I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
pu
t
er
E
ng
ineering
(
I
J
E
CE
)
Vo
l.
15
,
No
.
2
,
A
p
r
il
20
25
,
p
p
.
1
5
9
3
~
1
6
0
1
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
15
i
2
.
pp
1
5
9
3
-
1
6
0
1
1593
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
Two
-
sca
le decom
po
sitio
n and
deep
learning
f
usio
n
f
o
r vis
ible
a
nd infr
a
red
ima
g
es
Ruha
n B
ev
i A
za
d
1
,
H
a
ri
Un
nik
rish
na
n
2
,
L
o
k
esh
G
o
pin
a
t
h
1
1
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
o
n
i
c
s a
n
d
C
o
m
mu
n
i
c
a
t
i
o
n
En
g
i
n
e
e
r
i
n
g
,
S
R
M
I
n
st
i
t
u
t
e
o
f
S
c
i
e
n
c
e
a
n
d
Te
c
h
n
o
l
o
g
y
,
K
a
t
t
a
n
k
u
l
a
t
h
u
r
,
I
n
d
i
a
2
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
o
n
i
c
s a
n
d
C
o
m
mu
n
i
c
a
t
i
o
n
En
g
i
n
e
e
r
i
n
g
,
S
a
v
e
e
t
h
a
E
n
g
i
n
e
e
r
i
n
g
C
o
l
l
e
g
e
,
C
h
e
n
n
a
i
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ap
r
2
2
,
2
0
2
4
R
ev
is
ed
No
v
2
6
,
2
0
2
4
Acc
ep
ted
Dec
2
,
2
0
2
4
Th
e
p
a
p
e
r
f
o
c
u
se
s
o
n
t
h
e
fu
si
o
n
o
f
v
isi
b
le
a
n
d
i
n
fra
re
d
ima
g
e
s
to
g
e
n
e
ra
te
c
o
m
p
o
site
ima
g
e
s
th
a
t
p
re
se
rv
e
b
o
th
t
h
e
th
e
rm
a
l
ra
d
iatio
n
i
n
f
o
rm
a
ti
o
n
fr
o
m
th
e
in
fra
re
d
s
p
e
c
tru
m
a
n
d
t
h
e
d
e
t
a
il
e
d
tex
tu
re
fr
o
m
th
e
v
isib
le
sp
e
c
tru
m
.
Th
e
p
ro
p
o
se
d
a
p
p
r
o
a
c
h
c
o
m
b
i
n
e
s
trad
it
io
n
a
l
m
e
th
o
d
s,
su
c
h
a
s
two
-
sc
a
le
d
e
c
o
m
p
o
siti
o
n
,
with
d
e
e
p
lea
rn
i
n
g
tec
h
n
iq
u
e
s,
sp
e
c
ifi
c
a
ll
y
e
m
p
l
o
y
i
n
g
a
n
a
u
to
e
n
c
o
d
e
r
a
rc
h
it
e
c
tu
re
.
T
h
e
s
o
u
rc
e
ima
g
e
s
a
re
su
b
jec
ted
to
two
-
sc
a
le
d
e
c
o
m
p
o
siti
o
n
,
w
h
ich
e
x
trac
ts h
i
g
h
-
fre
q
u
e
n
c
y
d
e
tail
a
n
d
l
o
w
-
fre
q
u
e
n
c
y
b
a
se
in
fo
rm
a
ti
o
n
.
Ad
d
it
io
n
a
ll
y
,
a
n
a
lg
o
rit
h
m
ic u
n
ra
v
e
l
li
n
g
tec
h
n
iq
u
e
e
sta
b
li
sh
e
s
a
lo
g
ica
l
c
o
n
n
e
c
ti
o
n
b
e
twe
e
n
d
e
e
p
n
e
u
ra
l
n
e
two
r
k
s
a
n
d
tra
d
it
i
o
n
a
l
sig
n
a
l
p
ro
c
e
ss
in
g
a
lg
o
rit
h
m
s.
Th
e
m
o
d
e
l
c
o
n
sists
o
f
tw
o
e
n
c
o
d
e
rs
fo
r
d
e
c
o
m
p
o
siti
o
n
a
n
d
a
d
e
c
o
d
e
r
a
fte
r
th
e
u
n
ra
v
e
ll
i
n
g
o
p
e
ra
ti
o
n
.
Du
ri
n
g
tes
ti
n
g
,
a
fu
sio
n
la
y
e
r
m
e
rg
e
s
th
e
d
e
c
o
m
p
o
se
d
fe
a
tu
re
m
a
p
s,
a
n
d
th
e
d
e
c
o
d
e
r
g
e
n
e
ra
tes
th
e
fu
se
d
ima
g
e
.
Ev
a
l
u
a
ti
o
n
m
e
tri
c
s
i
n
c
lu
d
in
g
e
n
tro
p
y
,
a
v
e
ra
g
e
g
ra
d
ien
t
,
sp
a
ti
a
l
fre
q
u
e
n
c
y
a
n
d
sta
n
d
a
rd
d
e
v
iati
o
n
a
re
e
m
p
lo
y
e
d
t
o
su
b
jec
ti
v
e
l
y
a
ss
e
ss
fu
sio
n
q
u
a
li
ty
.
T
h
e
p
ro
p
o
se
d
a
p
p
r
o
a
c
h
d
e
m
o
n
stra
tes
p
ro
m
ise
f
o
r
e
ffe
c
ti
v
e
l
y
c
o
m
b
in
in
g
v
isib
le
a
n
d
i
n
fra
re
d
ima
g
e
ry
f
o
r
v
a
rio
u
s
a
p
p
li
c
a
ti
o
n
s.
K
ey
w
o
r
d
s
:
Alg
o
r
ith
m
u
n
r
av
ellin
g
Dee
p
lear
n
in
g
Near
-
in
f
r
ar
e
d
im
ag
e
T
r
ad
itio
n
al
m
eth
o
d
T
wo
-
s
ca
le
d
ec
o
m
p
o
s
itio
n
Vis
ib
le
im
ag
e
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
R
u
h
an
B
ev
i A
za
d
Dep
ar
tm
en
t o
f
E
lectr
o
n
ics an
d
C
o
m
m
u
n
icatio
n
E
n
g
in
ee
r
i
n
g
,
SR
M
I
n
s
titu
te
o
f
Scien
ce
an
d
T
ec
h
n
o
lo
g
y
Kattan
k
u
lath
u
r
,
C
h
e
n
g
alp
attu
,
6
0
3
2
0
3
,
T
am
il Na
d
u
,
I
n
d
ia
E
m
ail:
r
u
h
an
b
@
s
r
m
is
t.e
d
u
.
in
1.
I
NT
RO
D
UCT
I
O
N
I
n
th
e
f
ield
o
f
im
ag
e
p
r
o
ce
s
s
in
g
r
esear
ch
,
im
ag
e
f
u
s
io
n
is
an
em
er
g
in
g
to
p
ic.
Ad
o
p
ti
n
g
s
im
ilar
m
eth
o
d
o
l
o
g
ies
an
d
s
tr
ateg
ies
,
en
h
an
ce
s
th
e
ef
f
ec
tiv
e
n
ess
,
in
ter
p
r
etab
ilit
y
,
a
n
d
r
ep
r
o
d
u
cib
ilit
y
o
f
im
a
g
e
f
u
s
io
n
.
Ad
d
itio
n
ally
,
ev
alu
ati
n
g
m
o
d
els
o
n
b
e
n
ch
m
ar
k
d
atasets
an
d
m
ak
in
g
co
d
e
a
n
d
d
ata
o
p
en
l
y
av
ailab
le
will
co
n
tr
ib
u
te
to
th
e
a
d
v
an
ce
m
en
t
o
f
r
esear
ch
i
n
th
e
f
ield
o
f
r
em
o
te
s
en
s
in
g
[
1
]
.
I
n
co
r
p
o
r
atin
g
in
s
ig
h
ts
f
r
o
m
th
e
b
ase
p
ap
er
in
v
o
lv
es
lev
er
ag
in
g
u
n
s
u
p
er
v
is
ed
lear
n
in
g
t
ec
h
n
iq
u
es
an
d
ad
v
an
ce
d
l
o
s
s
f
u
n
ctio
n
s
f
o
r
im
ag
e
f
u
s
io
n
.
E
n
h
an
cin
g
m
o
d
el
in
te
r
p
r
etab
ilit
y
an
d
e
v
alu
atin
g
its
p
er
f
o
r
m
an
ce
o
n
b
en
c
h
m
ar
k
d
atasets
ar
e
cr
u
cial
s
tep
s
.
Ad
d
itio
n
ally
,
s
h
ar
in
g
co
d
e
an
d
d
ata
e
n
s
u
r
es
tr
an
s
p
ar
en
cy
a
n
d
r
ep
r
o
d
u
cib
ilit
y
,
f
ac
ilit
atin
g
f
u
r
th
e
r
ad
v
an
ce
m
e
n
ts
in
th
e
f
ield
[
2
]
.
T
h
e
r
ef
in
em
e
n
t
f
u
s
io
n
a
p
p
r
o
ac
h
ac
h
iev
es
s
u
p
er
io
r
p
er
f
o
r
m
an
ce
in
ter
m
s
o
f
im
ag
e
q
u
ality
,
tar
g
et
r
eg
io
n
p
r
eser
v
atio
n
an
d
ef
f
icien
c
y
[
3
]
.
E
v
alu
atin
g
th
e
p
er
f
o
r
m
an
ce
an
d
v
alid
atin
g
it
th
r
o
u
g
h
ex
p
er
im
en
ts
,
en
h
an
ce
s
th
e
ef
f
ec
tiv
en
ess
an
d
r
o
b
u
s
tn
ess
o
f
im
ag
e
f
u
s
io
n
.
c
o
n
s
id
e
r
in
g
th
e
s
im
ilar
ities
an
d
r
ef
i
n
in
g
a
p
p
r
o
ac
h
an
d
ad
d
r
ess
in
g
s
p
ec
if
ic
ch
allen
g
es
i
n
in
f
r
ar
e
d
an
d
v
is
ib
le
im
ag
e
f
u
s
io
n
[
4
]
.
T
h
er
e
ar
e
p
o
ten
tial
ar
ea
s
f
o
r
im
p
r
o
v
em
en
t
an
d
in
n
o
v
atio
n
in
m
eth
o
d
o
lo
g
ies.
L
ev
er
a
g
in
g
i
n
s
ig
h
ts
f
r
o
m
ea
ch
ap
p
r
o
ac
h
ca
n
co
n
tr
ib
u
te
to
th
e
d
e
v
elo
p
m
en
t
o
f
m
o
r
e
ef
f
ec
tiv
e
an
d
ad
v
an
ce
d
im
a
g
e
f
u
s
io
n
tech
n
iq
u
es
[
5
]
–
[
8
]
.
T
h
e
co
n
ce
p
t
o
f
a
u
n
i
f
ied
f
u
s
io
n
f
r
am
ewo
r
k
,
ad
ap
tiv
e
in
f
o
r
m
atio
n
p
r
eser
v
atio
n
,
m
itig
ati
o
n
o
f
d
ee
p
lear
n
in
g
lim
itatio
n
s
,
an
d
th
e
u
s
e
o
f
b
en
ch
m
ar
k
d
atasets
,
en
h
an
ce
s
th
e
e
f
f
ec
tiv
en
ess
,
v
er
s
atility
,
an
d
ev
alu
atio
n
o
f
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
2
,
Ap
r
il
20
25
:
1
5
9
3
-
1
6
0
1
1594
im
ag
e
f
u
s
io
n
.
