I
nte
rna
t
io
na
l J
o
urna
l o
f
I
nfo
rm
a
t
ics a
nd
Co
m
m
un
ica
t
io
n T
ec
hn
o
lo
g
y
(
I
J
-
I
CT
)
Vo
l.
9
,
No
.
3
,
Dec
em
b
er
2
0
2
0
,
p
p
.
19
5
~
20
4
I
SS
N:
2252
-
8
7
7
6
,
DOI
:
1
0
.
1
1
5
9
1
/
ijict.v
9
i3
.
p
p
19
5
-
20
4
195
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ict.
ia
esco
r
e.
co
m
A neuro
fuzzy im
a
g
e f
usio
n using
b
lo
ck bas
ed
feature
level me
th
o
d
S.
M
a
ry
P
ra
v
ee
na
,
R.
K
a
nm
a
ni
,
A.
K.
K
a
v
it
ha
De
p
a
rtme
n
t
o
f
ECE
,
S
ri
Ra
m
a
k
rish
n
a
In
sti
tu
te
o
f
Tec
h
n
o
l
o
g
y
,
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
Dec
28
,
2
0
1
8
R
ev
is
ed
Mar
2
0
,
2
0
19
Acc
ep
ted
J
an
1
2
,
2
0
20
Im
a
g
e
fu
sio
n
is
a
su
b
field
o
f
i
m
a
g
e
p
ro
c
e
ss
in
g
in
wh
ich
m
o
re
th
a
n
o
n
e
ima
g
e
s
a
re
fu
se
d
to
c
re
a
te
a
n
im
a
g
e
wh
e
re
a
ll
th
e
o
b
jec
ts
a
re
in
f
o
c
u
s.
Th
e
p
ro
c
e
ss
o
f
ima
g
e
fu
si
o
n
is
p
e
r
fo
rm
e
d
fo
r
m
u
lt
i
-
se
n
so
r
a
n
d
m
u
lt
i
-
f
o
c
u
s
ima
g
e
s
o
f
th
e
sa
m
e
s
c
e
n
e
.
M
u
lt
i
-
se
n
so
r
ima
g
e
s
o
f
th
e
sa
m
e
sc
e
n
e
a
re
c
a
p
tu
re
d
b
y
d
iffere
n
t
se
n
so
rs
wh
e
re
a
s
m
u
lt
i
-
fo
c
u
s
ima
g
e
s
a
re
c
a
p
tu
re
d
b
y
th
e
sa
m
e
s
e
n
so
r.
In
m
u
lt
i
-
f
o
c
u
s
ima
g
e
s,
th
e
o
b
jec
ts
in
t
h
e
sc
e
n
e
wh
ich
a
re
c
lo
se
r
to
t
h
e
c
a
m
e
ra
a
re
in
f
o
c
u
s
a
n
d
th
e
fa
rth
e
r
o
b
jec
ts
g
e
t
b
lu
rre
d
.
C
o
n
trary
t
o
it
,
wh
e
n
th
e
fa
rth
e
r
o
b
jec
ts
a
re
fo
c
u
se
d
th
e
n
c
lo
se
r
o
b
jec
ts
g
e
t
b
l
u
r
re
d
in
th
e
ima
g
e
.
To
a
c
h
iev
e
a
n
ima
g
e
wh
e
r
e
a
ll
th
e
o
b
jec
ts
a
re
in
f
o
c
u
s,
t
h
e
p
ro
c
e
ss
o
f
ima
g
e
s
fu
sio
n
is
p
e
rfo
rm
e
d
e
it
h
e
r
in
sp
a
ti
a
l
d
o
m
a
in
o
r
in
tr
a
n
sfo
rm
e
d
d
o
m
a
in
.
In
re
c
e
n
t
ti
m
e
s,
th
e
a
p
p
li
c
a
ti
o
n
s
o
f
ima
g
e
p
ro
c
e
ss
in
g
h
a
v
e
g
ro
w
n
imm
e
n
se
ly
.
Us
u
a
ll
y
d
u
e
t
o
li
m
it
e
d
d
e
p
t
h
o
f
field
o
f
o
p
ti
c
a
l
len
se
s
e
sp
e
c
ially
with
g
re
a
ter
fo
c
a
l
len
g
th
,
it
b
e
c
o
m
e
s
imp
o
ss
ib
le
t
o
o
b
tai
n
a
n
ima
g
e
wh
e
re
a
ll
th
e
o
b
jec
ts
a
re
i
n
fo
c
u
s.
