
th
e
KNN
,
w
h
ich
co
n
s
i
d
e
r
th
e
d
u
al
o
p
tim
izati
o
n
p
r
o
b
l
em
;
‖
‖
∑
‖
‖
∑
∑
(
(
)
)
(
1
6
)
w
h
er
e,
is
w
eig
h
t v
ec
t
o
r
th
a
t
lin
k
in
g
th
e
h
i
d
d
en
lay
er
to
c
o
r
r
es
p
o
n
d
in
g
o
u
t
p
u
t n
o
d
e
.
,
-
(
1
7
)
T
h
e
o
p
tim
ize
s
o
lu
ti
o
n
c
an
b
e
p
r
o
v
i
d
e
th
r
o
u
g
h
c
o
r
r
es
p
o
n
d
in
g
KK
T
-
A
p
p
r
o
a
ch
,
s
u
ch
as
;
∑
(
(
)
)
(
1
8
)
w
h
er
e,
is
L
ag
r
an
g
e
m
u
ltip
l
i
e
r
[
4
8
]
th
at
c
o
r
r
es
p
o
n
d
s
t
o
th
e
(
)
t
r
ain
in
g
s
am
p
les
.
(
1
9
)
(
)
(
2
0
)
I
n
E
q
u
at
io
n
s
(
1
4
)
to
(
1
8
)
s
h
o
w
s
f
o
r
th
e
s
p
ec
if
i
c
c
ase
o
f
th
e
m
u
ltip
l
e
-
o
u
t
p
u
t
n
o
d
es
an
d
it
al
s
o
c
an
b
e
u
s
e
d
f
o
r
a
s
in
g
le
-
o
u
t
p
u
t
n
o
d
e
v
i
a
s
e
ttin
g
v
alu
e
as
o
n
e
.
W
h
e
r
e
,
th
e
L
a
g
r
an
g
e
m
u
ltip
l
ie
r
[
4
8
]
f
o
r
n
o
n
lin
ea
r
ity
f
u
n
ctio
n
ca
n
b
e
g
iv
en
as;
[
]
(
2
1
)
,
-
(
2
2
)
T
h
er
ef
o
r
e
,
w
e
w
ill
co
n
s
i
d
e
r
t
h
e
m
u
ltip
le
-
o
u
t
p
u
t
n
o
d
es
f
o
r
t
h
e
m
u
lti
-
class
cl
ass
if
ie
r
;
in
th
at
th
e
h
id
d
en
l
ay
er
m
atr
ix
(
H
)
s
iz
e
ca
n
b
e
d
ec
i
d
e
t
h
r
o
u
g
h
th
e
-
n
u
m
b
er
o
f
t
r
ain
in
g
s
am
p
les
.
Q
is
n
u
m
b
er
o
f
th
e
h
id
d
en
n
o
d
es
th
at
ar
e
u
n
r
el
ate
d
t
o
o
u
t
p
u
t
n
o
d
es
n
u
m
b
er
(
i
.
e
.
,
n
u
m
b
er
o
f
clas
s
es)
.
He
r
e
,
w
e
co
n
s
id
e
r
th
e
c
ir
cu
m
s
tan
ce
w
h
er
e
tr
a
in
in
g
s
am
p
les
n
u
m
b
er
is
v
er
y
m
o
r
e
th
an
th
e
f
e
atu
r
e
s
p
a
ce
d
im
en
s
io
n
ality
.
T
h
e
n
u
m
b
er
o
f
t
r
ain
in
g
s
am
p
les
,
w
e
h
av
e
th
e
al
te
r
n
at
i
v
e
s
o
lu
ti
o
n
f
r
o
m
E
q
u
ati
o
n
s
(
1
8
)
an
d
(
1
9
)
;
,
-
(
2
3
)
,
-
(
2
4
)
Fro
m
E
q
u
at
io
n
(
2
0
)
w
e
ca
n
w
r
ite
it
as;
,
-
(
,
-
)
(
2
5
)
[
]
(
2
6
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
8
,
No
.
4
,
A
u
g
u
s
t 2
0
1
8
:
2
1
1
5
–
2
1
2
5
2120
F
o
r
th
is
ci
r
cu
m
s
tan
ce
s
,
a
d
ec
is
io
n
f
u
n
ctio
n
o
f
lin
e
ar
f
e
atu
r
e
s
p
a
ce
„
o
r
‟
th
e
o
u
tp
u
t
f
u
n
cti
o
n
a
l
ity
o
f
Ker
n
el
b
as
ed
NN
-
class
if
i
er
is
,
(
)
(
)
(
)
[
]
(
2
7
)
W
h
e
r
e
,
is
th
e
„
M
o
o
r
e
P
en
r
o
s
e‟
g
en
e
r
a
liz
e
in
v
er
s
e
m
atr
ix
[
4
9
]
,
[
5
0
]
an
d
m
atr
ix
,
-
,
-
.
T
h
i
s
p
r
o
p
o
s
e
d
m
o
d
e
l
ca
n
b
e
u
s
e
d
f
o
r
d
if
f
er
en
t
s
i
ze
o
f
a
p
p
lic
a
tio
n
s
,
h
o
w
ev
er
,
th
ese
a
p
p
r
o
a
ch
es
h
av
e
v
a
r
i
o
u
s
co
m
p
u
tati
o
n
a
l
c
o
s
ts
,
an
d
th
ei
r
ef
f
icien
cy
m
a
y
v
ar
y
f
o
r
th
e
d
if
f
er
en
t
a
p
p
lic
ati
o
n
s
.
I
t
f
o
u
n
d
th
o
u
g
h
th
e
KNN
im
p
lem
e
n
tatio
n
,
th
a
t
th
e
KN
N
p
e
r
f
o
r
m
an
ce
g
en
er
a
liz
ati
o
n
is
n
o
t
m
u
ch
s
en
s
itiv
e
to
w
ar
d
s
th
e
f
ea
tu
r
e
s
p
ac
e
d
im
en
s
io
n
ali
ty
(
)
an
d
th
e
b
e
tte
r
p
er
f
o
r
m
an
ce
c
an
b
e
a
ch
iev
e
d
as
l
o
n
g
as
„
‟
is
m
o
r
e
en
o
u
g
h
.
T
h
er
ef
o
r
e
,
if
th
e
d
at
a
-
s
et
o
f
t
r
ain
in
g
a
r
e
v
er
y
m
o
r
e
(
i.
e.
,
)
th
en
a
p
r
ef
e
r
ab
le
ch
o
ic
e
is
to
a
p
p
ly
E
q
u
ati
o
n
(
2
7
)
in
o
r
d
er
t
o
d
e
cr
ea
s
e
th
e
co
s
t
o
f
co
m
p
u
tat
i
o
n
.
4.
E
XP
E
RI
M
E
NT
A
L
RE
SUL
T
S AN
D
ANAL
YSI
S
4
.
1
.
Da
t
a
-
s
e
t
des
cr
ip
t
i
o
n
I
n
th
is
p
ap
e
r
,
w
e
w
ill
u
s
e
th
e
r
ea
l
HS
I
d
at
ase
t,
w
h
ich
is
av
aila
b
l
e
f
o
r
g
en
e
r
al
-
p
u
r
p
o
s
e
u
s
e.
A
ll
th
e
ex
p
e
r
im
en
ts
h
as
p
e
r
f
o
r
m
ed
u
s
in
g
MA
T
L
A
B
s
im
u
latio
n
o
n
an
I
n
tel
-
i5
2
.
8
0
GH
z
m
ac
h
in
e,
6
4
b
it
o
p
er
atin
g
s
y
s
te
m
w
ith
8
GB
R
A
M.