Ad
d
itio
n
ally
,
co
n
s
id
er
in
g
th
e
s
im
ilar
ities
an
d
d
i
f
f
er
en
ce
s
b
etwe
en
a
p
p
r
o
ac
h
es
ca
n
p
r
o
v
id
e
v
alu
ab
le
p
e
r
s
p
ec
tiv
es
f
o
r
r
ef
i
n
in
g
f
u
s
io
n
m
o
d
els
an
d
ad
d
r
ess
in
g
s
p
ec
if
ic
ch
allen
g
es
in
in
f
r
ar
ed
an
d
v
is
ib
le
im
ag
e
f
u
s
io
n
[
9
]
,
[
1
0
]
.
Sp
ar
s
e
r
eg
u
lar
izatio
n
[
1
1
]
,
[
1
2
]
,
d
ictio
n
ar
y
lear
n
in
g
[
1
3
]
,
[
1
4
]
,
alter
n
atin
g
o
p
tim
izatio
n
[
1
5
]
,
a
n
d
p
er
f
o
r
m
an
ce
ev
alu
atio
n
ag
ain
s
t
s
tate
-
of
-
th
e
-
ar
t
m
eth
o
d
s
,
en
h
a
n
c
e
th
e
ef
f
ec
tiv
en
ess
an
d
ef
f
icien
cy
o
f
in
f
r
a
r
ed
an
d
v
is
ib
le
im
ag
e
f
u
s
io
n
.
L
ev
er
ag
in
g
tech
n
iq
u
es
f
r
o
m
d
i
v
er
s
e
d
o
m
ain
s
,
s
u
ch
as
h
y
p
er
s
p
ec
tr
al
an
d
m
u
ltis
p
ec
tr
al
im
ag
e
f
u
s
io
n
,
ca
n
lead
t
o
i
n
n
o
v
ativ
e
s
o
lu
tio
n
s
an
d
im
p
r
o
v
em
en
ts
in
f
u
s
io
n
alg
o
r
ith
m
s
[
1
6
]
.
A
m
u
ltis
ca
le
f
u
s
io
n
s
tr
ateg
y
,
an
d
e
n
co
d
er
-
f
u
s
io
n
s
tr
ateg
y
-
d
ec
o
d
er
f
r
am
ewo
r
k
,
en
h
an
ce
th
e
p
er
f
o
r
m
an
ce
an
d
r
o
b
u
s
tn
ess
o
f
in
f
r
a
r
ed
an
d
v
is
ib
le
im
ag
e
f
u
s
io
n
[
1
7
]
.
C
o
m
p
ar
ativ
e
an
aly
s
is
o
f
f
u
s
io
n
tech
n
iq
u
es,
ev
al
u
atio
n
m
etr
ic
s
,
ef
f
icien
cy
,
a
n
d
s
u
itab
ilit
y
f
o
r
r
ea
l
a
p
p
licatio
n
s
,
a
n
d
th
e
ef
f
ec
tiv
en
ess
an
d
ap
p
licab
ilit
y
o
f
im
a
g
e
f
u
s
io
n
r
esear
ch
,
ev
en
if
th
ey
o
p
e
r
ate
in
a
d
if
f
e
r
en
t
d
o
m
ain
th
a
n
m
ed
ical
im
ag
in
g
.
ca
n
g
u
id
e
th
e
ev
alu
atio
n
p
r
o
ce
s
s
an
d
en
s
u
r
e
th
e
r
eliab
ilit
y
o
f
th
e
f
u
s
io
n
ap
p
r
o
ac
h
[
1
8
]
.
A
s
y
m
m
etr
ic
en
co
d
er
-
d
ec
o
d
er
a
r
ch
itectu
r
e
with
r
esi
d
u
al
b
lo
ck
s
,
atten
tio
n
m
ec
h
an
i
s
m
s
,
an
d
s
ep
ar
atio
n
o
f
tr
ain
in
g
an
d
f
u
s
io
n
s
tag
es,
en
h
an
ce
s
th
e
p
e
r
f
o
r
m
an
ce
a
n
d
ef
f
icien
cy
o
f
im
ag
e
f
u
s
io
n
[
1
9
]
.
T
h
e
u
s
e
o
f
g
en
er
ativ
e
a
d
v
er
s
ar
ial
n
etwo
r
k
(
GAN)
with
m
u
lticlas
s
if
icatio
n
co
n
s
tr
ain
ts
,
co
n
ten
t
lo
s
s
m
ec
h
an
is
m
s
,
an
d
co
m
p
r
e
h
en
s
iv
e
ev
alu
atio
n
m
eth
o
d
o
lo
g
ies,
en
h
an
ce
s
th
e
p
er
f
o
r
m
an
ce
a
n
d
ef
f
ec
tiv
en
ess
o
f
im
ag
e
f
u
s
io
n
[
2
0
]
.
Ad
ap
tiv
e
en
h
an
ce
m
e
n
t
tech
n
iq
u
es,
h
y
b
r
id
d
ec
o
m
p
o
s
itio
n
m
o
d
els,
co
u
p
led
d
ictio
n
ar
y
-
b
ased
f
u
s
io
n
,
an
d
n
o
v
el
f
u
s
io
n
s
ch
em
es
[
2
1
]
.
T
h
e
u
s
e
o
f
r
esid
u
al
n
etwo
r
k
a
r
ch
itectu
r
es,
in
n
o
v
ativ
e
lo
s
s
f
u
n
ctio
n
s
,
an
d
two
-
s
tag
e
tr
ain
in
g
s
tr
ateg
ies,
en
h
an
ce
s
th
e
p
er
f
o
r
m
a
n
ce
an
d
ef
f
icien
cy
o
f
im
ag
e
f
u
s
io
n
.
T
h
e
em
p
h
asis
o
n
task
-
s
p
ec
if
ic
f
u
s
io
n
s
tr
ateg
ies
a
n
d
th
e
a
d
o
p
tio
n
o
f
r
esid
u
al
n
etwo
r
k
a
r
ch
itectu
r
es,
ca
n
g
u
id
e
t
h
e
r
ef
in
em
en
t
o
f
th
e
f
u
s
io
n
ap
p
r
o
ac
h
to
b
etter
ad
ap
t
to
d
iv
er
s
e
f
u
s
io
n
task
s
[
2
2
]
.
Utilizin
g
u
n
s
u
p
er
v
is
ed
en
d
-
to
-
en
d
n
etwo
r
k
ar
ch
itectu
r
es,
d
esig
n
in
g
tailo
r
ed
l
o
s
s
f
u
n
ctio
n
s
,
an
d
im
p
lem
e
n
tin
g
c
o
n
v
o
lu
ti
o
n
al
lay
e
r
d
ec
o
m
p
o
s
itio
n
n
etwo
r
k
s
,
ca
n
en
h
an
ce
p
er
f
o
r
m
an
ce
an
d
e
f
f
ec
tiv
en
ess
[
2
3
]
.
T
h
e
co
m
p
r
eh
en
s
iv
en
ess
an
d
ef
f
ec
tiv
en
ess
o
f
th
ese
m
eth
o
d
s
s
h
o
u
ld
b
e
e
n
h
an
ce
d
.
T
h
is
in
clu
d
es
lev
er
ag
in
g
h
is
to
r
ical
co
n
tex
t,
e
x
p
lo
r
in
g
v
ar
io
u
s
f
u
s
io
n
tech
n
iq
u
es,
in
c
o
r
p
o
r
ati
n
g
r
ig
o
r
o
u
s
ev
alu
atio
n
p
r
o
ce
s
s
es,
an
d
co
n
s
id
er
in
g
p
r
ac
tical
ap
p
licatio
n
s
an
d
f
u
t
u
r
e
p
r
o
s
p
ec
ts
[
2
4
]
.
Fig
u
r
e
1
s
h
o
ws th
e
p
r
o
p
o
s
ed
f
u
s
io
n
m
o
d
el
wh
ich
c
o
n
s
is
ts
o
f
f
o
u
r
s
tag
es:
a.
I
n
p
u
t so
u
r
ce
s
: I
n
f
r
a
r
ed
im
a
g
es a
n
d
v
is
ib
le
im
ag
es a
r
e
f
ed
in
t
o
th
e
m
o
d
el
b.
Dec
o
m
p
o
s
itio
n
: T
h
e
im
a
g
es a
r
e
d
ec
o
m
p
o
s
ed
in
to
co
n
s
titu
en
t f
ea
tu
r
es
c.
Featu
r
e
f
u
s
io
n
: E
x
tr
ac
te
d
f
ea
t
u
r
es a
r
e
f
u
s
ed
t
o
en
h
a
n
ce
in
f
o
r
m
atio
n
d.
I
m
ag
e
f
u
s
io
n
: T
h
e
f
u
s
ed
f
ea
tu
r
es a
r
e
f
u
s
ed
to
g
eth
er
to
g
en
er
ate
th
e
f
in
al
im
ag
e
T
h
e
r
em
ai
n
in
g
s
ec
tio
n
s
o
f
th
i
s
wo
r
k
ar
e
s
tr
u
ct
u
r
ed
as
f
o
llo
ws.
Sectio
n
2
p
r
o
v
id
es
a
co
m
p
r
eh
en
s
iv
e
ex
p
lan
atio
n
o
f
th
e
m
eth
o
d
o
lo
g
y
th
at
h
as
b
ee
n
d
ev
is
ed
to
i
n
teg
r
ate
n
ea
r
-
i
n
f
r
ar
e
d
an
d
v
is
ib
le
im
ag
es
an
d
th
e
f
r
am
ewo
r
k
is
d
is
cu
s
s
ed
in
s
ec
tio
n
3
.
A
q
u
an
titativ
e
an
d
q
u
a
litativ
e
ev
alu
atio
n
o
f
th
e
alg
o
r
ith
m
is
p
r
o
v
id
ed
in
s
ec
tio
n
4
,
an
d
it is
co
n
clu
d
e
d
i
n
s
ec
tio
n
5
.
Fig
u
r
e
1
.
Pro
p
o
s
ed
f
u
s
io
n
m
o
d
el
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
Tw
o
-
s
ca
le
d
ec
o
mp
o
s
itio
n
a
n
d
d
ee
p
le
a
r
n
in
g
fu
s
io
n
fo
r
visi
b
le
a
n
d
in
fr
a
r
ed
ima
g
es
(
R
u
h
a
n
B
ev
i A
z
a
d
)
1595
2.
P
RO
P
O
SE
D
M
E
T
H
O
D
2
.
1
.
T
wo
-
s
ca
le
deco
m
po
s
it
io
n m
e
t
ho
d
T
h
r
o
u
g
h
o
u
t
th
e
y
ea
r
s
,
n
u
m
er
o
u
s
f
u
s
io
n
alg
o
r
ith
m
s
h
av
e
b
ee
n
p
r
o
p
o
s
ed
,
ea
ch
aim
i
n
g
to
en
h
an
ce
th
e
q
u
ality
an
d
lev
el
o
f
in
f
o
r
m
ati
o
n
in
im
ag
es.
T
h
e
p
r
im
ar
y
o
b
jectiv
e
o
f
two
-
s
ca
le
d
ec
o
m
p
o
s
itio
n
is
to
p
ar
titi
o
n
th
e
o
r
ig
in
al
im
ag
e
in
to
a
s
er
ies
o
r
en
s
em
b
le
o
f
im
ag
es,
ea
ch
o
f
wh
ich
h
ig
h
lig
h
ts
a
s
p
ec
if
ic
attr
ib
u
te
o
r
ch
ar
ac
ter
is
tic.
I
n
s
tead
o
f
d
ir
ec
tly
m
er
g
in
g
th
e
s
o
u
r
ce
im
ag
es,
th
is
d
ec
o
m
p
o
s
itio
n
m
e
th
o
d
is
em
p
l
o
y
ed
.