T
h
u
s,
i
t
p
la
y
s
a
n
imp
o
rtan
t
ro
le
t
o
p
e
rfo
rm
o
t
h
e
r
tas
k
s
o
f
ima
g
e
p
r
o
c
e
ss
in
g
su
c
h
a
s
ima
g
e
se
g
m
e
n
tatio
n
,
e
d
g
e
d
e
tec
ti
o
n
,
ste
re
o
m
a
tch
in
g
a
n
d
ima
g
e
e
n
h
a
n
c
e
m
e
n
t.
He
n
c
e
,
a
n
o
v
e
l
fe
a
tu
re
-
lev
e
l
m
u
lt
i
-
fo
c
u
s
ima
g
e
fu
sio
n
t
e
c
h
n
iq
u
e
h
a
s
b
e
e
n
p
r
o
p
o
se
d
wh
ich
f
u
se
s
m
u
lt
i
-
f
o
c
u
s
ima
g
e
s.
Th
u
s,
t
h
e
re
su
lt
s
o
f
e
x
ten
si
v
e
e
x
p
e
rime
n
tatio
n
p
e
rfo
rm
e
d
to
h
i
g
h
li
g
h
t
t
h
e
e
fficie
n
c
y
a
n
d
u
ti
li
ty
o
f
th
e
p
r
o
p
o
se
d
tec
h
n
iq
u
e
is
p
re
se
n
ted
.
Th
e
p
ro
p
o
se
d
wo
rk
f
u
rth
e
r
e
x
p
l
o
re
s
c
o
m
p
a
ris
o
n
b
e
twe
e
n
fu
z
z
y
b
a
se
d
ima
g
e
fu
sio
n
a
n
d
n
e
u
ro
f
u
z
z
y
f
u
sio
n
tec
h
n
iq
u
e
a
lo
n
g
with
q
u
a
li
ty
e
v
a
lu
a
ti
o
n
in
d
ice
s.
K
ey
w
o
r
d
s
:
Feed
f
o
r
war
d
n
eu
r
al
n
etwo
r
k
Fu
s
io
n
Mu
lti
-
f
o
cu
s
im
ag
es
Op
tim
al
b
lo
ck
Var
ian
ce
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
:
S.
Ma
r
y
Pra
v
ee
n
a
,
Dep
ar
tm
en
t o
f
E
C
E
,
Sri
R
am
ak
r
is
h
n
a
I
n
s
titu
te
o
f
T
ec
h
n
o
lo
g
y
,
Pach
ap
alay
am
(
Po
s
t)
,
Per
u
r
C
h
ettip
alay
am
,
C
o
im
b
ato
r
e
–
6
4
1
0
1
0
,
I
n
d
ia
.
p
r
av
ee
n
a.
i
n
f
an
t@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
A
wid
e
v
ar
iety
o
f
d
ata
ac
q
u
i
s
itio
n
d
ev
ices
ar
e
av
ailab
le
a
t
p
r
esen
t,
an
d
h
en
ce
im
a
g
e
f
u
s
io
n
h
as
b
ec
o
m
e
a
n
im
p
o
r
tan
t
s
u
b
ar
ea
o
f
im
ag
e
p
r
o
ce
s
s
in
g
.
T
h
er
e
ar
e
s
en
s
o
r
s
wh
ich
ca
n
n
o
t
g
en
er
ate
im
ag
es
o
f
all
o
b
jects
at
v
ar
i
o
u
s
d
is
tan
ce
s
with
eq
u
al
cla
r
ity
.
T
h
u
s
s
ev
er
al
im
ag
es
o
f
a
s
ce
n
e
ar
e
ca
p
t
u
r
ed
,
with
f
o
cu
s
o
n
d
if
f
er
en
t
p
ar
ts
o
f
it
[
1
]
.
W
ith
th
e
a
v
ailab
ilit
y
o
f
m
u
lti
-
s
en
s
o
r
d
ata
in
m
an
y
f
ield
s
s
u
c
h
as
r
em
o
te
s
en
s
in
g
,
m
ed
ical
im
ag
in
g
,
m
ac
h
i
n
e
v
is
io
n
an
d
m
ilit
ar
y
ap
p
licatio
n
s
,
s
en
s
o
r
f
u
s
io
n
h
as
em
er
g
ed
as
a
n
ew
a
n
d
p
r
o
m
is
in
g
r
esear
c
h
ar
ea
.