T
h
e
t
w
o
HI
S
d
ata
-
s
e
t
n
am
el
y
as
Salin
as
-
s
ce
n
e
[
1
7
]
an
d
P
av
i
a
Un
i
v
er
s
ity
[
1
7
]
,
w
e
ar
e
g
o
in
g
to
u
s
e
th
r
o
u
g
h
o
u
t
th
is
p
ap
er
to
v
a
li
d
at
e
o
u
r
c
lass
if
i
ca
t
i
o
n
r
esu
lts
w
ith
r
esp
ec
t
o
th
e
r
s
t
ate
-
of
-
ar
t
m
eth
o
d
s
.
A
t
f
ir
s
t
Salin
as
s
ce
n
e
d
at
aset
h
as
co
lle
ct
ed
o
f
S
alin
as
v
all
ey
(
w
h
ich
is
s
itu
ated
a
t
C
alif
o
r
n
i
a
)
th
r
o
u
g
h
th
e
„
A
VI
R
I
S
-
s
en
s
o
r
‟
w
ith
2
2
4
-
b
a
n
d
s
.
M
o
r
eo
v
er
,
it
ca
n
b
e
ch
a
r
a
cte
r
iz
e
b
y
th
e
h
ig
h
„
s
p
atial
l
ev
el
r
es
o
lu
t
io
n
‟
th
at
is
3
.
7
m
p
;
th
is
d
at
aset
c
o
n
s
is
t
o
f
m
an
y
r
ad
i
an
ce
d
ata
s
u
ch
as
v
i
n
ey
ar
d
,
b
a
r
e
s
o
ils
,
f
iel
d
s
an
d
v
eg
eta
b
l
es.
T
h
e
r
e
a
r
e
to
t
al
s
ix
te
en
c
lass
es
p
r
esen
t
in
th
e
g
r
o
u
n
d
t
r
u
th
o
f
S
alin
as
d
at
aset
.
T
h
e
s
e
co
n
d
c
o
n
s
i
d
e
r
d
at
ase
t
i
s
Pav
iaU
,
w
h
ich
d
atas
et
h
as
c
o
ll
ec
t
e
d
o
f
Pav
ia
Un
iv
er
s
i
ty
at
No
r
th
e
r
n
I
ta
ly
(
w
h
ich
is
d
u
r
in
g
f
lig
h
t
ca
m
p
aig
n
)
th
r
o
u
g
h
th
e
„
R
OSI
S
-
s
en
s
o
r
‟
w
ith
1
0
3
-
b
an
d
s
.
Mo
r
e
o
v
e
r
,
it
ca
n
b
e
ch
ar
ac
t
er
iz
e
b
y
th
e
h
ig
h
„
g
eo
m
etr
ic
r
es
o
lu
ti
o
n
th
at
is
1
.
3
m
w
ith
6
1
0
×
6
1
0
p
ix
els;
th
e
r
e
a
r
e
t
o
t
al
n
in
e
cl
ass
es
p
r
esen
t
in
th
e
g
r
o
u
n
d
tr
u
th
o
f
P
av
iaU
d
at
aset
.
T
ab
le
1
s
h
o
w
th
e
to
tal
n
u
m
b
er
o
f
s
am
p
les
at
ea
ch
clas
s
o
f
Salin
as
Sc
en
e
d
at
aset
[
1
7
]
an
d
it
co
m
p
r
is
es
o
f
5
4
1
2
9
to
tal
s
a
m
p
les.
Sim
ilar
ly
,
T
a
b
l
e
2
s
h
o
w
s
th
e
to
tal
n
u
m
b
er
o
f
s
am
p
les
a
t
ea
ch
c
lass
o
f
P
av
iaU
d
atas
et
[
1
7
]
an
d
it
c
o
m
p
r
is
es
o
f
4
2
7
7
6
to
tal
s
am
p
les
.
T
ab
le
1
.
T
o
tal
Sa
m
p
les at
ea
c
h
C
la
s
s
o
f
Sali
n
as
Scen
e
d
ataset
[
1
7
]
C
L
A
S
S
ES
T
O
T
A
L
S
A
M
P
L
ES
C
l
a
s
s
1
„
B
r
o
c
o
l
i
-
w
e
e
d
s
_
1
‟
2
0
0
9
.
0
C
l
a
s
s
2
„
B
r
o
c
o
l
i
-
w
e
e
d
s
_
2
‟
3
7
2
6
.
0
C
l
a
s
s
3
„
F
a
l
l
o
w
‟
1
9
7
6
.
0
C
l
a
s
s
4
„
R
o
u
g
h
-
F
a
l
l
o
w
‟
1
3
9
4
.
0
C
l
a
s
s
5
„
S
mo
o
t
h
-
F
a
l
l
o
w
‟
2
6
7
8
.
0
C
l
a
s
s
6
„
S
t
u
b
b
l
e
‟
3
9
5
9
.
0
C
l
a
s
s
7
„
C
e
l
e
r
y
‟
3
5
7
9
.
0
C
l
a
s
s
8
„
G
r
a
p
e
s
-
u
n
t
r
a
i
n
e
d
‟
1
1
2
7
1
.
0
C
l
a
s
s
9
„
-
‟
6
2
0
3
.
0
C
l
a
s
s
1
0
„
C
o
r
n
-
g
r
e
e
n
-
w
e
e
d
s‟
3
2
7
8
.
0
C
l
a
s
s
1
1
„
L
e
t
t
u
c
e
_
r
o
m
a
i
n
e
-
4
w
k
‟
1
0
6
8
3
.
0
C
l
a
s
s
1
2
„
L
e
t
t
u
c
e
_
r
o
m
a
i
n
e
-
5
w
k
‟
1
9
2
7
.
0
C
l
a
s
s
1
3
„
L
e
t
t
u
c
e
_
r
o
m
a
i
n
e
-
6
w
k
‟
9
1
6
.
0
C
l
a
s
s
1
4
„
L
e
t
t
u
c
e
_
r
o
m
a
i
n
e
-
7
w
k
‟
1
0
7
0
.
0
C
l
a
s
s
1
5
„
-
‟
7
2
6
8
.
0
C
l
a
s
s
1
6
„
-
‟
1
8
0
7
.
0
T
O
T
A
L
S
A
M
P
L
ES
5
4
1
2
9
.
0
T
a
b
le
2
.
T
o
t
al
Sam
p
les
at
e
ac
h
C
lass
o
f
P
av
i
a
Un
iv
er
s
i
ty
d
at
aset
[
1
7
]
C
L
A
S
S
ES
T
O
T
A
L
S
A
M
P
L
ES
C
l
a
s
s
1
„
A
sp
h
a
l
t
‟
6
6
3
1
.
0
C
l
a
s
s
2
„
M
e
a
d
o
w
s‟
1
8
6
4
9
.
0
Cl
a
s
s
2
„
G
r
a
v
e
l
‟
2
0
9
9
.
0
C
l
a
s
s
2
„
T
r
e
e
s‟
3
0
6
4
.
0
C
l
a
s
s
2
„
M
e
t
a
l
s
h
e
e
t
s‟
1
3
4
5
.
0
C
l
a
s
s
2
„
B
a
r
e
S
o
i
l
‟
5
0
2
9
.
0
C
l
a
s
s
2
„
B
i
t
u
me
n
‟
1
3
3
0
.
0
C
l
a
s
s
2
„
S
e
l
f
-
B
l
o
c
k
i
n
g
-
B
r
i
c
k
s‟
3
6
8
2
.
0
C
l
a
s
s
2
„
S
h
a
d
o
w
s‟
9
4
7
.
0
T
O
T
A
L
S
A
M
P
L
ES
4
2
7
7
6
.
0
4
.
2
.
Co
m
pa
ra
t
i
v
e
s
t
ud
y
Her
e
w
e
co
m
p
ar
e
th
e
c
lass
if
i
c
ati
o
n
p
e
r
f
o
r
m
an
ce
o
f
s
ev
er
al
s
tate
o
f
ar
t
w
ith
r
esp
ec
t
to
o
u
r
p
r
o
p
o
s
e
d
m
o
d
el.