Dec
o
m
p
o
s
itio
n
p
r
io
r
itizes
th
e
f
u
s
io
n
o
f
th
e
d
ec
o
n
s
tr
u
cted
i
m
ag
es
r
ath
er
t
h
an
m
a
k
in
g
d
ir
ec
t
m
o
d
if
icatio
n
s
to
th
e
s
o
u
r
ce
im
ag
es.
A
m
o
r
e
in
tr
icate
an
d
co
m
p
r
eh
en
s
iv
e
f
u
s
ed
im
ag
e
is
g
en
er
ate
d
b
y
m
er
g
in
g
th
e
u
n
iq
u
e
in
f
o
r
m
atio
n
co
n
s
er
v
ed
in
ea
c
h
d
ec
o
n
s
tr
u
cted
im
ag
e.
T
h
is
p
ap
er
ex
p
lo
r
es
two
d
is
tin
ct
d
ee
p
lear
n
in
g
m
et
h
o
d
o
lo
g
ies
with
in
th
e
f
r
am
ewo
r
k
o
f
m
u
ltis
ca
le
d
ec
o
m
p
o
s
itio
n
f
o
r
f
u
s
io
n
.
Alt
h
o
u
g
h
th
e
f
ir
s
t
m
eth
o
d
is
u
s
e
f
u
l
f
o
r
ex
tr
ac
tin
g
g
en
er
al
f
ea
t
u
r
es,
it
r
elies
o
n
a
p
r
etr
ain
ed
d
ee
p
n
eu
r
al
n
etwo
r
k
th
at
m
ay
n
o
t
b
e
o
p
tim
ize
d
b
ec
au
s
e
o
f
th
e
in
t
r
icate
n
a
tu
r
e
o
f
m
u
ltis
ca
le
d
ec
o
m
p
o
s
itio
n
task
s
.
T
h
is
c
o
u
ld
r
esu
lt
i
n
s
u
b
o
p
tim
al
f
u
s
io
n
o
u
tc
o
m
es
d
u
e
to
th
e
n
etwo
r
k
'
s
lim
ited
ad
ap
tab
ilit
y
in
v
a
r
io
u
s
b
r
ea
k
d
o
wn
s
ce
n
ar
io
s
.
On
th
e
o
th
er
h
an
d
,
t
h
e
s
ec
o
n
d
m
eth
o
d
u
tili
ze
s
an
a
u
to
en
co
d
er
.
Au
to
en
co
d
e
r
s
ar
e
p
ar
ticu
lar
ly
well
-
s
u
ited
f
o
r
task
s
th
at
r
e
q
u
ir
e
ac
cu
r
ate
p
r
eser
v
atio
n
o
f
i
n
f
o
r
m
atio
n
,
as
t
h
ey
ar
e
s
p
ec
if
ically
d
esig
n
ed
f
o
r
ex
tr
ac
tin
g
f
ea
tu
r
es
an
d
r
ec
o
n
s
tr
u
ctin
g
im
a
g
es.
An
au
t
o
en
co
d
er
'
s
en
co
d
in
g
p
r
o
ce
s
s
d
ec
o
m
p
o
s
es
im
ag
es
in
to
f
ea
tu
r
e
m
ap
s
,
e
n
ab
lin
g
t
h
e
ef
f
icien
t
ca
p
tu
r
e
o
f
in
f
o
r
m
atio
n
at
d
i
f
f
er
en
t
s
ca
les.
An
ad
v
an
tag
e
o
f
u
s
i
n
g
an
au
to
en
c
o
d
er
-
b
ased
ap
p
r
o
ac
h
,
as
o
p
p
o
s
ed
to
a
p
r
e
tr
ain
ed
d
ee
p
n
eu
r
al
n
etwo
r
k
,
is
its
s
u
p
er
io
r
s
u
ita
b
ilit
y
f
o
r
m
u
ltis
ca
le
d
ec
o
m
p
o
s
itio
n
task
s
d
u
e
to
its
in
h
er
en
t
ad
ap
tab
ilit
y
an
d
co
n
tex
tu
al
r
elev
an
ce
.
2
.
2
.
O
pti
m
iza
t
io
n
m
o
del
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
u
tili
ze
s
o
p
tim
izatio
n
tech
n
iq
u
es
to
d
ec
o
m
p
o
s
e
an
in
p
u
t
im
ag
e
i
n
to
a
b
ase
im
ag
e
an
d
a
d
etail
im
ag
e.
T
h
e
ac
q
u
is
itio
n
o
f
t
h
e
b
asis
i
m
ag
e
ca
n
b
e
ac
h
iev
ed
b
y
ad
d
r
ess
in
g
th
e
is
s
u
e
o
f
lo
w
-
f
r
eq
u
e
n
cy
b
ac
k
g
r
o
u
n
d
in
f
o
r
m
atio
n
.
∗
=
a
r
g
min
2
‖
−
‖
2
+
∑
‖
∗
‖
2
=
1
(
1
)
wh
er
e
∗
is
th
e
d
is
in
teg
r
ated
b
a
s
e
im
ag
e
,
(m
=
1
,
2
….
,
n
)
ar
e
h
ig
h
p
ass
f
ilter
s
,
is
th
e
in
p
u
t
im
ag
e,
∗
is
th
e
co
n
v
o
lu
tio
n
o
p
er
atio
n
,
is
r
ep
r
esen
ts
th
e
tu
n
in
g
h
y
p
er
p
ar
a
m
eter
an
d
∑
‖
∗
‖
2
=
1
is
u
s
ed
to
r
ed
u
ce
th
e
h
ig
h
f
r
eq
u
en
cy
o
f
.
No
w,
r
ep
r
esen
ts
th
e
d
etailed
im
ag
e
a
n
d
m
ea
n
s
h
i
g
h
f
r
e
q
u
en
c
y
q
u
ality
/tex
tu
r
e
a
n
d
co
l
o
u
r
p
r
o
g
r
ess
io
n
an
d
is
g
iv
e
n
b
y
(
2
)
.
∗
=
2
‖
−
‖
2
+
∑
‖
∗
‖
2
(
2
)
h
er
e
,
(
m
=1
,
2
….
,
n
)
ar
e
lo
w
p
ass
f
ilter
s
,
an
d
is
ag
ain
a
tu
n
in
g
h
y
p
e
r
p
ar
a
m
eter
.
T
o
ac
c
o
u
n
t
f
o
r
th
ei
r
f
lex
ib
ilit
y
,
th
e
weig
h
ts
f
o
r
a
ce
n
tr
al
p
ix
el
m
u
s
t b
e
n
o
r
m
alize
d
to
s
u
m
to
1
.
T
h
er
e
f
o
r
e,
th
e
b
a
s
e
f
ea
tu
r
e
m
ap
is
=
−
∑
(
)
=
1
∗
(
∗
)
−
(
−
)
(
3
)
wh
er
e
(
)
r
ep
r
esen
ts
th
e
k
er
n
el
o
f
r
o
tated
b
y
1
8
0
°
an
d
is
th
e
s
tep
s
ize.
2
.
3
.
Unra
v
elling
a
l
g
o
rit
hm
T
h
e
“u
n
r
a
v
ellin
g
”
alg
o
r
ith
m
is
a
r
ec
en
tly
d
ev
elo
p
ed
m
et
h
o
d
th
at
p
r
o
v
i
d
es
an
ap
p
ea
lin
g
p
r
o
ce
s
s
f
o
r
d
esig
n
in
g
d
ee
p
n
e
u
r
al
n
etwo
r
k
s
b
ased
o
n
m
o
d
els.
Alg
o
r
i
th
m
u
n
r
o
llin
g
is
th
e
p
r
o
ce
s
s
o
f
tr
an
s
f
o
r
m
in
g
a
r
ep
etitiv
e
alg
o
r
ith
m
in
to
a
d
e
ep
n
eu
r
al
n
etwo
r
k
(
DNN)
b
y
ex
p
an
d
i
n
g
its
co
m
p
u
tatio
n
al
g
r
ap
h
.
T
h
is
en
a
b
les
th
e
tr
ain
in
g
o
f
p
r
ed
ef
in
e
d
h
y
p
er
p
ar
am
eter
s
an
d
u
n
k
n
o
wn
co
ef
f
icien
ts
in
a
co
m
p
r
eh
e
n
s
iv
e
way
.
T
h
e
b
ase
co
n
v
o
l
u
tio
n
lay
er
an
d
d
etail
c
o
n
v
o
lu
ti
o
n
lay
er
ar
e
r
e
p
lace
d
with
f
ilter
s
an
d
as (
4
)
.
=
−
[
2
(
1
(
)
)
−
(
−
)
]
(
4
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
2
,
Ap
r
il
20
25
:
1
5
9
3
-
1
6
0
1
1596
wh
er
e
(
m
=1
,
2
)
wh
ich
d
e
n
o
t
es
th
e
k
er
n
el
s
ize.
Fu
r
th
er
m
o
r
e,
we
s
et
th
e
1
k
er
n
el
e
q
u
a
l
to
2
wh
er
e
th
is
m
ad
e
a
1
8
0
°
tu
r
n
.
Similar
ly
,
th
e
d
etailed
f
ea
tu
r
e
m
ap
u
p
d
atin
g
p
r
o
ce
d
u
r
e
is
ca
r
r
ied
o
u
t
as (
5
)
:
=
−
[
2
(
1
(
)
)
−
(
−
)
]
(
5
)
wh
er
e,
an
d
ar
e
p
r
e
d
ef
in
e
d
h
y
p
er
p
ar
am
eter
s
a
n
d
an
d
ar
e
s
tep
s
izes.
3.
F
RAM
E
WO
RK
T
h
e
d
etail
an
d
b
ase
im
ag
es
ar
e
r
eg
ar
d
ed
as
f
ea
tu
r
e
m
ap
s
o
b
tain
ed
f
r
o
m
th
e
s
o
u
r
ce
i
m
ag
e.
T
h
is
ap
p
r
o
ac
h
in
v
o
lv
es
o
r
g
a
n
izin
g
N
d
etail
c
o
n
v
o
lu
tio
n
lay
er
(
DC
L
)
an
d
b
ase
co
n
v
o
lu
tio
n
l
ay
er
(
B
C
L
)
as
two
en
co
d
er
s
.
T
h
e
o
b
jectiv
e
is
to
r
ep
r
o
d
u
ce
th
e
r
e
p
etitiv
e
p
r
o
ce
d
u
r
e
o
f
c
o
n
v
e
n
tio
n
al
o
p
tim
izatio
n
m
o
d
els
a
n
d
ex
tr
ac
t
f
u
n
d
am
e
n
tal
an
d
in
tr
i
ca
te
f
ea
tu
r
e
m
ap
s
.
Su
b
s
eq
u
en
tly
,
a
s
u
p
p
lem
en
tar
y
d
ec
o
d
er
is
g
en
er
ated
u
s
in
g
in
p
u
ts
th
at
in
clu
d
e
th
e
s
u
m
m
atio
n
o
f
two
d
ec
o
n
s
tr
u
cted
f
ea
tu
r
e
m
a
p
s
.
T
h
e
r
esu
lt
o
f
th
is
d
ec
o
d
er
is
th
e
r
ec
o
n
s
tr
u
cted
s
o
u
r
ce
im
ag
e.
Fig
u
r
e
2
illu
s
tr
ates
th
e
n
etwo
r
k
ar
ch
itectu
r
e
d
u
r
in
g
t
h
e
tr
ain
in
g
p
h
ase
an
d
Fig
u
r
e
3
d
e
p
icts
a
s
in
g
le
B
C
L
,
wh
ile
a
DC
L
h
as
a
s
im
ilar
s
tr
u
ctu
r
e
b
u
t
d
is
tin
ct
p
ar
a
m
eter
s
.