T
h
e
cu
r
r
en
t d
ef
in
itio
n
o
f
s
en
s
o
r
f
u
s
io
n
is
v
er
y
b
r
o
ad
an
d
t
h
e
f
u
s
i
o
n
ca
n
ta
k
e
p
lace
at
th
e
s
ig
n
al,
p
i
x
el,
f
ea
tu
r
e,
a
n
d
s
y
m
b
o
l
lev
el.
T
h
e
g
o
al
o
f
im
ag
e
f
u
s
io
n
is
to
cr
ea
te
n
ew
i
m
ag
es
th
at
ar
e
m
o
r
e
s
u
itab
le
f
o
r
th
e
p
u
r
p
o
s
es
o
f
h
u
m
a
n
v
is
u
al
p
e
r
ce
p
tio
n
,
o
b
ject
d
etec
tio
n
an
d
tar
g
et
r
ec
o
g
n
itio
n
.
I
n
th
is
p
ap
er
we
ad
d
r
ess
th
e
p
r
o
b
lem
o
f
p
ix
el
-
l
ev
el
f
u
s
io
n
[
2
,
3]
o
r
th
e
s
o
-
ca
lled
im
ag
e
f
u
s
io
n
p
r
o
b
lem
.
T
o
ac
h
iev
e
an
im
ag
e
wh
er
e
all
th
e
o
b
jects
ar
e
in
f
o
cu
s
,
th
e
p
r
o
ce
s
s
o
f
im
ag
es
f
u
s
io
n
is
p
er
f
o
r
m
ed
eith
er
in
s
p
ec
ial
d
o
m
ain
o
r
in
tr
an
s
f
o
r
m
ed
d
o
m
ain
.
Sp
atial
d
o
m
ain
in
clu
d
es
th
e
tech
n
iq
u
es
wh
ich
d
ir
ec
tly
in
co
r
p
o
r
ate
th
e
p
ix
el
v
alu
es
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
9
,
No
.
3
,
Dec
em
b
e
r
2
0
2
0
:
19
5
–
20
4
196
Mu
lti
-
s
ca
le
o
r
m
u
lti
-
r
eso
l
u
tio
n
ap
p
r
o
ac
h
es
p
r
o
v
id
e
a
m
e
an
s
to
ex
p
lo
it
th
is
f
ac
t.
Af
t
er
ap
p
l
y
in
g
ce
r
tain
o
p
er
atio
n
s
o
n
th
e
tr
a
n
s
f
o
r
m
e
d
im
ag
es,
th
e
f
u
s
ed
im
ag
e
is
cr
ea
ted
b
y
tak
in
g
th
e
in
v
er
s
e
tr
an
s
f
o
r
m
.
I
m
ag
e
f
u
s
io
n
is
g
en
er
ally
p
e
r
f
o
r
m
ed
at
th
r
ee
d
if
f
er
en
t
lev
els
o
f
i
n
f
o
r
m
atio
n
r
ep
r
esen
tatio
n
in
c
lu
d
in
g
p
ix
el
lev
el,
f
ea
tu
r
e
lev
el
an
d
d
ec
is
io
n
lev
el.
I
n
p
ix
el
-
lev
el
im
a
g
e
f
u
s
io
n
,
s
im
p
le
m
ath
em
atica
l
o
p
er
atio
n
s
s
u
ch
as
m
ax
im
u
m
o
r
av
e
r
ag
e
a
r
e
ap
p
lied
o
n
t
h
e
p
ix
el
v
alu
es
o
f
th
e
s
o
u
r
ce
s
to
g
en
e
r
ate
f
u
s
ed
im
a
g
e
[
4
,
5
]
.
Ho
wev
er
th
ese
tech
n
iq
u
es
u
s
u
ally
s
m
o
o
th
th
e
s
h
ar
p
ed
g
es
o
r
leav
e
th
e
b
lu
r
r
in
g
ef
f
ec
ts
in
th
e
f
u
s
ed
im
ag
e.
I
n
th
e
f
ea
tu
r
e
lev
el
m
u
lti
-
f
o
c
u
s
im
ag
e
f
u
s
io
n
,
th
e
s
o
u
r
ce
im
ag
es
ar
e
f
ir
s
t
s
eg
m
en
ted
in
to
d
if
f
er
en
t
r
eg
io
n
s
an
d
th
en
th
e
f
ea
tu
r
e
v
alu
es
o
f
th
ese
r
e
g
io
n
s
ar
e
ca
lcu
lated
.