I
n
th
is
s
ec
ti
o
n
,
a
p
r
o
p
o
s
e
d
p
r
e
p
r
o
ce
s
s
in
g
MB
F
-
A
lg
o
r
ith
m
w
ill
co
m
p
ar
e
w
ith
s
o
m
e
ex
is
tin
g
p
o
p
u
la
r
f
ea
tu
r
e
ex
t
r
a
cti
o
n
m
eth
o
d
s
s
u
c
h
as
MN
F
[
1
0
]
,
P
C
A
[
1
1
]
,
S
PP
[
1
2
]
,
L
PP
[
1
3
]
,
MSM
E
[
1
4
]
,
an
d
S
P
A
[
1
5
]
.
T
h
e
k
ap
p
a
c
o
ef
f
ic
ien
t
o
f
class
if
ica
tio
n
(
Ka
p
p
a
)
an
d
o
v
e
r
al
l
ac
c
u
r
ac
y
(
OA
)
h
as
u
s
ed
t
o
s
h
o
w
th
e
ac
cu
r
a
cy
af
ter
class
if
i
ca
t
io
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
l
ec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
A
N
eu
r
a
l Netw
o
r
k
A
p
p
r
o
a
ch
t
o
I
d
en
tify H
yp
ers
p
ec
tr
a
l I
ma
g
e
C
o
n
ten
t (
P
u
tta
s
w
a
my
M
.
R
.
)
2121
T
ak
in
g
a
Sa
lin
asS
d
a
tase
t,
th
e
class
if
i
ca
t
io
n
r
esu
l
t
o
f
d
if
f
er
e
n
t
m
eth
o
d
h
as
c
o
m
p
ar
e
d
w
ith
p
r
o
p
o
s
e
d
MB
F
h
as
s
h
o
w
n
in
T
a
b
l
e
3
.
T
h
e
M
B
F
(
P
A
)
is
co
m
p
a
r
e
d
w
ith
r
esp
ec
t
t
o
M
NF
[
1
0
]
,
PC
A
[
1
1
]
,
S
PP
[
1
2
]
,
L
PP
[
1
3
]
,
MSM
E
[
1
4
]
,
an
d
S
P
A
[
1
5
]
m
eth
o
d
,
b
as
ed
u
p
o
n
t
h
e
Ka
p
p
a
(
%
)
,
OA
(
%
)
an
d
th
eir
in
d
iv
id
u
al
class
.
I
n
T
a
b
le
3
,
f
iv
e
cl
ass
es
g
o
t
a
cc
u
r
ac
y
o
f
1
0
0
%
at
M
B
F
alg
o
r
i
th
m
(
class
es
n
u
m
b
er
ar
e
3
,
4
,
9
,
1
1
an
d
1
6
)
.
T
h
e
o
v
e
r
a
ll
ac
cu
r
ac
y
o
f
p
r
o
p
o
s
ed
M
B
F
m
o
d
el
h
as
c
o
m
p
ar
e
d
w
ith
r
esp
ec
t
t
o
MN
F
[
1
0
]
,
PC
A
[
1
1
]
,
S
PP
[
1
2
]
,
L
PP
[
1
3
]
,
MSM
E
[
1
4
]
,
an
d
S
P
A
[
1
5
]
an
d
w
e
g
o
t
im
p
r
o
v
e
d
class
if
i
ca
t
io
n
r
esu
l
ts
o
f
6
.
2
3
%
,
7
.
8
%
,
9
.
4
%
,
7
.
9
%
,
4
.
9
%
,
an
d
9
.
4
%
.
T
h
e
Ka
p
p
a
ac
cu
r
a
cy
o
f
p
r
o
p
o
s
e
d
M
B
F
m
o
d
el
h
as
c
o
m
p
ar
ed
w
ith
r
es
p
e
ct
t
o
MN
F
[
1
0
]
,
P
C
A
[
1
1
]
,
S
PP
[
1
2
]
,
L
PP
[
1
3
]
,
MSM
E
[
1
4
]
,
S
P
A
[
1
5
]
an
d
w
e
g
o
t
im
p
r
o
v
e
d
class
if
ica
ti
o
n
r
esu
lts
o
f
7
.
0
3
%
,
8
.
7
5
%,
1
0
.
5
7
%,
8
.
8
%,
5
.
5
%
,
a
n
d
1
0
.
5
7
%.
T
a
b
le
3
.
C
lass
w
is
e
C
o
m
p
ar
is
o
n
o
n
th
e
d
at
aset
(
)
C
l
a
s
s
N
u
m
b
e
r
M
N
F
[
1
0
]
P
C
A
[
1
1
]
S
P
P
[
1
2
]
L
P
P
[
1
3
]
M
S
M
E
[
1
4
}
S
P
A
[
1
5
]
M
B
F
(
P
A
)
1
9
9
.
8
0
9
9
.
3
0
9
8
.
9
0
9
9
.
4
0
9
9
.
9
0
8
8
.
7
0
9
9
.
7
4
2
1
0
0
.
0
9
9
.
6
0
1
0
0
.
0
1
0
0
.
0
1
0
0
.
0
9
8
.
6
0
9
9
.
8
9
3
1
0
0
.
0
9
9
.
3
0
9
7
.
2
0
1
0
0
.
0
1
0
0
.
0
9
8
.
4
0
1
0
0
.
0
4
9
9
.
8
0
9
9
.
7
0
9
9
.
3
0
9
9
.
6
0
9
9
.
9
0
9
7
.
5
0
1
0
0
.
0
5
9
7
.
7
0
9
6
.
7
0
9
2
.
7
0
9
7
.
8
0
9
8
.
4
0
9
4
.
6
0
9
6
.
6
0
6
9
9
.
8
0
9
9
.
7
0
9
9
.
9
0
9
9
.
9
0
9
9
.
9
0
9
7
.
0
0
9
6
.
7
0
7
9
9
.
6
0
9
8
.
4
0
9
9
.
2
0
9
9
.
6
0
9
9
.
7
0
9
8
.
5
0
9
9
.
9
7
8
8
5
.
0
0
7
7
.
4
0
7
8
.
9
0
8
3
.
1
0
8
6
.
3
0
7
5
.
2
0
9
8
.
7
3
9
9
9
.
4
0
9
8
.
1
0
9
6
.
2
0
9
9
.
6
0
9
9
.
9
0
9
6
.
4
0
1
0
0
.
0
10
9
5
.
6
0
9
1
.
1
0
9
1
.
6
0
9
3
.
0
0
9
7
.
3
0
8
2
.
8
0
9
8
.
1
7
11
9
9
.
0
0
9
8
.
1
0
9
6
.
9
0
9
8
.
6
0
9
9
.
0
0
9
3
.
8
0
1
0
0
.
0
12
1
0
0
.
0
1
0
0
.
0
9
9
.
9
0
1
0
0
.
0
9
9
.
9
0
9
2
.
6
0
9
9
.
6
7
13
9
8
.
2
0
9
9
.
5
0
9
8
.
0
0
9
8
.
2
0
1
0
0
.
0
8
7
.
4
0
9
9
.
5
0
14
9
5
.
3
0
9
3
.
1
0
9
3
.
9
0
9
4
.
6
0
9
8
.
5
0
9
6
.
8
0
9
8
.
9
7
15
7
4
.
8
0
8
1
.
9
0
6
9
.
1
0
6
6
.
8
0
8
0
.
5
0
7
0
.
4
0
9
8
.
6
0
16
9
9
.
0
0
9
8
.
7
0
9
8
.
5
0
9
8
.
9
0
9
7
.
6
0
9
6
.
3
0
1
0
0
.
0
O
A
(
%
)
9
2
.
8
0
9
1
.
2
0
8
9
.
6
0
9
1
.
1
0
9
4
.
1
0
8
9
.
6
0
9
8
.
9
7
K
a
p
p
a
(
%)
9
1
.
9
0
9
0
.
2
0
8
8
.
4
0
9
0
.
1
0
9
3
.
4
0
8
8
.
4
0
9
8
.