T
h
e
n
u
m
b
er
o
f
in
p
u
t
an
d
o
u
tp
u
t
ch
a
n
n
els
f
o
r
th
e
f
ir
s
t
co
n
v
o
l
u
tio
n
u
n
its
1
an
d
1
is
(
1
,
H)
.
T
h
e
s
ec
o
n
d
co
n
v
o
l
u
tio
n
al
u
n
its
,
2
an
d
2
ar
e
s
et
as
(
H,
1
)
.
T
h
e
v
alu
e
o
f
H
is
s
et
to
6
4
.
DC
L
an
d
B
C
L
d
o
n
o
t
h
a
v
e
a
n
y
p
ar
am
e
ter
s
th
at
a
r
e
s
h
ar
ed
b
etwe
en
th
em
.
T
h
e
L
a
p
lacia
n
an
d
b
lu
r
f
ilter
s
ar
e
a
p
p
lied
to
th
e
s
o
u
r
ce
im
a
g
e
a
n
d
th
e
d
etail
en
co
d
er
0
an
d
b
ase
e
n
co
d
e
r
0
ar
e
in
itialized
.
T
h
e
s
ig
m
o
id
f
u
n
ctio
n
,
a
b
atch
r
eg
u
la
r
izatio
n
lay
e
r
,
an
d
a
3
×
3
co
n
v
o
l
u
tio
n
u
n
it
m
ak
e
u
p
th
e
d
ec
o
d
er
.
T
h
e
co
n
v
o
l
u
tio
n
u
n
it
h
as
o
n
e
in
p
u
t
c
h
an
n
el
an
d
o
n
e
o
u
t
p
u
t
ch
an
n
el.
T
h
e
r
esto
r
ed
im
a
g
e’
s
p
ix
el
v
alu
es a
r
e
n
o
r
m
alize
d
to
a
r
an
g
e
o
f
0
-
1
b
y
m
ea
n
s
o
f
th
e
s
ig
m
o
i
d
f
u
n
ctio
n
.
Fig
u
r
e
4
illu
s
tr
ates
th
e
wo
r
k
f
lo
w
o
f
th
e
test
in
g
f
r
am
ewo
r
k
.
Du
r
in
g
th
e
test
p
h
ase,
in
p
u
t
p
air
s
o
f
in
f
r
ar
ed
an
d
v
is
ib
le
im
ag
es
ar
e
g
iv
e
n
an
d
th
e
u
ltima
te
f
u
s
io
n
r
esu
lts
ar
e
s
u
b
s
eq
u
en
tl
y
o
b
tain
e
d
.
U
p
o
n
co
m
p
letio
n
o
f
th
e
tr
ain
in
g
p
r
o
ce
s
s
,
two
p
r
o
f
icien
t
e
n
co
d
e
r
s
an
d
a
d
ec
o
d
er
ar
e
o
b
tai
n
ed
.
I
n
th
is
c
o
n
tex
t,
,
,
,
an
d
r
ep
r
esen
t
th
e
in
f
r
ar
e
d
d
etail,
b
ase
f
ea
tu
r
e
m
ap
s
an
d
v
is
ib
le
d
etail,
b
ase
f
ea
tu
r
e
m
ap
s
,
r
esp
ec
tiv
ely
.
Fig
u
r
e
2
.
T
r
ain
in
g
f
r
am
ewo
r
k
Fig
u
r
e
3
.
Sin
g
le
B
C
L
lay
er
Fig
u
r
e
4
.
T
r
ain
in
g
f
r
am
ewo
r
k
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
Tw
o
-
s
ca
le
d
ec
o
mp
o
s
itio
n
a
n
d
d
ee
p
le
a
r
n
in
g
fu
s
io
n
fo
r
visi
b
le
a
n
d
in
fr
a
r
ed
ima
g
es
(
R
u
h
a
n
B
ev
i A
z
a
d
)
1597
4.
E
XP
E
R
I
M
E
N
T
S
T
h
e
ef
f
ec
tiv
en
ess
o
f
th
is
s
tr
ateg
y
was
ev
alu
ated
th
r
o
u
g
h
a
co
m
p
r
eh
e
n
s
iv
e
co
m
p
a
r
ativ
e
a
n
aly
s
is
o
f
th
e
s
tan
d
ar
d
f
u
s
io
n
alg
o
r
ith
m
an
d
r
ec
en
tly
p
r
o
p
o
s
ed
f
u
s
io
n
alg
o
r
ith
m
.
Su
b
jectiv
e
ass
ess
m
en
ts
f
o
cu
s
ed
o
n
v
is
u
al
q
u
ality
an
d
p
e
r
ce
p
tu
al
c
lar
ity
,
wh
ile
o
b
jectiv
e
ass
ess
m
en
ts
u
tili
ze
d
estab
lis
h
ed
m
etr
i
cs,
s
u
ch
as e
n
tr
o
p
y
an
d
s
p
atial
f
id
elity
.
T
h
e
r
esu
lts
co
n
s
is
ten
tly
d
em
o
n
s
tr
ate
th
e
s
u
p
er
io
r
p
er
f
o
r
m
an
ce
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
,
h
ig
h
lig
h
tin
g
its
p
o
ten
tial f
o
r
p
r
ac
tical
ap
p
licatio
n
s
r
eq
u
ir
i
n
g
h
ig
h
im
ag
e
q
u
ality
.
4
.
1
.
Da
t
a
s
et
s
T
h
e
FLI
R
an
d
T
NO
d
atasets
[
2
5
]
wer
e
s
elec
ted
as
th
e
test
s
u
b
jects
to
th
o
r
o
u
g
h
ly
ass
ess
th
e
ef
f
icac
y
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
.
T
h
e
T
NO
d
ataset
co
n
s
is
ts
o
f
n
u
m
e
r
o
u
s
p
r
ea
lig
n
ed
p
air
s
o
f
n
ea
r
-
i
n
f
r
ar
ed
(
NI
R
)
a
n
d
v
is
ib
le
im
ag
es.
Similar
ly
,
th
e
FLI
R
d
ataset
co
n
s
is
ts
o
f
th
er
m
al
in
f
r
ar
ed
a
n
d
v
is
ib
le
im
ag
es.
T
h
e
ex
p
er
im
en
tal
s
etu
p
in
v
o
lv
ed
s
elec
tin
g
th
i
r
ty
p
air
s
o
f
im
ag
es
f
r
o
m
th
e
F
L
I
R
d
ataset.
T
h
ese
im
ag
es
w
er
e
th
en
u
n
i
f
o
r
m
l
y
r
esized
to
2
5
6
×
2
5
6
p
ix
els
to
f
ac
ilit
ate
th
e
an
aly
s
is
o
f
m
u
ltis
ca
le
tr
an
s
f
o
r
m
atio
n
s
.
A
lap
t
o
p
f
ea
tu
r
in
g
a
C
o
r
e
i7
p
r
o
ce
s
s
o
r
an
d
1
6
GB
o
f
R
AM
was u
tili
ze
d
f
o
r
co
n
d
u
ctin
g
th
e
ev
alu
atio
n
p
r
o
ce
s
s
.
4
.
2
.
Q
ua
lit
y
m
e
t
rics
Fo
u
r
o
b
jectiv
e
ass
ess
m
en
t
m
etr
ics
wer
e
u
s
ed
to
an
al
y
ze
th
e
im
p
ac
t
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
.
T
h
ey
ar
e
en
tr
o
p
y
(
E
N)
is
th
e
ev
alu
a
tio
n
o
f
im
a
g
e
q
u
ality
to
q
u
an
t
if
y
th
e
lev
el
o
f
in
f
o
r
m
atio
n
co
n
tain
ed
with
in
th
e
f
u
s
ed
im
ag
e.
T
h
e
av
er
a
g
e
g
r
a
d
ien
t
(
AG)
m
etr
ic
ass
es
s
es
an
im
ag
e'
s
v
is
u
al
clar
ity
b
y
ex
am
in
in
g
its
tex
tu
r
al
an
d
co
n
tr
ast
f
ea
tu
r
es.
T
h
is
ev
a
lu
atio
n
d
eter
m
i
n
es
h
o
w
well
t
h
e
co
m
b
i
n
ed
im
a
g
e
p
r
eser
v
es
th
e
in
tr
icate
d
etails
an
d
b
o
u
n
d
ar
ies
f
o
u
n
d
in
t
h
e
o
r
ig
in
al
im
ag
es.
I
m
a
g
es
with
h
i
g
h
er
AG
s
co
r
es
a
r
e
g
en
er
ally
co
n
s
id
er
ed
to
h
av
e
b
etter
p
er
ce
p
tu
al
q
u
ality
.
Sp
e
ctr
al
f
id
elity
(
SF
)
is
a
m
etr
ic
th
at
q
u
an
tifie
s
th
e
ex
ten
t
to
wh
ich
th
e
s
p
ec
tr
al
in
f
o
r
m
atio
n
o
f
th
e
in
p
u
t
im
ag
es
is
ac
cu
r
ately
m
ain
tain
ed
in
th
e
f
u
s
ed
im
ag
e,
th
u
s
en
s
u
r
in
g
th
e
f
id
elity
o
f
th
e
f
u
s
ed
im
ag
e
t
o
th
e
o
r
ig
in
al
s
p
ec
tr
al
ch
ar
ac
ter
is
tics
.
Sp
atial
d
is
to
r
tio
n
(
SD)
is
a
m
ea
s
u
r
e
o
f
th
e
ex
ten
t
to
wh
ic
h
s
p
atial
d
is
to
r
tio
n
o
r
m
is
alig
n
m
en
t
o
cc
u
r
s
d
u
r
in
g
th
e
f
u
s
io
n
p
r
o
ce
s
s
.
I
t
e
v
alu
ates
th
e
d
e
g
r
ee
to
wh
ich
th
o
s
e
s
p
atial
d
etails ar
e
m
ain
tain
ed
in
th
e
f
u
s
ed
i
m
ag
e.
4
.
3
.
E
v
a
lua
t
i
o
n a
g
a
ins
t
riv
a
l
a
lg
o
rit
hm
s
us
ing
t
he
F
L
I
R
a
nd
T
NO
da
t
a
s
et
s
T
h
e
ef
f
ec
tiv
en
ess
o
f
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
is
s
u
b
s
tan
tiated
th
r
o
u
g
h
th
e
a
p
p
licatio
n
o
f
th
e
FLI
R
d
ataset.
T
h
e
f
u
s
io
n
o
u
tc
o
m
es
o
f
th
e
s
o
u
r
ce
im
ag
e
ar
e
p
r
esen
ted
in
Fig
u
r
e
5
,
wh
ich
d
e
p
icts
a
s
ce
n
e
f
ea
tu
r
in
g
two
p
eo
p
le
s
tan
d
in
g
alo
n
g
s
id
e
th
eir
b
icy
cle
n
ea
r
th
e
ed
g
e
o
f
an
ap
ar
tm
e
n
t
co
m
p
lex
r
o
a
d
.
T
h
e
v
is
ib
le
im
ag
e
clea
r
ly
d
is
p
lay
s
t
h
e
s
p
ec
if
ic
f
e
atu
r
es
o
f
th
e
h
o
u
s
e
an
d
c
y
cle.
Ho
wev
er
,
t
h
e
in
d
iv
id
u
als
wer
e
n
o
t
d
is
ce
r
n
ib
le
in
th
e
v
is
ib
le
im
ag
e,
wh
e
r
ea
s
th
e
y
wer
e
d
etec
tab
le
in
th
e
NI
R
i
m
ag
e
b
ec
au
s
e
o
f
th
eir
h
ig
h
s
en
s
itiv
ity
to
th
er
m
al
r
ad
iatio
n
,
wh
ich
ca
p
tu
r
es d
ata
r
elate
d
to
a
p
er
s
o
n
.