Usi
n
g
s
o
m
e
f
u
s
io
n
r
u
le,
th
e
r
e
g
io
n
s
ar
e
s
elec
ted
to
g
en
er
ate
th
e
f
u
s
ed
im
ag
e.
I
n
d
ec
is
io
n
l
ev
el
im
ag
e
f
u
s
io
n
,
th
e
o
b
jects
in
th
e
s
o
u
r
ce
im
ag
es
ar
e
f
ir
s
t
d
etec
ted
a
n
d
t
h
en
b
y
u
s
in
g
s
o
m
e
s
u
itab
le
f
u
s
io
n
alg
o
r
ith
m
,
t
h
e
f
u
s
ed
im
ag
e
is
g
en
er
ated
[
6
,
7
]
.
I
n
th
is
p
a
p
er
,
th
er
e
is
a
n
ew
p
r
o
p
o
s
ed
m
eth
o
d
f
o
r
m
u
lti
-
f
o
cu
s
im
ag
e
f
u
s
io
n
2.
F
UZ
Z
Y
B
AS
E
D
I
M
AG
E
F
USI
O
N
Fu
zz
y
im
ag
e
p
r
o
ce
s
s
in
g
is
n
o
t
a
u
n
i
q
u
e
th
e
o
r
y
.
Fu
zz
y
i
m
ag
e
p
r
o
ce
s
s
in
g
is
th
e
co
lle
ctio
n
o
f
all
ap
p
r
o
ac
h
es
th
at
u
n
d
er
s
tan
d
,
r
ep
r
esen
t
an
d
p
r
o
ce
s
s
th
e
im
ag
es,
th
eir
s
eg
m
en
ts
an
d
f
ea
tu
r
es
as
f
u
zz
y
s
et
s
.
T
h
e
r
ep
r
esen
tatio
n
a
n
d
p
r
o
ce
s
s
in
g
d
ep
en
d
o
n
th
e
s
elec
ted
f
u
zz
y
t
ec
h
n
iq
u
e
an
d
o
n
th
e
p
r
o
b
lem
t
o
b
e
s
o
lv
ed
.
I
t
h
as
th
r
ee
m
ain
s
tag
es:
−
I
m
ag
e
f
u
zz
if
icatio
n
(
(
Usi
n
g
m
em
b
er
s
h
ip
f
u
n
ctio
n
s
to
g
r
ap
h
ically
d
escr
ib
e
a
s
itu
atio
n
)
−
Mo
d
if
icatio
n
o
f
m
em
b
e
r
s
h
ip
v
alu
es(A
p
p
licatio
n
o
f
f
u
zz
y
r
u
l
es)
−
I
m
ag
e
d
e
f
u
zz
if
ica
tio
n
(
(
Ob
tain
in
g
th
e
cr
is
p
o
r
ac
tu
al
r
esu
lts
)
T
h
e
co
d
in
g
o
f
im
ag
e
d
ata
(
f
u
zz
if
icatio
n
)
an
d
d
ec
o
d
in
g
o
f
th
e
r
esu
lts
(
d
ef
u
zz
if
icatio
n
)
ar
e
s
tep
s
th
at
m
ak
e
p
o
s
s
ib
le
to
p
r
o
ce
s
s
im
ag
es
with
f
u
zz
y
tech
n
iq
u
es.
T
h
e
m
ain
p
o
wer
o
f
f
u
zz
y
im
a
g
e
p
r
o
ce
s
s
in
g
is
in
th
e
mi
d
d
le
s
tep
(
m
o
d
if
icatio
n
o
f
m
em
b
er
s
h
ip
v
alu
es).
Af
te
r
th
e
im
ag
e
d
ata
ar
e
tr
an
s
f
o
r
m
ed
f
r
o
m
g
r
ay
-
lev
el
p
lan
e
to
th
e
m
e
m
b
er
s
h
ip
p
lan
e
(
f
u
z
zif
icatio
n
)
,
ap
p
r
o
p
r
iate
f
u
zz
y
t
ec
h
n
iq
u
es
m
o
d
if
y
th
e
m
em
b
er
s
h
ip
v
alu
es.