8
5
T
a
b
le
4
s
h
o
w
s
th
e
class
w
is
e
class
if
i
ca
ti
o
n
r
esu
lts
o
n
a
Pav
iaU
d
at
aset
.
Ou
t
o
f
n
in
e
cl
ass
es
,
o
u
r
p
r
o
p
o
s
e
d
m
o
d
el
M
B
F
h
as
m
o
r
e
ac
cu
r
a
cy
in
f
iv
e
class
es
(
c
las
s
n
u
m
b
er
2
,
3
,
6
,
7
an
d
8
)
.
T
h
e
o
v
e
r
al
l
ac
cu
r
a
cy
o
f
p
r
o
p
o
s
e
d
M
B
F
m
o
d
el
h
as
c
o
m
p
ar
e
d
w
ith
r
esp
ec
t
t
o
MN
F
[
1
0
]
,
P
C
A
[
1
1
]
,
S
PP
[
1
2
]
,
L
PP
[
1
3
]
,
MSM
E
[
1
4
]
,
SP
A
[
1
5
]
an
d
w
e
g
o
t
i
m
p
r
o
v
ed
cl
ass
if
ic
ati
o
n
r
esu
lts
o
f
4
.
8
9
%,
1
.
8
%
,
1
4
.
2
%,
8
.
4
%
,
0
.
3
%
,
an
d
1
5
.
8
%
.
T
h
e
Ka
p
p
a
a
cc
u
r
a
cy
o
f
p
r
o
p
o
s
ed
M
B
F
m
o
d
el
h
as
c
o
m
p
ar
e
d
w
ith
r
esp
e
ct
t
o
MN
F
[
1
0
]
,
P
C
A
[
1
1
]
,
S
PP
[
1
2
]
,
L
PP
[
1
3
]
,
MS
ME
[
1
4
]
,
S
P
A
[
1
5
]
an
d
w
e
g
o
t
im
p
r
o
v
e
d
cl
as
s
if
icat
io
n
r
esu
lts
o
f
6
.
4
%
,
2
.
3
8
%,
1
8
.
3
8
%
,
1
0
.
8
%,
0
.
4
2
%,
an
d
2
0
.
7
%
.
T
a
b
le
4
.
C
lass
w
is
e
C
o
m
p
ar
is
o
n
o
n
th
e
d
at
aset
(
)
C
l
a
s
s
N
u
m
b
e
r
M
N
F
[
1
0
]
P
C
A
[
1
1
]
S
P
P
[
1
2
]
L
P
P
[
1
3
]
M
S
M
E
[
1
4
}
S
P
A
[
1
5
]
M
B
F
(
P
A
)
1
8
7
.
3
0
8
5
.
5
0
7
6
.
8
0
8
4
.
1
0
9
2
.
3
0
6
9
.
2
0
9
0
.
4
7
2
8
6
.
7
0
9
3
.
2
0
7
6
.
7
0
8
1
.
3
0
9
4
.
4
0
8
2
.
4
0
9
9
.
5
8
3
8
4
.
3
0
8
9
.
0
0
7
9
.
6
0
8
0
.
8
0
7
8
.
1
0
7
1
.
1
0
9
3
.
8
8
4
9
2
.
7
0
9
3
.
4
0
9
3
.
2
0
92.
60
9
4
.
0
0
8
8
.
7
0
6
6
.
6
8
5
1
0
0
.
0
9
9
.
8
0
9
9
.
8
0
1
0
0
.
0
1
0
0
.
0
9
2
.
5
0
9
0
.
8
0
6
9
2
.
8
0
9
1
.
7
0
8
3
.
9
0
9
2
.
5
0
9
3
.
9
0
7
8
.
7
0
9
9
.
9
0
7
9
7
.
4
0
9
7
.
5
0
8
7
.
1
0
9
6
.
9
0
9
6
.
9
0
8
2
.
5
0
9
9
.
8
0
8
9
1
.
9
0
9
3
.
9
0
7
7
.
9
0
8
9
.
5
0
9
4
.
3
0
6
5
.
5
0
9
6
.
2
8
9
9
9
.
9
0
1
0
0
.
0
1
0
0
.
0
1
0
0
.
0
9
9
.
9
0
9
6
.
5
0
4
8
.
9
7
O
A
(
%
)
89
.
3
0
9
2
.
2
0
8
0
.
5
0
8
6
.
0
0
9
3
.
6
0
7
9
.
0
0
9
3
.
9
0
K
a
p
p
a
(
%)
8
6
.
0
0
8
9
.
7
0
7
5
.
0
0
8
1
.
9
0
9
1
.
5
0
7
2
.
8
0
9
1
.
8
9
C
lass
if
ic
ati
o
n
r
esu
lts
at
1
0
-
s
a
m
p
le
s
ize
o
f
Sa
lin
asS
-
d
at
aset
h
as
s
h
o
w
n
in
Fig
u
r
e
1
,
w
h
e
r
e
to
p
l
in
e
m
ar
k
at
e
ac
h
b
a
r
r
ep
r
es
en
t
th
e
er
r
o
r
b
a
r
.
P
r
o
p
o
s
e
d
m
o
d
el
is
c
o
m
p
a
r
e
d
in
te
r
m
s
o
f
OA
an
d
Ka
p
p
a
(
KA
)
w
ith
MN
F
[
1
0
]
,
P
C
A
[
1
1
]
,
S
PP
[
1
2
]
,
L
PP
[
1
3
]
,
MSM
E
[
1
4
]
,
an
d
SP
A
[
1
5
]
.
A
s
p
er
Fig
u
r
e
1
,
MB
F
is
c
o
m
p
ar
ed
in
ter
m
s
OA
(
%
)
an
d
w
e
g
o
t 1
1
.
4
3
%,
9
.
9
%,
1
5
.
8
9
%
,
1
5
.
6
9
%
,
1
3
.
1
6
% a
n
d
1
1
.
1
3
%
im
p
r
o
v
em
en
t
w
.
r
.
t
c
o
n
s
i
d
e
r
e
d
s
tate
o
f
a
r
t.
Sim
ilar
ly
,
MB
F
is
co
m
p
a
r
e
d
in
te
r
m
s
KA
(
%)
an
d
w
e
g
o
t
1
2
.
7
%
,
1
0
.
9
7
%,
1
7
.
6
8
%,
1
7
.
4
7
%
,
1
4
.
3
%
an
d
1
2
.
3
9
%
im
p
r
o
v
em
en
t
w
.
r
.
t
c
o
n
s
id
e
r
e
d
s
t
ate
o
f
a
r
t.
Fig
u
r
e
2
s
h
o
w
s
th
e
cl
ass
if
ic
ati
o
n
r
e
s
u
lts
at
2
0
-
s
am
p
le
s
ize
(
S
alin
asS
-
d
a
tase
t)
,
h
er
e
MB
F
is
c
o
m
p
ar
e
d
in
t
er
m
s
O
A
(
%)
an
d
w
e
g
o
t
1
1
.
6
1
%,
7
.
9
6
%
,
1
2
.
1
1
%
,
1
3
%,
1
0
%
an
d
7
.
9
%
im
p
r
o
v
em
en
t
w
.
r
.
t
c
o
n
s
id
e
r
e
d
s
t
ate
o
f
ar
t.
S
im
ilar
ly
,
MBF
is
co
m
p
a
r
e
d
in
te
r
m
s
KA
(
%)
an
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
8
,
No
.
4
,
A
u
g
u
s
t 2
0
1
8
:
2
1
1
5
–
2
1
2
5
2122
w
e
g
o
t 1
2
.
8
%,
8
.
8
%
,
1
3
.
4
1
%
,
1
4
.
4
%,
1
1
.
2
% a
n
d
8
.
6
5
% im
p
r
o
v
em
en
t
w
.
r
.
t c
o
n
s
i
d
e
r
e
d
s
ta
te
o
f
ar
t
.