Qu
an
titativ
e
an
d
q
u
alitativ
e
m
etr
ics we
r
e
u
s
ed
to
ass
es
s
th
e
f
u
s
io
n
p
er
f
o
r
m
a
n
ce
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
.
Fig
u
r
e
5
.
Fu
s
io
n
o
u
tp
u
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
2
,
Ap
r
il
20
25
:
1
5
9
3
-
1
6
0
1
1598
Fro
m
a
s
u
b
jectiv
e
s
tan
d
p
o
in
t
,
th
e
im
ag
e
p
r
o
d
u
ce
d
b
y
Fu
s
io
n
GAN
ap
p
ea
r
s
to
b
e
o
f
lo
wer
v
is
u
al
q
u
ality
an
d
lack
s
s
ev
er
al
elem
en
ts
f
o
u
n
d
in
th
e
v
is
ib
le
im
ag
e.
I
m
ag
es
g
en
er
ated
u
s
in
g
Dee
p
Fu
s
e
g
en
er
ally
ap
p
ea
r
b
lu
r
r
y
,
with
in
d
is
tin
ct
ed
g
es a
n
d
co
n
s
id
er
a
b
le
n
o
is
e.
T
h
is
lo
s
s
o
f
ed
g
e
in
f
o
r
m
atio
n
m
ak
es it d
if
f
icu
lt to
d
is
ce
r
n
th
e
in
ter
io
r
o
f
th
e
ap
a
r
tm
en
t.
T
h
e
Den
s
eFu
s
e
alg
o
r
ith
m
f
ails
to
m
ain
tain
co
n
s
is
ten
t
b
r
ig
h
tn
ess
lev
els,
p
ar
ticu
lar
ly
n
o
ticea
b
le
o
n
r
o
ad
s
u
r
f
ac
es.
T
h
e
o
u
tp
u
t
f
r
o
m
to
tal
v
ar
iatio
n
alter
n
atin
g
d
ir
ec
tio
n
m
et
h
o
d
o
f
m
u
ltip
lier
s
(
T
VADM
M
)
d
is
p
lay
s
a
d
eg
r
ee
o
f
f
o
g
g
in
ess
,
w
h
ile
two
-
s
ca
le
in
f
r
ar
ed
a
n
d
v
is
ib
le
im
ag
e
f
u
s
io
n
s
ch
em
e
(
T
SIFVS
)
r
esu
lts
s
ee
m
u
n
d
e
r
ex
p
o
s
ed
.
I
n
co
m
p
ar
is
o
n
to
all
o
th
er
m
eth
o
d
s
,
t
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
d
em
o
n
s
tr
ates
s
u
p
er
io
r
co
n
tr
as
t
in
id
en
tify
in
g
p
r
o
m
in
en
t
tar
g
ets.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
y
clea
r
ly
p
r
o
d
u
ce
s
a
f
u
s
ed
im
ag
e
th
at
s
u
cc
ess
f
u
lly
d
is
p
lay
s
th
e
r
ich
tex
tu
r
es
o
f
two
p
eo
p
le
n
ea
r
th
eir
cy
cle
s
in
th
e
d
esig
n
ated
o
r
an
g
e
b
o
x
.
As
in
d
icate
d
in
T
ab
le
1
,
th
e
p
r
o
p
o
s
ed
m
eth
o
d
o
u
tp
er
f
o
r
m
s
ex
is
tin
g
tech
n
iq
u
es
in
E
N,
AG,
an
d
SD,
h
ig
h
lig
h
tin
g
its
s
u
p
er
io
r
p
er
f
o
r
m
an
ce
in
m
ai
n
tain
in
g
d
etail
an
d
clar
ity
in
t
h
e
f
u
s
ed
i
m
ag
e
s
.
Sp
ec
if
ically
,
th
e
E
N
v
alu
e
o
f
7
.
4
5
,
AG
v
alu
e
o
f
5
.
9
1
,
an
d
SD
v
alu
e
o
f
3
8
.
1
8
ar
e
th
e
h
ig
h
est
am
o
n
g
all
th
e
co
m
p
ar
ed
m
et
h
o
d
s
,
d
em
o
n
s
tr
atin
g
th
e
m
et
h
o
d
’
s
ab
ilit
y
to
p
r
eser
v
e
f
in
e
tex
t
u
r
es,
ed
g
es,
an
d
s
p
atial
co
n
s
is
ten
cy
,
wh
ich
ar
e
cr
u
cial
f
o
r
task
s
lik
e
s
u
r
v
eillan
ce
.
T
h
e
v
is
u
al
clar
ity
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
,
as
s
h
o
w
n
in
Fig
u
r
e
6
,
is
m
ar
k
ed
ly
h
ig
h
er
th
a
n
th
at
o
f
alter
n
ativ
e
tech
n
iq
u
es,
s
u
ch
as
Dee
p
Fu
s
e
an
d
Fu
s
io
n
GAN,
wh
ich
s
h
o
w
co
m
p
r
o
m
is
es
in
s
p
atial
co
n
s
is
ten
cy
an
d
ed
g
e
d
etail.
T
ab
le
1
.
Av
e
r
ag
e
r
esu
lts
o
b
tai
n
ed
b
y
ap
p
l
y
in
g
m
u
ltip
le
tech
n
iq
u
es to
th
e
FLI
R
d
ataset
M
e
t
h
o
d
s
EN
AG
SF
SD
D
e
e
p
F
u
se
7
.
2
1
4
.
8
0
1
5
.
4
7
3
7
.
3
5
F
u
si
o
n
G
a
n
7
.
0
2
3
.
2
0
1
1
.
5
1
3
4
.
3
8
D
e
n
seF
u
se
7
.
2
1
4
.
8
2
1
5
.
5
0
3
7
.
3
2
TSI
F
V
S
7
.
1
5
5
.
5
7
1
8
.
7
9
3
5
.
8
9
I
mag
e
F
u
se
6
.
9
9
4
.
1
5
1
4
.
5
2
3
2
.
5
8
TV
-
a
d
mm
6
.
8
0
3
.
5
2
1
4
.
0
4
2
8
.
0
7
P
r
o
p
o
se
d
7
.
4
5
5
.
9
1
1
4
.
4
6
3
8
.
1
8
Fig
u
r
e
6
.
Av
e
r
ag
e
r
esu
lts
b
y
tech
n
iq
u
es
-
FLI
R
d
ataset
Usi
n
g
a
T
NO
d
ataset,
T
ab
le
2
d
is
p
lay
s
th
e
av
er
ag
e
v
alu
es
o
f
s
ix
o
b
jectiv
e
ass
ess
m
en
t
cr
it
er
ia,
with
th
e
r
ed
v
alu
es
em
p
h
asizin
g
th
e
h
ig
h
est
v
alu
es.
C
o
m
p
ar
ed
to
Dee
p
Fu
s
e
an
d
Fu
s
io
n
GAN,
o
u
r
m
eth
o
d
co
n
s
is
ten
tly
d
eliv
er
s
b
etter
p
e
r
f
o
r
m
a
n
ce
in
ter
m
s
o
f
AG
a
n
d
E
N.
T
h
ese
f
i
n
d
in
g
s
alig
n
wi
th
p
r
ev
i
o
u
s
s
tu
d
ies,
wh
er
e
m
u
ltis
ca
le
d
ec
o
m
p
o
s
itio
n
tech
n
iq
u
es
d
e
m
o
n
s
tr
ated
a
n
ad
v
a
n
tag
e
in
s
p
atial
clar
ity
.
Ho
wev
er
,
as
s
ee
n
with
SF
,
f
u
tu
r
e
im
p
r
o
v
e
m
en
t
s
m
ay
f
o
cu
s
o
n
s
p
ec
tr
al
f
id
e
lity
,
p
o
ten
tially
th
r
o
u
g
h
h
y
b
r
id
tech
n
iq
u
es
th
at
m
er
g
e
d
ec
o
m
p
o
s
itio
n
-
b
ased
a
n
d
s
p
ec
tr
al
r
ete
n
tio
n
-
f
o
cu
s
ed
m
eth
o
d
s
.
T
h
e
p
r
o
p
o
s
ed
alg
o
r
i
th
m
o
u
t
p
er
f
o
r
m
e
d
all
alter
n
ativ
es
in
ter
m
s
o
f
f
u
s
io
n
p
er
f
o
r
m
a
n
ce
,
with
th
e
e
x
ce
p
tio
n
o
f
AG
an
d
SF
,
an
d
ac
h
ie
v
ed
th
e
m
a
x
im
u
m
ac
h
iev
ab
le
s
co
r
e.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
Tw
o
-
s
ca
le
d
ec
o
mp
o
s
itio
n
a
n
d
d
ee
p
le
a
r
n
in
g
fu
s
io
n
fo
r
visi
b
le
a
n
d
in
fr
a
r
ed
ima
g
es
(
R
u
h
a
n
B
ev
i A
z
a
d
)
1599
T
ab
le
2
.
Av
e
r
ag
e
r
esu
lts
o
b
tai
n
ed
b
y
ap
p
l
y
in
g
m
u
ltip
le
tech
n
iq
u
es to
th
e
T
NO
d
ataset
M
e
t
h
o
d
s
EN
AG
SF
SD
D
e
e
p
F
u
se
6
.
8
6
3
.
6
0
1
1
.
1
3
3
2
.
2
5
F
u
si
o
n
G
a
n
6
.
5
8
2
.
4
2
8
.
7
6
2
9
.
0
4
D
e
n
seF
u
se
6
.
8
4
3
.
6
0
1
1
.
0
9
3
1
.
8
2
TSI
F
V
S
6
.
6
7
3
.
9
8
1
2
.
6
0
2
8
.
0
4
I
mag
e
F
u
se
6
.
3
8
2
.
7
2
9
.
8
0
2
2
.
9
4
TV
-
a
d
mm
6
.
4
0
2
.
5
2
9
.
0
3
2
3
.
0
1
P
r
o
p
o
se
d
6
.
9
0
3
.
3
3
9
.
8
6
3
3
.
5
0
Fig
u
r
e
7
illu
s
tr
ates
th
e
co
m
p
ar
is
o
n
o
f
av
e
r
ag
e
r
esu
lts
f
o
r
v
ar
io
u
s
tech
n
iq
u
es
b
ased
o
n
th
e
T
NO
d
ataset,
f
u
r
th
e
r
h
i
g
h
lig
h
tin
g
t
h
e
s
u
p
er
i
o
r
p
er
f
o
r
m
an
ce
o
f
t
h
e
p
r
o
p
o
s
ed
m
eth
o
d
i
n
m
o
s
t
m
etr
ics.
T
h
e
r
esu
lts
f
r
o
m
b
o
t
h
th
e
FLI
R
an
d
T
NO
d
atasets
u
n
d
er
s
co
r
e
th
e
p
o
ten
tial
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
f
o
r
ap
p
licatio
n
s
r
eq
u
ir
in
g
h
ig
h
s
p
atial
clar
ity
,
s
u
ch
as
in
s
u
r
v
eillan
ce
an
d
m
ed
ical
im
ag
in
g
.
Fu
t
u
r
e
wo
r
k
co
u
ld
a
d
d
r
ess
th
e
s
lig
h
t
tr
ad
e
-
o
f
f
in
s
p
ec
tr
al
f
id
elity
b
y
e
x
p
lo
r
in
g
h
y
b
r
id
ap
p
r
o
ac
h
es
th
at
in
c
o
r
p
o
r
ate
ad
d
itio
n
al
s
p
ec
tr
al
lo
s
s
f
u
n
ctio
n
s
,
e
n
ab
lin
g
a
m
o
r
e
b
a
lan
ce
d
f
u
s
io
n
ac
r
o
s
s
d
if
f
er
e
n
t
m
o
d
alities
.
E
x
p
an
d
in
g
t
h
e
s
co
p
e
o
f
th
is
m
eth
o
d
to
o
th
er
d
ata
ty
p
es,
s
u
ch
as
h
y
p
er
s
p
ec
tr
al
im
ag
es,
c
o
u
ld
r
ev
ea
l
n
ew
ap
p
licatio
n
s
an
d
f
u
r
th
er
en
h
an
ce
th
e
r
o
b
u
s
tn
ess
o
f
th
e
p
r
o
p
o
s
ed
f
u
s
io
n
s
tr
ateg
y
.
Fig
u
r
e
7
.
Av
e
r
ag
e
r
esu
lts
b
y
tech
n
iq
u
es
-
T
NO
d
ataset
5.
CO
NCLU
SI
O
N
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
f
o
r
v
is
ib
le
an
d
in
f
r
a
r
ed
im
a
g
e
f
u
s
io
n
,
w
h
ich
co
m
b
in
es
two
-
s
ca
le
d
ec
o
m
p
o
s
itio
n
with
d
ee
p
lear
n
in
g
tech
n
iq
u
es,
h
as
d
em
o
n
s
t
r
ated
s
ig
n
if
ican
t
im
p
r
o
v
em
en
ts
in
p
er
f
o
r
m
an
ce
o
v
er
ex
is
tin
g
ap
p
r
o
ac
h
es.
Ou
r
ap
p
r
o
ac
h
ef
f
ec
tiv
ely
p
r
eser
v
es
b
o
th
th
e
s
p
atial
an
d
th
er
m
a
l
in
f
o
r
m
atio
n
f
r
o
m
th
e
in
p
u
t
im
ag
es,
as
r
ef
lecte
d
b
y
th
e
h
i
g
h
est
AG,
E
N,
an
d
SD
v
alu
es
am
o
n
g
all
co
m
p
ar
ed
m
eth
o
d
s
.
T
h
ese
r
esu
lts
in
d
icate
th
at
o
u
r
f
u
s
io
n
m
o
d
el
is
p
ar
ticu
lar
ly
well
-
s
u
ited
f
o
r
ap
p
licatio
n
s
wh
er
e
s
p
atial
d
etail
an
d
clar
ity
ar
e
cr
itical,
s
u
ch
as su
r
v
eillan
ce
,
m
ilit
ar
y
,
an
d
m
ed
ic
al
im
ag
in
g
.
B
ey
o
n
d
d
em
o
n
s
tr
atin
g
its
ef
f
icac
y
in
im
ag
e
f
u
s
io
n
,
th
is
r
esear
ch
also
o
p
en
s
th
e
d
o
o
r
f
o
r
f
u
tu
r
e
in
v
esti
g
atio
n
s
.
W
h
ile
th
e
m
e
th
o
d
ex
ce
ls
in
m
ain
tain
in
g
t
ex
tu
r
e
an
d
clar
ity
,
f
u
r
t
h
er
w
o
r
k
is
r
eq
u
ir
ed
to
en
h
an
ce
s
p
ec
t
r
al
f
id
elity
.
E
x
p
lo
r
in
g
h
y
b
r
id
f
u
s
io
n
tec
h
n
iq
u
es
th
at
in
co
r
p
o
r
ate
s
p
ec
tr
al
r
et
en
tio
n
m
ec
h
an
is
m
s
alo
n
g
s
id
e
m
u
ltis
ca
le
d
ec
o
m
p
o
s
itio
n
co
u
ld
ad
d
r
ess
th
is
lim
itatio
n
.
Ad
d
itio
n
ally
,
ex
p
an
d
in
g
th
e
m
eth
o
d
to
h
an
d
le
o
t
h
er
m
u
ltimo
d
al
im
a
g
e
f
u
s
io
n
task
s
,
s
u
ch
as
h
y
p
e
r
s
p
ec
tr
al
o
r
r
a
d
ar
im
a
g
er
y
,
p
r
esen
ts
an
ex
citin
g
d
ir
ec
tio
n
f
o
r
f
u
tu
r
e
r
esear
c
h
.
I
n
co
n
clu
s
io
n
,
o
u
r
f
in
d
in
g
s
c
o
n
tr
ib
u
te
to
th
e
o
n
g
o
in
g
ad
v
an
ce
m
en
ts
in
im
a
g
e
f
u
s
io
n
t
ec
h
n
iq
u
es,
o
f
f
er
in
g
a
r
o
b
u
s
t
s
o
lu
tio
n
f
o
r
co
m
b
in
in
g
v
is
ib
le
an
d
in
f
r
a
r
e
d
d
ata.
T
h
e
r
ea
l
-
wo
r
ld
im
p
lica
tio
n
s
o
f
th
is
m
eth
o
d
ar
e
v
ast,
an
d
its
ap
p
licatio
n
a
cr
o
s
s
d
if
f
er
en
t
d
o
m
ain
s
ca
n
p
o
ten
tially
lead
to
en
h
a
n
ce
d
d
e
tectio
n
,
m
o
n
ito
r
i
n
g
,
an
d
im
a
g
in
g
s
y
s
tem
s
.
Mo
v
in
g
f
o
r
war
d
,
im
p
r
o
v
em
en
ts
i
n
s
p
ec
tr
al
f
id
elity
an
d
ex
ten
s
io
n
to
b
r
o
ad
er
d
atasets
will f
u
r
th
er
s
o
lid
if
y
t
h
e
u
tili
ty
o
f
th
is
ap
p
r
o
ac
h
with
in
t
h
e
im
ag
e
f
u
s
io
n
r
esear
ch
co
m
m
u
n
it
y
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
2
,
Ap
r
il
20
25
:
1
5
9
3
-
1
6
0
1
1600
ACK
NO
WL
E
DG
E
M
E
NT
S
T
h
e
au
th
o
r
e
x
p
r
ess
es
s
in
ce
r
e
g
r
atitu
d
e
to
ca
r
r
y
th
is
ex
ten
d
e
d
r
esear
ch
b
ased
o
n
th
e
f
in
a
n
c
ial
s
u
p
p
o
r
t
r
ec
eiv
ed
f
r
o
m
th
e
SR
M
I
n
s
titu
te
o
f
Scien
ce
an
d
T
ec
h
n
o
l
o
g
y
,
Kattan
k
u
lath
u
r
,
I
n
d
ia,
u
n
d
er
th
e
Selectiv
e
E
x
ce
llen
ce
R
esear
ch
I
n
itiativ
e
(
SERI)
.
RE
F
E
R
E
NC
E
S
[
1
]
H
.
V
.
N
g
u
y
e
n
,
M
.
O
.
U
l
f
a
r
ss
o
n
,
J.
R
.
S
v
e
i
n
sso
n
,
a
n
d
M
.
D
.
M
u
r
a
,
“
U
n
s
u
p
e
r
v
i
s
e
d
s
e
n
t
i
n
e
l
-
2
i
m
a
g
e
f
u
si
o
n
u
s
i
n
g
a
d
e
e
p
u
n
r
o
l
l
i
n
g
met
h
o
d
,
”
I
EE
E
G
e
o
sc
i
e
n
c
e
a
n
d
Re
m
o
t
e
S
e
n
si
n
g
L
e
t
t
e
rs
,
v
o
l
.
2
0
,
p
p
.
1
–
5
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
LG
R
S
.
2
0
2
3
.
3
3
2
6
8
4
5
.
[
2
]
J.
M
.
R
a
mi
r
e
z
,
J.
I
.
M
a
r
t
í
n
e
z
-
To
r
r
e
,
a
n
d
H
.
A
r
g
u
e
l
l
o
,
“
LA
D
M
M
-
N
e
t
:
a
n
u
n
r
o
l
l
e
d
d
e
e
p
n
e
t
w
o
r
k
f
o
r
s
p
e
c
t
r
a
l
i
mag
e
f
u
si
o
n
f
r
o
m
c
o
m
p
r
e
ss
i
v
e
d
a
t
a
,
”
S
i
g
n
a
l
Pr
o
c
e
ssi
n
g
,
v
o
l
.
1
8
9
,
D
e
c
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
s
i
g
p
r
o
.
2
0
2
1
.
1
0
8
2
3
9
.
[
3
]
S
.
Zh
a
n
g
,
X
.
Li
,
X
.
Z
h
a
n
g
,
a
n
d
S
.
Z
h
a
n
g
,
“
I
n
f
r
a
r
e
d
a
n
d
v
i
si
b
l
e
i
ma
g
e
f
u
si
o
n
b
a
se
d
o
n
s
a
l
i
e
n
c
y
d
e
t
e
c
t
i
o
n
a
n
d
t
w
o
-
sca
l
e
t
r
a
n
sf
o
r
m
d
e
c
o
m
p
o
si
t
i
o
n
,
”
I
n
f
r
a
r
e
d
P
h
y
si
c
s &
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
1
1
4
,
M
a
y
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
n
f
r
a
r
e
d
.
2
0
2
0
.
1
0
3
6
2
6
.
[
4
]
L.
M
a
,
Y
.
H
u
,
B
.
Z
h
a
n
g
,
J.
Li
,
Z
.
C
h
e
n
,
a
n
d
W
.
S
u
n
,
“
A
n
e
w
m
u
l
t
i
-
f
o
c
u
s i
mag
e
f
u
si
o
n
me
t
h
o
d
b
a
se
d
o
n
m
u
l
t
i
-
c
l
a
ssi
f
i
c
a
t
i
o
n
f
o
c
u
s
l
e
a
r
n
i
n
g
a
n
d
m
u
l
t
i
-
s
c
a
l
e
d
e
c
o
mp
o
s
i
t
i
o
n
,
”
Ap
p
l
i
e
d
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
5
3
,
n
o
.
2
,
p
p
.
1
4
5
2
–
1
4
6
8
,
Ja
n
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
0
4
8
9
-
0
2
2
-
0
3
6
5
8
-
2.
[
5
]
L.
G
o
p
i
n
a
t
h
a
n
d
A
.
R
.
B
e
v
i
,
“
A
n
i
s
o
t
r
o
p
i
c
g
u
i
d
e
d
f
i
l
t
e
r
i
n
g
a
n
d
m
u
l
t
i
-
l
e
v
e
l
d
i
si
n
t
e
g
r
a
t
i
o
n
met
h
o
d
f
o
r
N
I
R
a
n
d
v
i
si
b
l
e
i
ma
g
e
f
u
si
o
n
,
”
i
n
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
C
o
g
n
i
t
i
v
e
C
o
m
p
u
t
i
n
g
a
n
d
I
n
f
o
rm
a
t
i
o
n
Pr
o
c
e
ss
i
n
g
,
2
0
2
4
,
p
p
.
1
6
5
–
1
7
5
.
[
6
]
K
.
L
i
,
G
.
Li
u
,
X
.
G
u
,
H
.
T
a
n
g
,
J
.
X
i
o
n
g
,
a
n
d
Y
.
Q
i
a
n
,
“
D
A
N
T
-
G
A
N
:
a
d
u
a
l
a
t
t
e
n
t
i
o
n
-
b
a
s
e
d
o
f
n
e
s
t
e
d
t
r
a
i
n
i
n
g
n
e
t
w
o
r
k
f
o
r
i
n
f
r
a
r
e
d
a
n
d
v
i
si
b
l
e
i
ma
g
e
f
u
si
o
n
,
”
D
i
g
i
t
a
l
S
i
g
n
a
l
Pr
o
c
e
ssi
n
g
,
v
o
l
.