I
n
F
ig
u
r
e
3
,
a
p
r
o
p
o
s
e
d
M
B
F
alg
o
r
ith
m
p
e
r
f
o
r
m
s
1
0
.
7
4
%
,
6
.
7
%
,
9
.
9
%
,
1
0
.
4
%
,
9
%,
7
.
3
%
b
e
tte
r
o
v
e
r
a
ll
ac
cu
r
ac
y
th
an
th
e
ex
is
tin
g
f
ea
tu
r
e
alg
o
r
i
th
m
MN
F
[
1
0
]
,
P
C
A
[
1
1
]
,
S
PP
[
1
2
]
,
L
PP
[
1
3
]
,
MSM
E
[
1
4
]
,
an
d
S
PA
[
1
5
]
;
in
ter
m
s
KA
(
%)
,
w
e
g
o
t
1
1
.
9
%
,
7
.
5
%
,
1
1
%,
1
1
.
5
%,
1
0
%
an
d
8
.
1
%
i
m
p
r
o
v
em
en
t.
A
t
40
-
s
am
p
lin
g
s
ize
th
at
s
h
o
w
s
in
F
ig
u
r
e
4
,
M
B
F
p
e
r
f
o
r
m
s
1
0
.
4
%,
5
.
3
%
,
8
%,
9
.
5
%,
6
.
4
%,
6
.
6
%
b
ett
er
o
v
er
all
ac
cu
r
ac
y
th
an
th
e
c
o
n
s
i
d
e
r
e
d
f
ea
tu
r
e
alg
o
r
i
th
m
an
d
in
te
r
m
s
KA
(
%)
,
w
e
g
o
t
1
1
.
6
%
,
6
.
7
%
,
1
0
%
,
1
1
.
3
8
%
,
8
.
3
%
an
d
8
.
1
4
% im
p
r
o
v
em
en
t.
A
s
g
iv
en
in
Fig
u
r
e
1
t
o
Fig
u
r
e
4
f
o
r
-
d
at
aset
,
s
im
ila
r
ly
in
Fig
u
r
e
5
t
o
F
ig
u
r
e
8
h
as
g
iv
en
f
o
r
-
d
at
aset
.
Fig
u
r
e
5
,
s
h
o
w
s
th
e
class
if
i
ca
t
io
n
r
esu
lts
a
t
1
0
-
s
a
m
p
le
s
iz
e
(
-
d
at
aset
)
,
h
er
e
M
B
F
is
co
m
p
a
r
e
d
in
te
r
m
s
OA
(
%)
an
d
w
e
g
o
t
2
5
%
,
1
4
%,
4
1
%
,
3
4
%,
1
9
.
5
%
an
d
3
3
.
6
%
im
p
r
o
v
e
m
en
t
w
ith
r
esp
e
ct
t
o
co
n
s
i
d
e
r
e
d
s
t
ate
o
f
ar
t
m
eth
o
d
s
.
Sim
ilar
ly
,
MB
F
is
c
o
m
p
a
r
e
d
in
te
r
m
s
K
A
(
%)
an
d
w
e
g
o
t
3
1
%
,
1
8
%
,
5
0
%
,
4
1
.
4
%,
2
4
%
an
d
4
1
.
6
%
im
p
r
o
v
em
en
t
w
.
r
.
t
c
o
n
s
i
d
e
r
e
d
s
t
ate
o
f
a
r
t.
I
n
Fig
u
r
e
6
an
d
7
,
th
e
k
ap
p
a
c
o
ef
f
ici
e
n
t
o
f
MB
F
is
9
1
.
5
2
an
d
9
1
.
8
,
an
d
t
h
e
o
v
er
all
a
cc
u
r
ac
y
is
9
3
.
8
an
d
9
3
.
9
.
I
n
F
ig
u
r
e
8
,
a
p
r
o
p
o
s
ed
M
B
F
alg
o
r
i
th
m
p
e
r
f
o
r
m
s
1
7
%
,
4
.
4
%,
1
8
.
7
%
,
2
2
%,
5
.
8
%,
1
4
.
3
6
%
b
et
te
r
KA
(
%)
th
an
th
e
ex
is
t
in
g
f
ea
tu
r
e
al
g
o
r
i
th
m
MN
F [
1
0
]
,
P
C
A
[
1
1
]
,
S
PP
[
1
2
]
,
L
PP
[
1
3
]
,
MS
ME
[
1
4
]
,
an
d
S
P
A
[
1
5
]
.
T
h
e
an
aly
s
is
o
f
a
b
o
v
e
r
esu
lt
p
r
o
v
i
d
es
th
e
p
e
r
f
o
r
m
an
ce
ev
alu
at
i
o
n
o
f
MB
F
m
o
d
el
,
w
h
er
e
th
e
p
e
r
f
o
r
m
an
ce
in
ev
e
r
y
s
a
m
p
lin
g
s
ize
is
in
cr
e
asin
g
as
p
e
r
in
cr
em
en
t in
th
e
t
r
ain
in
g
-
s
am
p
le
-
s
ize
.
Fig
u
r
e
1
.
C
lass
if
i
ca
ti
o
n
r
es
u
lts
at
1
0
-
s
am
p
le
s
i
ze
(
-
d
atas
et
)
Fig
u
r
e
2
.
C
lass
if
i
ca
ti
o
n
r
esu
lts
at
2
0
-
s
am
p
le
s
i
ze
(
-
d
atas
et
)
Fig
u
r
e
3
.
C
lass
if
i
ca
ti
o
n
r
esu
lts
at
3
0
-
s
am
p
le
s
i
ze
(
-
d
atas
et
)
Fig
u
r
e
4
.
C
lass
if
i
ca
ti
o
n
r
esu
lts
at
4
0
-
s
am
p
le
s
i
ze
(
-
d
atas
et
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
l
ec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
A
N
eu
r
a
l Netw
o
r
k
A
p
p
r
o
a
ch
t
o
I
d
en
tify H
yp
ers
p
ec
tr
a
l I
ma
g
e
C
o
n
ten
t (
P
u
tta
s
w
a
my
M
.
R
.
)
2123
Fig
u
r
e
5
.
C
lass
if
i
ca
ti
o
n
r
esu
lts
at
1
0
-
s
am
p
le
s
i
ze
(
-
d
atas
et
)
Fig
u
r
e
6
.
C
lass
if
i
ca
ti
o
n
r
esu
lts
at
2
0
-
s
am
p
le
s
i
ze
(
-
d
atas
et
)
Fig
u
r
e
7
.
C
lass
if
i
ca
ti
o
n
r
esu
lts
at
3
0
-
s
am
p
le
s
i
ze
(
-
d
atas
et
)
Fig
u
r
e
8
.
C
lass
if
i
ca
ti
o
n
r
esu
lts
at
4
0
-
s
am
p
le
s
i
ze
(
-
d
atas
et
)
5.
CO
NCLU
SI
O
N
T
h
e
HSI
ar
e
al
w
a
y
s
h
a
v
in
g
a
lar
g
e
n
u
m
b
er
o
f
co
r
r
elate
d
an
d
r
ed
u
n
d
an
t
f
ea
tu
r
es,
w
h
ic
h
ca
u
s
e
s
th
e
d
ec
r
em
e
n
t
i
n
t
h
e
clas
s
i
f
icatio
n
ac
cu
r
ac
y
.
He
n
ce
,
HI
S
d
ata
p
r
o
ce
s
s
in
g
b
ec
o
m
e
s
s
o
m
e
w
h
a
t
d
if
f
ic
u
lt,
e
s
p
ec
iall
y
w
it
h
t
h
e
s
u
p
er
v
is
ed
lear
n
in
g
m
et
h
o
d
b
ec
au
s
e
th
e
ac
c
u
r
ac
y
o
f
class
i
f
icatio
n
d
e
cr
ea
s
es
w
it
h
s
p
ec
i
f
ic
s
et
o
f
tr
ain
i
n
g
a
s
i
n
cr
ea
s
i
n
g
f
ea
tu
r
e
s
n
u
m
b
er
.
I
n
o
r
d
er
to
ac
h
iev
e
h
i
g
h
er
cla
s
s
i
f
icat
io
n
ac
c
u
r
a
c
y
,
d
i
m
e
n
s
io
n
alit
y
r
ed
u
ctio
n
ap
p
r
o
ac
h
b
ec
o
m
i
n
g
v
er
y
b
en
e
f
icial
a
n
d
it
als
o
r
ed
u
ce
s
t
h
e
r
eq
u
ir
e
m
e
n
t
o
f
d
ata
s
to
r
ag
e
a
n
d
co
m
p
u
tatio
n
al
ti
m
e.
I
n
t
h
i
s
p
a
p
er
,
an
u
n
s
u
p
er
v
i
s
ed
b
ased
B
S
A
is
co
n
s
id
er
ed
w
ith
t
h
e
L
P
th
at
d
ep
en
d
s
u
p
o
n
th
e
m
etr
ic
-
b
a
n
d
s
i
m
ilar
itie
s
.
S
u
b
s
eq
u
e
n
tl
y
MB
F
h
a
s
co
n
s
id
er
ed
p
er
f
o
r
m
i
n
g
t
h
e
„
te
x
tu
r
e
a
n
al
y
s
e
s
o
f
t
h
e
HSI
,
w
h
er
e
th
r
ee
o
p
er
atio
n
al
co
m
p
o
n
en
ts
r
ep
r
esen
t
s
th
e
m
o
n
o
g
en
et
ic
s
i
g
n
al
s
u
c
h
as;
p
h
as
e,
am
p
lit
u
d
e
an
d
o
r
ien
tatio
n
.
I
n
p
o
s
t
p
r
o
ce
s
s
in
g
class
if
icatio
n
s
ta
g
e,
th
e
KNN
h
as
ap
p
lied
to
o
p
tim
ize
t
h
e
g
e
n
er
aliza
tio
n
ab
ilit
y
b
y
a
n
alter
n
ati
v
e
m
et
h
o
d
o
f
m
u
lticla
s
s
ap
p
licatio
n
t
h
r
o
u
g
h
co
n
s
id
er
in
g
th
e
m
u
lti
-
o
u
tp
u
t
n
o
d
es
i
n
s
tead
o
f
tak
i
n
g
s
in
g
le
-
o
u
tp
u
t n
o
d
e.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
cla
s
s
i
f
ic
atio
n
ac
cu
r
ac
y
h
as
co
m
p
ar
ed
w
it
h
t
h
e
ac
c
u
r
ac
y
o
f
s
ev
er
al
s
tate
-
of
-
ar
t
m
et
h
o
d
s
an
d
f
r
o
m
t
h
e
ab
o
v
e
an
al
y
s
is
o
f
r
es
u
lt
s
o
u
r
p
r
o
p
o
s
ed
m
o
d
el
h
as
p
er
f
o
r
m
co
n
s
id
er
ab
ly
w
e
ll
i
n
ev
er
y
tr
ain
i
n
g
s
et.
P
r
o
p
o
s
ed
m
o
d
el
OA
a
n
d
Kap
p
a
v
al
u
es a
t 4
0
-
tr
ain
i
n
g
s
a
m
p
le
s
is
9
8
.
9
7
% a
n
d
9
8
.
8
5
% a
t Salin
as
-
S
d
ataset;
s
i
m
ilar
l
y
at
P
av
ia
-
U
d
ataset,
w
e
g
o
t
9
3
.
9
%
an
d
9
1
.
9
%
o
f
o
v
er
all
a
n
d
Kap
p
a
clas
s
if
ica
tio
n
ac
c
u
r
ac
y
.
I
n
f
u
t
u
r
e
w
o
r
k
,
e
x
p
er
i
m
e
n
ts
s
h
o
u
ld
b
e
i
m
p
le
m
e
n
t o
n
v
er
y
la
r
g
e
d
atasets
in
o
r
d
er
to
ca
lcu
late
th
e
s
ca
lab
ilit
y
o
f
s
ev
er
al
ap
p
r
o
ac
h
es.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
8
,
No
.
4
,
A
u
g
u
s
t 2
0
1
8
:
2
1
1
5
–
2
1
2
5
2124
RE
F
E
R
E
NC
E
S
[1
]
A
.
M
u
n
tas
a
,
I
.
A
.
S
iraju
d
in
,
M
.
H
.
P
u
rn
o
m
o
,
“
A
p
p
e
a
ra
n
c
e
g
lo
b
a
l
a
n
d
l
o
c
a
l
stru
c
tu
re
f
u
sio
n
f
o
r
f
a
c
e
i
m
a
g
e
re
c
o
g
n
it
io
n
”
,
T
EL
KOM
NIKA
(
T
e
lec
o
mm
u
n
ica
ti
o
n
,
Co
mp
u
ti
n
g
,
El
e
c
tro
n
ics
a
n
d
C
o
n
tr
o
l)
,
2
0
1
1
,
v
o
l.
9
,
n
o
.
1
,
pp.
1
2
5
-
13
2
.
[2
]
H
.
G
r
a
h
n
a
n
d
P
.
G
e
lad
i,
“
T
e
c
h
n
iq
u
e
s
a
n
d
a
p
p
li
c
a
ti
o
n
s
o
f
h
y
p
e
rsp
e
c
tral
i
m
a
g
e
a
n
a
l
y
sis,
”
Ch
ich
e
ste
r,
En
g
lan
d
;
Ho
b
o
k
e
n
,
J.
W
il
e
y
,
2
0
0
7
.
[3
]
K.
F
u
k
u
n
a
g
a
a
n
d
R.
R.
Ha
y
e
s,
“
Eff
e
c
ts
o
f
sa
m
p
le
siz
e
in
c
las
si
fi
e
r
d
e
sig
n
,
”
IEE
E
T
ra
n
s.
P
a
tt
e
rn
An
a
l.
M
a
c
h
.
In
tell.
,
v
o
l
.
11
,
n
o
.
8
,
p
p
.
8
7
3
-
8
8
5
,
1
9
8
9
.
[4
]
M
.
P
a
l
a
n
d
G
.
M
.
F
o
o
d
y
,
“
F
e
a
tu
re
S
e
lec
ti
o
n
f
o
r
Clas
sif
ic
a
ti
o
n
o
f
H
y
p
e
rsp
e
c
tral
Da
ta
b
y
S
V
M
,
”
IEE
E
T
ra
n
s
a
c
ti
o
n
s
o
n
Ge
o
sc
ien
c
e
a
n
d
Rem
o
te S
e
n
si
n
g
,
v
o
l
.
48
,
n
o
.
5
,
p
p
.
2
2
9
7
-
2
3
0
7
,
2
0
1
0
.
[5
]
A
.
Ja
n
e
c
e
k
,
e
t
a
l.
,
“
On
th
e
Re
latio
n
sh
i
p
b
e
tw
e
e
n
F
e
a
tu
re
S
e
lec
ti
o
n
a
n
d
Clas
sif
ica
ti
o
n
A
c
c
u
ra
c
y
,”
J
o
u
rn
a
l
o
f
M
a
c
h
in
e
L
e
a
r
n
in
g
Res
e
a
rc
h
-
Pro
c
e
e
d
in
g
s
T
ra
c
k
,
v
o
l.
4
,
p
p
.
90
-
1
0
5
,
2
0
0
8.
[6
]
B.
M
.
S
h
a
h
sh
a
h
a
n
i
a
n
d
D.
A
.
L
a
n
d
g
re
b
e
,
“
T
h
e
e
ff
e
c
t
o
f
u
n
lab
e
led
sa
m
p
les
in
re
d
u
c
i
n
g
th
e
sm
a
ll
sa
m
p
le
siz
e
p
ro
b
lem
a
n
d
m
it
ig
a
ti
n
g
th
e
H
u
g
h
e
s
p
h
e
n
o
m
e
n
o
n
,
”
IEE
E
T
r
a
n
s.