1
4
5
,
F
e
b
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
d
s
p
.
2
0
2
3
.
1
0
4
3
1
6
.
[
7
]
L.
G
o
p
i
n
a
t
h
a
n
d
A
.
R
u
h
a
n
B
e
v
i
,
“
A
d
i
m
e
n
s
i
o
n
a
l
i
t
y
r
e
d
u
c
t
i
o
n
me
t
h
o
d
f
o
r
t
h
e
f
u
s
i
o
n
o
f
N
I
R
a
n
d
v
i
s
i
b
l
e
i
m
a
g
e
,
”
i
n
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
I
m
a
g
e
Pro
c
e
ss
i
n
g
a
n
d
C
a
p
s
u
l
e
N
e
t
w
o
r
k
s
,
2
0
2
3
,
p
p
.
6
2
9
–
645.
[
8
]
P
.
Z
h
u
,
W
.
O
u
y
a
n
g
,
Y
.
G
u
o
,
a
n
d
X
.
Z
h
o
u
,
“
A
t
w
o
-
to
-
o
n
e
d
e
e
p
l
e
a
r
n
i
n
g
g
e
n
e
r
a
l
f
r
a
m
e
w
o
r
k
f
o
r
i
ma
g
e
f
u
si
o
n
,
”
Fro
n
t
i
e
rs
i
n
Bi
o
e
n
g
i
n
e
e
r
i
n
g
a
n
d
Bi
o
t
e
c
h
n
o
l
o
g
y
,
v
o
l
.
1
0
,
J
u
l
.
2
0
2
2
,
d
o
i
:
1
0
.
3
3
8
9
/
f
b
i
o
e
.
2
0
2
2
.
9
2
3
3
6
4
.
[
9
]
H
.
X
u
,
J
.
M
a
,
J
.
Ji
a
n
g
,
X
.
G
u
o
,
a
n
d
H
.
Li
n
g
,
“
U
2
F
u
s
i
o
n
:
a
u
n
i
f
i
e
d
u
n
su
p
e
r
v
i
s
e
d
i
m
a
g
e
f
u
s
i
o
n
n
e
t
w
o
r
k
,
”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
Pa
t
t
e
r
n
An
a
l
y
si
s
a
n
d
Ma
c
h
i
n
e
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
4
4
,
n
o
.
1
,
p
p
.
5
0
2
–
5
1
8
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
TPA
M
I
.
2
0
2
0
.
3
0
1
2
5
4
8
.
[
1
0
]
H
.
U
n
n
i
k
r
i
s
h
n
a
n
a
n
d
R
.
B
.
A
z
a
d
,
“
N
o
n
l
o
c
a
l
r
e
t
i
n
e
x
b
a
s
e
d
d
e
h
a
z
i
n
g
a
n
d
l
o
w
l
i
g
h
t
e
n
h
a
n
c
e
me
n
t
o
f
i
m
a
g
e
s,
”
T
ra
i
t
e
m
e
n
t
d
u
S
i
g
n
a
l
,
v
o
l
.
3
9
,
n
o
.
3
,
p
p
.
8
7
9
–
8
9
2
,
2
0
2
2
,
d
o
i
:
1
0
.
1
8
2
8
0
/
t
s
.
3
9
0
3
1
3
.
[
1
1
]
N
.
A
n
a
n
t
r
a
si
r
i
c
h
a
i
,
R
.
Z
h
e
n
g
,
I
.
S
e
l
e
sn
i
c
k
,
a
n
d
A
.
A
c
h
i
m,
“
I
mag
e
f
u
s
i
o
n
v
i
a
s
p
a
r
se
r
e
g
u
l
a
r
i
z
a
t
i
o
n
w
i
t
h
n
o
n
-
c
o
n
v
e
x
p
e
n
a
l
t
i
e
s,
”
Pa
t
t
e
r
n
Re
c
o
g
n
i
t
i
o
n
L
e
t
t
e
rs
,
v
o
l
.
1
3
1
,
p
p
.
3
5
5
–
3
6
0
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
a
t
r
e
c
.
2
0
2
0
.
0
1
.
0
2
0
.
[
1
2
]
Y
.
P
a
n
,
T
.
La
n
,
C
.
X
u
,
C
.
Zh
a
n
g
,
a
n
d
Z.
F
e
n
g
,
“
R
e
c
e
n
t
a
d
v
a
n
c
e
s
v
i
a
c
o
n
v
o
l
u
t
i
o
n
a
l
sp
a
r
se
r
e
p
r
e
sen
t
a
t
i
o
n
mo
d
e
l
f
o
r
p
i
x
e
l
-
l
e
v
e
l
i
m
a
g
e
f
u
si
o
n
,
”
Mu
l
t
i
m
e
d
i
a
T
o
o
l
s
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
8
3
,
n
o
.
1
7
,
p
p
.
5
2
8
9
9
–
5
2
9
3
0
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
1
0
4
2
-
0
2
3
-
17584
-
z.
[
1
3
]
Q
.
H
u
,
S
.
H
u
,
X
.
M
a
,
F
.
Z
h
a
n
g
,
a
n
d
J
.
F
a
n
g
,
“
M
R
I
i
m
a
g
e
f
u
si
o
n
b
a
sed
o
n
o
p
t
i
mi
z
e
d
d
i
c
t
i
o
n
a
r
y
l
e
a
r
n
i
n
g
a
n
d
b
i
n
a
r
y
m
a
p
r
e
f
i
n
i
n
g
i
n
g
r
a
d
i
e
n
t
d
o
mai
n
,
”
Mu
l
t
i
m
e
d
i
a
T
o
o
l
s
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
8
2
,
n
o
.
2
,
p
p
.
2
5
3
9
–
2
5
6
1
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
1
0
4
2
-
0
2
2
-
12225
-
3.
[
1
4
]
Y
.
Ji
e
,
X
.
L
i
,
H
.
Ta
n
,
F
.
Zh
o
u
,
a
n
d
G
.
W
a
n
g
,
“
M
u
l
t
i
-
m
o
d
a
l
me
d
i
c
a
l
i
mag
e
f
u
si
o
n
v
i
a
m
u
l
t
i
-
d
i
c
t
i
o
n
a
r
y
a
n
d
t
r
u
n
c
a
t
e
d
H
u
b
e
r
f
i
l
t
e
r
i
n
g
,
”
Bi
o
m
e
d
i
c
a
l
S
i
g
n
a
l
Pr
o
c
e
ssi
n
g
a
n
d
C
o
n
t
r
o
l
,
v
o
l
.
8
8
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
b
s
p
c
.
2
0
2
3
.
1
0
5
6
7
1
.
[
1
5
]
A
.
C
a
m
a
c
h
o
,
E.
V
a
r
g
a
s,
a
n
d
H
.
A
r
g
u
e
l
l
o
,
“
H
y
p
e
r
s
p
e
c
t
r
a
l
a
n
d
m
u
l
t
i
sp
e
c
t
r
a
l
i
m
a
g
e
f
u
s
i
o
n
a
d
d
r
e
ss
i
n
g
s
p
e
c
t
r
a
l
v
a
r
i
a
b
i
l
i
t
y
b
y
a
n
a
u
g
me
n
t
e
d
l
i
n
e
a
r
mi
x
i
n
g
mo
d
e
l
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
Re
m
o
t
e
S
e
n
s
i
n
g
,
v
o
l
.
4
3
,
n
o
.
5
,
p
p
.
1
5
7
7
–
1
6
0
8
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
8
0
/
0
1
4
3
1
1
6
1
.
2
0
2
2
.
2
0
4
1
7
6
2
.
[
1
6
]
Q
.
W
e
i
,
J.
B
i
o
u
c
a
s
-
D
i
a
s,
N
.
D
o
b
i
g
e
o
n
,
a
n
d
J.
Y
.
T
o
u
r
n
e
r
e
t
,
“
H
y
p
e
r
s
p
e
c
t
r
a
l
a
n
d
mu
l
t
i
sp
e
c
t
r
a
l
i
ma
g
e
f
u
si
o
n
b
a
se
d
o
n
a
sp
a
r
se
r
e
p
r
e
s
e
n
t
a
t
i
o
n
,
”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
G
e
o
s
c
i
e
n
c
e
a
n
d
R
e
m
o
t
e
S
e
n
si
n
g
,
v
o
l
.
5
3
,
n
o
.
7
,
p
p
.
3
6
5
8
–
3
6
6
8
,
2
0
1
5
,
d
o
i
:
1
0
.
1
1
0
9
/
TG
R
S
.
2
0
1
4
.
2
3
8
1
2
7
2
.
[
1
7
]
H
.
Li
,
X
.
J.
W
u
,
a
n
d
T.
D
u
r
r
a
n
i
,
“
N
e
st
F
u
s
e
:
a
n
i
n
f
r
a
r
e
d
a
n
d
v
i
si
b
l
e
i
m
a
g
e
f
u
si
o
n
a
r
c
h
i
t
e
c
t
u
r
e
b
a
se
d
o
n
n
e
s
t
c
o
n
n
e
c
t
i
o
n
a
n
d
sp
a
t
i
a
l
/
c
h
a
n
n
e
l
a
t
t
e
n
t
i
o
n
m
o
d
e
l
s,”
I
E
EE
T
r
a
n
sa
c
t
i
o
n
s
o
n
I
n
s
t
r
u
m
e
n
t
a
t
i
o
n
a
n
d
Me
a
s
u
reme
n
t
,
v
o
l
.
6
9
,
n
o
.
1
2
,
p
p
.
9
6
4
5
–
9
6
5
6
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
0
9
/
TI
M
.
2
0
2
0
.
3
0
0
5
2
3
0
.
[
1
8
]
H
.
K
a
u
r
,
D
.
K
o
u
n
d
a
l
,
a
n
d
V
.
K
a
d
y
a
n
,
“
M
u
l
t
i
m
o
d
a
l
i
m
a
g
e
f
u
si
o
n
:
C
o
m
p
a
r
a
t
i
v
e
a
n
a
l
y
s
i
s,
”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
1
9
I
E
EE
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
m
u
n
i
c
a
t
i
o
n
a
n
d
S
i
g
n
a
l
Pr
o
c
e
s
si
n
g
,
I
C
C
S
P
2
0
1
9
,
2
0
1
9
,
p
p
.
7
5
8
–
7
6
1
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
S
P
.
2
0
1
9
.
8
6
9
7
9
6
7
.
[
1
9
]
L.
Ji
a
n
,
X
.
Y
a
n
g
,
Z
.
Li
u
,
G
.
J
e
o
n
,
M
.
G
a
o
,
a
n
d
D
.
C
h
i
s
h
o
l
m
,
“
S
ED
R
F
u
se
:
a
sy
mm
e
t
r
i
c
e
n
c
o
d
e
r
–
d
e
c
o
d
e
r
w
i
t
h
r
e
si
d
u
a
l
b
l
o
c
k
n
e
t
w
o
r
k
f
o
r
i
n
f
r
a
r
e
d
a
n
d
v
i
si
b
l
e
i
ma
g
e
f
u
s
i
o
n
,
”
I
E
EE
T
r
a
n
sa
c
t
i
o
n
s
o
n
I
n
st
ru
m
e
n
t
a
t
i
o
n
a
n
d
M
e
a
s
u
rem
e
n
t
,
v
o
l
.
7
0
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
TI
M
.
2
0
2
0
.
3
0
2
2
4
3
8
.
[
2
0
]
J.
M
a
,
H
.
Z
h
a
n
g
,
Z.
S
h
a
o
,
P
.
Li
a
n
g
,
a
n
d
H
.