Ge
o
sc
i.
Rem
o
te
S
e
n
s.
,
v
o
l
.
32
,
n
o
.
5
,
p
p
.
1
0
8
7
-
1
0
9
5
,
1
9
9
4
.
[7
]
G
.
Hu
g
h
e
s,
“
On
th
e
m
e
a
n
a
c
c
u
r
a
c
y
o
f
sta
ti
stica
l
p
a
tt
e
rn
re
c
o
g
n
ize
rs,”
IEE
E
T
ra
n
s.
In
f.
T
h
e
o
ry
,
v
o
l
.
14
,
n
o
.
1
,
p
p
.
5
5
-
6
3
,
1
9
6
8
.
[8
]
Z.
S
u
n
,
e
t
a
l.
,
“
Ob
jec
t
d
e
tec
ti
o
n
u
sin
g
f
e
a
tu
re
su
b
se
t
se
lec
ti
o
n
,
”
P
a
tt
e
rn
Rec
o
g
n
it
.
,
v
o
l
.
37
,
n
o
.
1
1
,
p
p
.
2
1
6
5
-
2
1
7
6
,
2
0
0
4
.
[9
]
I.
G
u
y
o
n
,
e
t
a
l
.
,
“
G
e
n
e
se
lec
ti
o
n
f
o
r
c
a
n
c
e
r
c
las
sifi
c
a
ti
o
n
u
si
n
g
su
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
s,”
M
a
c
h
.
L
e
a
rn
.
,
v
o
l
.
46
,
p
p
.
1
-
3
,
p
p
.
3
8
9
-
4
2
2
,
2
0
0
2
.
[1
0
]
G
.
L
ix
in
,
e
t
a
l.
,
“
S
e
g
m
e
n
ted
m
in
im
u
m
n
o
ise
f
ra
c
ti
o
n
tran
sf
o
rm
a
ti
o
n
f
o
r
e
ff
icie
n
t
f
e
a
tu
re
e
x
tra
c
ti
o
n
o
f
h
y
p
e
rsp
e
c
tral
im
a
g
e
s,
”
Pa
tt
e
rn
Rec
o
g
n
it
.
,
v
o
l
.
48
,
n
o
.
10
,
p
p
.
3
2
1
6
-
3
2
2
6
,
2
0
1
5
.
[1
1
]
X
.
M
.
Ch
e
n
g
,
e
t
a
l.
,
“
A
n
o
v
e
l
i
n
teg
ra
ted
P
CA
a
n
d
F
L
D
m
e
th
o
d
o
n
h
y
p
e
rsp
e
c
tral
i
m
a
g
e
f
e
a
tu
re
e
x
tra
c
ti
o
n
f
o
r
c
u
c
u
m
b
e
r
c
h
il
li
n
g
d
a
m
a
g
e
in
sp
e
c
ti
o
n
,
”
'
T
ra
n
s.
AS
AE
,
v
o
l
.
47
,
n
o
.
4
,
p
p
.
1
3
1
3
-
1
3
2
0
,
2
0
0
4
.
[1
2
]
L
.
Qia
o
,
e
t
a
l.
,
“
S
p
a
rsity
p
re
s
e
rv
in
g
p
ro
jec
ti
o
n
s
w
it
h
a
p
p
li
c
a
ti
o
n
s
t
o
f
a
c
e
re
c
o
g
n
it
io
n
,
”
Pa
t
ter
n
Rec
o
g
n
it
.
,
v
o
l
.
43
,
n
o
.
1
,
p
p
.
3
3
1
-
3
4
1
,
2
0
1
0
.
[1
3
]
Z.
W
a
n
g
a
n
d
B.
He
,
“
L
o
c
a
li
t
y
p
e
rse
rv
in
g
p
ro
jec
ti
o
n
s
a
lg
o
rit
h
m
f
o
r
h
y
p
e
rsp
e
c
tral
i
m
a
g
e
d
i
m
e
n
sio
n
a
li
ty
r
e
d
u
c
ti
o
n
,
”
in
Pro
c
.
1
9
th
In
t.
Co
n
f.
Ge
o
in
f
o
rm
.
,
S
h
a
n
g
h
a
i,
Ch
i
n
a
,
p
p
.
1
-
4
,
2
0
1
1
.
[1
4
]
Y.
G
a
n
,
e
t
a
l.
,
“F
e
a
tu
re
Ex
trac
ti
o
n
Ba
se
d
M
u
lt
i
-
S
tru
c
t
u
re
M
a
n
if
o
ld
Em
b
e
d
d
i
n
g
f
o
r
Hy
p
e
rsp
e
c
tral
Re
m
o
te
S
e
n
sin
g
Im
a
g
e
Cla
ss
i
f
ica
ti
o
n
,
”
in
IEE
E
A
c
c
e
ss
,
v
o
l.
5
,
p
p
.
2
5
0
6
9
-
2
5
0
8
0
,
2
0
1
7
.
[1
5
]
F
.
L
u
o
,
e
t
a
l.
,
“
F
u
si
o
n
o
f
g
ra
p
h
e
m
b
e
d
d
in
g
a
n
d
s
p
a
rse
re
p
re
se
n
tatio
n
f
o
r
f
e
a
tu
re
e
x
trac
ti
o
n
a
n
d
c
l
a
ss
i_
c
a
ti
o
n
o
f
h
y
p
e
rsp
e
c
tral
i
m
a
g
e
r
y
,
”
Ph
o
t
o
g
ra
m.
En
g
.
Rem
o
te
S
e
n
s
.
,
v
o
l
.
83
,
n
o
.
1
,
p
p
.
3
7
-
4
6
,
2
0
1
7
.
[1
6
]
K.
R.
M
u
ll
e
r,
e
t
a
l.
,
“
A
n
in
tr
o
d
u
c
ti
o
n
t
o
k
e
rn
e
l
-
b
a
se
d
lea
rn
i
n
g
a
lg
o
rit
h
m
s,
”
in
IE
EE
T
r
a
n
sa
c
t
i
o
n
s
o
n
Ne
u
r
a
l
Ne
two
rk
s
,
v
o
l
.
12
,
n
o
.
2
,
p
p
.
1
8
1
-
2
0
1
,
2
0
0
1
.
[1
7
]
h
tt
p
:
//
ww
w
.
e
h
u
.
e
u
s/c
c
w
in
tco
/i
n
d
e
x
.
p
h
p
?
ti
tl
e
=
Hy
p
e
rsp
e
c
tral_
Re
m
o
te_
S
e
n
si
n
g
_
S
c
e
n
e
s
[1
8
]
Q.
Zh
a
n
g
,
e
t
a
l
.
,
“
A
u
to
m
a
ti
c
sp
a
ti
a
l
–
sp
e
c
tral
f
e
a
tu
re
se
lec
ti
o
n
f
o
r
h
y
p
e
rsp
e
c
tral
i
m
a
g
e
v
ia
d
isc
rim
in
a
ti
v
e
sp
a
rse
m
u
lt
i
m
o
d
a
l
lea
rn
in
g
,
”
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Ge
o
sc
ien
c
e
a
n
d
Rem
o
te S
e
n
si
n
g
,
v
o
l
.
53
,
n
o
.
1
,
p
p
.
261
-
2
7
9
,
2
0
1
5
.
[1
9
]
M
.
Zh
a
n
g
,
e
t
a
l
.
,
“
M
e
m
e
ti
c
a
l
g
o
rit
h
m
b
a
se
d
fe
a
tu
re
se
le
c
ti
o
n
f
o
r
h
y
p
e
r
sp
e
c
tral
i
m
a
g
e
s
c
las
si
f
ic
a
ti
o
n
,
”
2
0
1
7
IE
EE
Co
n
g
re
ss
o
n
Evo
lu
ti
o
n
a
ry
Co
m
p
u
ta
ti
o
n
(
CEC),
S
a
n
S
e
b
a
stia
n
,
p
p
.
4
9
5
-
5
0
2
,
2
0
1
7
.
[2
0
]
L
.
S
h
e
n
,
e
t
a
l.
,
“
Disc
rim
in
a
ti
v
e
G
a
b
o
r
F
e
a
tu
re
S
e
lec
ti
o
n
f
o
r
H
y
p
e
rsp
e
c
tral
I
m
a
g
e
Cla
ss
i
f
ic
a
ti
o
n
,
”
in
IEE
E
Ge
o
sc
ien
c
e
a
n
d
Rem
o
te S
e
n
si
n
g
L
e
tt
e
rs
,
v
o
l
.
10
,
n
o
.
1
,
p
p
.
2
9
-
3
3
,
2
0
1
3
.
[2
1
]
M
.
F
a
u
v
e
l,
e
t
a
l
.
,
“
F
a
st
F
o
rw
a
rd
F
e
a
tu
re
S
e
lec
ti
o
n
o
f
Hy
p
e
rsp
e
c
tral
Im
a
g
e
s
f
o
r
Clas
si
f
ica
ti
o
n
W
it
h
G
a
u
ss
ian
M
ix
tu
re
M
o
d
e
ls,
”
in
IEE
E
J
o
u
rn
a
l
o
f
S
e
lec
ted
T
o
p
ics
in
A
p
p
l
ied
Ea
rth
O
b
se
rv
a
ti
o
n
s
a
n
d
Rem
o
te
S
e
n
sin
g
,
v
o
l
.
8
,
n
o
.
6
,
p
p
.
2
8
2
4
-
2
8
3
1
,
2
0
1
5
.
[2
2
]
K.
T
a
n
,
e
t
a
l.
,
“
H
y
p
e
rsp
e
c
tral
I
m
a
g
e
Clas
si
f
ica
ti
o
n
Us
in
g
Ba
n
d
S
e
lec
ti
o
n
a
n
d
M
o
rp
h
o
l
o
g
ica
l
P
r
o
f
il
e
s,
”
in
IEE
E
J
o
u
rn
a
l
o
f
S
e
lec
ted
T
o
p
ics
in
Ap
p
li
e
d
Ea
rt
h
O
b
se
rv
a
ti
o
n
s
a
n
d
R
e
m
o
te S
e
n
si
n
g
,
v
o
l
.
7
,
n
o
.
1
,
p
p
.
4
0
-
4
8
,
2
0
1
4
.
[2
3
]
M
.
P
a
l
a
n
d
G
.
M
.
F
o
o
d
y
,
“
F
e
a
tu
re
S
e
lec
ti
o
n
f
o
r
Clas
sif
ica
ti
o
n
o
f
Hy
p
e
rsp
e
c
tr
a
l
Da
ta
b
y
S
V
M
,
”
i
n
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Ge
o
sc
ien
c
e
a
n
d
Rem
o
te S
e
n
si
n
g
,
v
o
l
.
48
,
n
o
.
5
,
p
p
.
2
2
9
7
-
2
3
0
7
,
2
0
1
0
.
[2
4
]
M
.
Im
a
n
i
a
n
d
H.
G
h
a
ss
e
m
i
a
n
,
“
F
a
st
f
e
a
tu
re
se
le
c
ti
o
n
m
e
th
o
d
s
f
o
r
c
las
sif
i
c
a
ti
o
n
o
f
h
y
p
e
rsp
e
c
tral
ima
g
e
s,
”
T
e
lec
o
mm
u
n
ica
ti
o
n
s (
IS
T
)
,
2
0
1
4
7
th
I
n
ter
n
a
ti
o
n
a
l
S
y
mp
o
siu
m
o
n
,
T
e
h
ra
n
,
p
p
.
7
8
-
83
,
2
0
1
4
.
[2
5
]
N
.
Zh
a
n
g
,
e
t
a
l.
,
“
A
n
En
h
a
n
c
e
d
K
-
Ne
a
re
st
Ne
ig
h
b
o
r
Clas
sif
ic
a
ti
o
n
M
e
t
h
o
d
Ba
se
d
o
n
M
a
x
im
a
l
Co
h
e
re
n
c
e
a
n
d
V
a
li
d
it
y
Ra
ti
n
g
s
,”
1
4
th
I
n
ter
n
a
ti
o
n
a
l
S
y
mp
o
si
u
m,
J
a
p
a
n
,
2
0
1
7
.
[2
6
]
A
.
W
o
o
d
ley
,
e
t
a
l.
,
“
Ef
f
i
c
ien
t
F
e
a
tu
re
S
e
lec
ti
o
n
a
n
d
Ne
a
re
st
Ne
ig
h
b
o
u
r
S
e
a
rc
h
f
o
r
H
y
p
e
rsp
e
c
tral
Im
a
g
e
Clas
sif
ic
a
ti
o
n
,
”
2
0
1
6
I
n
ter
n
a
t
io
n
a
l
Co
n
fer
e
n
c
e
o
n
Dig
it
a
l
Im
a
g
e
Co
mp
u
ti
n
g
:
T
e
c
h
n
iq
u
e
s
a
n
d
Ap
p
li
c
a
ti
o
n
s
(
DICTA
),
Go
ld
C
o
a
st,
QLD
,
p
p
.
1
-
8
,
2
0
1
6
.
[2
7
]
S
.
Ke
ll
e
r,
e
t
a
l.
,
“
I
n
v
e
stig
a
ti
o
n
o
f
th
e
im
p
a
c
t
o
f
d
im
e
n
sio
n
a
li
ty
re
d
u
c
ti
o
n
a
n
d
f
e
a
tu
re
se
lec
ti
o
n
o
n
th
e
c
las
si
f
ica
ti
o
n
o
f
h
y
p
e
rsp
e
c
tral
En
M
A
P
d
a
ta,
”
2
0
1
6
8
t
h
W
o
rk
sh
o
p
o
n
Hy
p
e
rs
p
e
c
tra
l
Ima
g
e
a
n
d
S
ig
n
a
l
Pro
c
e
ss
in
g
:
Evo
l
u
ti
o
n
i
n
Rem
o
te S
e
n
si
n
g
(
W
HIS
PE
RS
),
L
o
s A
n
g
e
les
,
CA
,
p
p
.
1
-
5
,
2
0
1
6
.
[2
8
]
P
.
J.
Hs
ieh
,
e
t
a
l.
,
“
A
n
o
n
li
n
e
a
r
f
e
a
tu
re
se
lec
ti
o
n
m
e
th
o
d
b
a
se
d
o
n
k
e
rn
e
l
se
p
a
ra
b
il
it
y
m
e
a
su
re
f
o
r
h
y
p
e
rsp
e
c
tral
im
a
g
e
c
las
si
f
ica
ti
o
n
,
”
2
0
1
5
IEE
E
In
ter
n
a
t
io
n
a
l
Ge
o
sc
ien
c
e
a
n
d
Rem
o
te
S
e
n
sin
g
S
y
mp
o
siu
m
(
IGARS
S
),
M
il
a
n
,
p
p
.
4
6
1
-
464
,
2
0
1
5
.
[2
9
]
C.
A
.
K
e
n
d
z
io
ra
,
e
t
a
l.
,
“
A
d
v
a
n
c
e
s
in
sta
n
d
o
f
f
d
e
te
c
ti
o
n
o
f
trac
e
e
x
p
lo
siv
e
s
b
y
in
f
r
a
re
d
p
h
o
to
-
th
e
r
m
a
l
i
m
a
g
in
g
,
”
Pro
c
e
e
d
in
g
s
S
PIE
7
6
6
4
,
De
tec
ti
o
n
a
n
d
S
e
n
si
n
g
o
f
M
in
e
s,
Exp
l
o
siv
e
Ob
jec
ts,
a
n
d
Ob
sc
u
re
d
T
a
r
g
e
ts
XV
,
2
0
1
0
.
Evaluation Warning : The document was created with Spire.PDF for Python.