X
u
,
“
G
A
N
M
c
C
:
a
g
e
n
e
r
a
t
i
v
e
a
d
v
e
r
sari
a
l
n
e
t
w
o
r
k
w
i
t
h
m
u
l
t
i
c
l
a
ssi
f
i
c
a
t
i
o
n
c
o
n
s
t
r
a
i
n
t
s
f
o
r
i
n
f
r
a
r
e
d
a
n
d
v
i
si
b
l
e
i
ma
g
e
f
u
s
i
o
n
,
”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
I
n
st
r
u
m
e
n
t
a
t
i
o
n
a
n
d
M
e
a
s
u
rem
e
n
t
,
v
o
l
.
7
0
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
TI
M
.
2
0
2
0
.
3
0
3
8
0
1
3
.
[
2
1
]
W
.
Y
i
n
,
K
.
H
e
,
D
.
X
u
,
Y
.
L
u
o
,
a
n
d
J
.
G
o
n
g
,
“
A
d
a
p
t
i
v
e
e
n
h
a
n
c
e
d
i
n
f
r
a
r
e
d
a
n
d
v
i
si
b
l
e
i
m
a
g
e
f
u
s
i
o
n
u
si
n
g
h
y
b
r
i
d
d
e
c
o
m
p
o
s
i
t
i
o
n
a
n
d
c
o
u
p
l
e
d
d
i
c
t
i
o
n
a
r
y
,
”
N
e
u
ra
l
C
o
m
p
u
t
i
n
g
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
3
4
,
n
o
.
2
3
,
p
p
.
2
0
8
3
1
–
2
0
8
4
9
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
0
7
/
s0
0
5
2
1
-
022
-
0
7
5
5
9
-
w.
[
2
2
]
H
.
L
i
,
X
.
J
.
W
u
,
a
n
d
J.
K
i
t
t
l
e
r
,
“
R
F
N
-
N
e
st
:
a
n
e
n
d
-
to
-
e
n
d
r
e
s
i
d
u
a
l
f
u
s
i
o
n
n
e
t
w
o
r
k
f
o
r
i
n
f
r
a
r
e
d
a
n
d
v
i
si
b
l
e
i
m
a
g
e
s,”
I
n
f
o
rm
a
t
i
o
n
Fu
si
o
n
,
v
o
l
.
7
3
,
p
p
.
7
2
–
8
6
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
n
f
f
u
s
.
2
0
2
1
.
0
2
.
0
2
3
.
[
2
3
]
J.
D
i
e
t
a
l
.
,
“
F
D
N
e
t
:
a
n
e
n
d
-
to
-
e
n
d
f
u
si
o
n
d
e
c
o
m
p
o
si
t
i
o
n
n
e
t
w
o
r
k
f
o
r
i
n
f
r
a
r
e
d
a
n
d
v
i
s
i
b
l
e
i
m
a
g
e
s,
”
PL
o
S
O
N
E
,
v
o
l
.
1
8
,
n
o
.
9
S
e
p
t
e
m
b
e
r
,
2
0
2
3
,
d
o
i
:
1
0
.
1
3
7
1
/
j
o
u
r
n
a
l
.
p
o
n
e
.
0
2
9
0
2
3
1
.
[
2
4
]
Y
.
Lu
o
a
n
d
Z.
L
u
o
,
“
I
n
f
r
a
r
e
d
a
n
d
v
i
s
i
b
l
e
i
ma
g
e
f
u
si
o
n
:
m
e
t
h
o
d
s
,
d
a
t
a
set
s,
a
p
p
l
i
c
a
t
i
o
n
s
,
a
n
d
p
r
o
s
p
e
c
t
s
,
”
A
p
p
l
i
e
d
S
c
i
e
n
c
e
s
(
S
w
i
t
zer
l
a
n
d
)
,
v
o
l
.
1
3
,
n
o
.
1
9
,
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
a
p
p
1
3
1
9
1
0
8
9
1
.
[
2
5
]
A
.
To
e
t
,
“
Th
e
T
N
O
m
u
l
t
i
b
a
n
d
i
m
a
g
e
d
a
t
a
c
o
l
l
e
c
t
i
o
n
,
”
D
a
t
a
i
n
Br
i
e
f
,
v
o
l
.
1
5
,
p
p
.
2
4
9
–
2
5
1
,
2
0
1
7
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
d
i
b
.
2
0
1
7
.
0
9
.
0
3
8
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
Tw
o
-
s
ca
le
d
ec
o
mp
o
s
itio
n
a
n
d
d
ee
p
le
a
r
n
in
g
fu
s
io
n
fo
r
visi
b
le
a
n
d
in
fr
a
r
ed
ima
g
es
(
R
u
h
a
n
B
ev
i A
z
a
d
)
1601
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Ruh
a
n
Be
v
i
Az
a
d
is
a
n
a
ss
o
c
iate
p
ro
fe
ss
o
r
in
t
h
e
De
p
a
rtme
n
t
o
f
El
e
c
tro
n
ics
a
n
d
Co
m
m
u
n
ica
ti
o
n
En
g
i
n
e
e
rin
g
,
S
R
M
In
stit
u
te
o
f
S
c
ien
c
e
a
n
d
Tec
h
n
o
l
o
g
y
,
KTR
c
a
m
p
u
s.
S
h
e
h
o
l
d
s
a
m
a
ste
r’s
d
e
g
re
e
i
n
e
m
b
e
d
d
e
d
s
y
ste
m
tec
h
n
o
lo
g
ies
fro
m
An
n
a
Un
i
v
e
rsity
,
Ch
e
n
n
a
i
a
n
d
a
P
h
.
D.
in
se
c
u
rit
y
in
e
m
b
e
d
d
e
d
sy
ste
m
s
fro
m
S
RM
In
stit
u
te
o
f
S
c
ien
c
e
a
n
d
Tec
h
n
o
l
o
g
y
.
He
r
re
se
a
rc
h
in
tere
sts
in
c
lu
d
e
d
e
e
p
l
e
a
rn
in
g
,
ima
g
e
c
o
g
n
it
i
o
n
,
sig
n
a
l
a
n
d
ima
g
e
p
ro
c
e
ss
in
g
with
m
a
c
h
in
e
lea
rn
in
g
,
n
e
u
ra
l
n
e
two
r
k
s,
Io
T
fo
r
a
u
to
m
a
ti
o
n
,
se
c
u
rit
y
in
e
m
b
e
d
d
e
d
sy
ste
m
s,
a
n
d
re
c
o
n
fig
u
ra
b
le
c
o
m
p
u
ti
n
g
.
S
h
e
h
a
s
twe
n
ty
-
th
re
e
y
e
a
rs
o
f
tea
c
h
i
n
g
e
x
p
e
rien
c
e
.
S
h
e
c
a
n
b
e
re
a
c
h
e
d
v
ia em
a
il
a
t
ru
h
a
n
b
@s
rm
ist.
e
d
u
.
i
n
.
H
a
r
i
Unn
ik
r
ish
n
a
n
is
wo
rk
in
g
se
lec
ti
o
n
g
ra
d
e
a
ss
istan
t
p
ro
fe
s
so
r
in
De
p
a
rtme
n
t
o
f
El
e
c
tro
n
ics
a
n
d
C
o
m
m
u
n
ica
ti
o
n
En
g
in
e
e
rin
g
,
a
t
S
a
v
e
e
th
a
En
g
in
e
e
rin
g
C
o
ll
e
g
e
,
Ch
e
n
n
a
i
.
He
h
a
s
m
o
re
th
a
n
twe
n
ty
-
fiv
e
y
e
a
rs
o
f
tea
c
h
in
g
e
x
p
e
rien
c
e
.
He
re
c
e
iv
e
d
h
is
b
a
c
h
e
lo
r’s
d
e
g
re
e
in
e
lec
tro
n
ics
e
n
g
i
n
e
e
rin
g
fro
m
M
a
d
ra
s
in
stit
u
te
o
f
Tec
h
n
o
l
o
g
y
,
An
n
a
Un
i
v
e
rsity
,
Ch
e
n
n
a
i,
i
n
y
e
a
r
1
9
9
3
.
He
c
o
m
p
lete
d
a
m
a
ste
r’s d
e
g
re
e
in
p
o
we
r
e
lec
tro
n
ics
a
n
d
d
ri
v
e
s fro
m
th
e
C
o
ll
e
g
e
o
f
En
g
i
n
e
e
rin
g
,
An
n
a
Un
i
v
e
rsity
,
C
h
e
n
n
a
i.
He
c
o
m
p
lete
d
P
h
.
D.
a
t
S
RM
In
stit
u
te
o
f
S
c
ien
c
e
a
n
d
Tec
h
n
o
l
o
g
y
,
Tam
il
Na
d
u
,
I
n
d
ia
.
He
h
a
s
p
u
b
li
sh
e
d
m
a
n
y
r
e
se
a
rc
h
p
a
p
e
rs
i
n
n
a
ti
o
n
a
l
a
n
d
in
tern
a
ti
o
n
a
l
c
o
n
fe
re
n
c
e
s
a
n
d
in
d
e
x
e
d
j
o
u
r
n
a
ls.
His
re
se
a
rc
h
i
n
tere
sts
in
c
l
u
d
e
sig
n
a
l
p
ro
c
e
ss
in
g
,
ima
g
e
p
r
o
c
e
ss
in
g
,
m
a
c
h
in
e
lea
rn
i
n
g
a
n
d
wire
les
s
c
o
m
m
u
n
ica
ti
o
n
s.
He
is
a
m
e
m
b
e
r
o
f
In
st
it
u
t
io
n
o
f
En
g
i
n
e
e
rs
(IE
),
I
n
d
ia.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
h
a
riu
@s
a
v
e
e
th
a
.
a
c
.
i
n
o
r
u
h
a
rih
8
6
@
g
m
a
il
.
c
o
m
.
Lo
k
e
sh
G
o
p
i
n
a
th
c
o
m
p
lete
d
h
is
B.
E
i
n
e
lec
tro
n
ics
a
n
d
c
o
m
m
u
n
ica
ti
o
n
e
n
g
in
e
e
rin
g
fr
o
m
TJ
I
n
stit
u
te
o
f
Tec
h
n
o
l
o
g
y
,
a
ffil
iate
d
t
o
An
n
a
U
n
iv
e
rsit
y
,
C
h
e
n
n
a
i,
In
d
ia
i
n
2
0
1
3
.
He
o
b
tai
n
e
d
h
is
M
.
E
d
e
g
r
e
e
in
e
m
b
e
d
d
e
d
s
y
ste
m
tec
h
n
o
lo
g
ies
fro
m
An
a
n
d
I
n
stit
u
te
o
f
Hig
h
e
r
Tec
h
n
o
l
o
g
y
,
a
ffil
iate
d
wi
th
An
n
a
Un
iv
e
rsit
y
,
Ch
e
n
n
a
i,
I
n
d
ia
in
2
0
1
6
.
C
u
rre
n
tl
y
,
h
e
is
p
u
rsu
i
n
g
a
P
h
.
D.
d
e
g
re
e
i
n
e
lec
t
ro
n
ics
a
n
d
c
o
m
m
u
n
ica
ti
o
n
e
n
g
in
e
e
rin
g
a
t
S
RM
I
n
stit
u
te
o
f
S
c
ien
c
e
a
n
d
Tec
h
n
o
lo
g
y
,
Ka
tt
a
n
k
u
lath
u
r,
I
n
d
ia.
His
re
se
a
rc
h
fo
c
u
s
e
s
o
n
ima
g
e
p
r
o
c
e
ss
in
g
a
n
d
d
e
e
p
lea
rn
in
g
.
He
c
a
n
b
e
re
a
c
h
e
d
a
t
lg
0
6
5
4
@s
rm
ist.
e
d
u
.
in
.
Evaluation Warning : The document was created with Spire.PDF for Python.