Mu
lti
-
s
en
s
o
r
d
ata
f
u
s
io
n
ca
n
b
e
p
e
r
f
o
r
m
ed
at
f
o
u
r
d
if
f
er
e
n
t
p
r
o
ce
s
s
in
g
lev
els,
ac
co
r
d
in
g
to
th
e
s
tag
e
at
wh
ich
th
e
f
u
s
io
n
tak
es
p
lace
:
s
ig
n
al
le
v
el,
p
ix
el
le
v
el,
f
ea
tu
r
e
lev
el
,
an
d
d
ec
is
io
n
lev
el.
Fig
u
r
e
1
illu
s
tr
ates
o
f
th
e
co
n
ce
p
t o
f
th
e
f
o
u
r
d
if
f
er
e
n
t f
u
s
io
n
lev
els
[
8
-
1
0
]
.
Fig
u
r
e
1
.
An
o
v
er
v
iew
o
f
ca
te
g
o
r
izatio
n
o
f
th
e
f
u
s
io
n
alg
o
r
it
h
m
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
A
n
eu
r
o
fu
z
z
y
ima
g
e
fu
s
io
n
u
s
in
g
b
lo
ck
b
a
s
ed
fea
t
u
r
e
leve
l m
eth
o
d
(
S
.
Ma
r
y
P
r
a
ve
en
a
)
197
Sig
n
al
lev
el
f
u
s
io
n
.
I
n
s
ig
n
al
-
b
ased
f
u
s
io
n
,
s
ig
n
als
f
r
o
m
d
if
f
er
en
t
s
en
s
o
r
s
ar
e
co
m
b
in
ed
t
o
cr
ea
te
a
n
ew
s
ig
n
al
with
a
b
etter
s
ig
n
al
-
to
n
o
is
e
r
atio
t
h
an
th
e
o
r
i
g
in
al
s
ig
n
als.
(
2
)
Pix
el
lev
el
f
u
s
io
n
.
P
ix
el
-
b
ased
f
u
s
io
n
is
p
er
f
o
r
m
ed
o
n
a
p
ix
el
-
by
-
p
i
x
el
b
asis
.
I
t
g
en
er
ates
a
f
u
s
ed
im
ag
e
in
wh
ich
i
n
f
o
r
m
atio
n
ass
o
ciate
d
with
ea
ch
p
ix
el
is
d
eter
m
in
e
d
f
r
o
m
a
s
et
o
f
p
i
x
els
in
s
o
u
r
c
e
im
ag
es
to
im
p
r
o
v
e
th
e
p
er
f
o
r
m
an
ce
o
f
im
ag
e
p
r
o
ce
s
s
in
g
task
s
s
u
ch
as
s
eg
m
en
tati
o
n
(
3
)
Featu
r
e
le
v
el
f
u
s
io
n
.
Featu
r
e
-
b
ased
f
u
s
io
n
at
f
e
atu
r
e
lev
el
r
eq
u
ir
es
an
ex
tr
ac
tio
n
o
f
o
b
jects
r
ec
o
g
n
ized
in
th
e
v
ar
io
u
s
d
ata
s
o
u
r
ce
s
.
I
t
r
eq
u
ir
es
th
e
ex
tr
ac
tio
n
o
f
s
alien
t
f
ea
tu
r
es
wh
ich
ar
e
d
ep
e
n
d
in
g
o
n
t
h
eir
en
v
ir
o
n
m
en
t
s
u
c
h
as
p
ix
el
in
t
en
s
ities
,
ed
g
es
o
r
tex
tu
r
es.
T
h
ese
s
im
ilar
f
ea
tu
r
es
f
r
o
m
in
p
u
t im
ag
es a
r
e
f
u
s
ed
.
2
.
1
.
Ste
ps
in f
uzzy
im
a
g
e
f
us
io
n
T
h
e
o
r
ig
in
al
im
a
g
e
in
th
e
g
r
ay
lev
el
p
lan
e
is
s
u
b
jecte
d
t
o
f
u
zz
if
icatio
n
a
n
d
th
e
m
o
d
i
f
icatio
n
o
f
m
em
b
er
s
h
ip
f
u
n
ctio
n
s
is
ca
r
r
ied
o
u
t
in
th
e
m
em
b
e
r
s
h
ip
p
lan
e.
T
h
e
r
esu
lt
is
th
e
o
u
tp
u
t
i
m
ag
e
o
b
tain
ed
af
te
r
th
e
d
ef
u
zz
if
icatio
n
p
r
o
ce
s
s
.
T
h
e
alg
o
r
ith
m
f
o
r
p
ix
el
-
lev
e
l
im
ag
e
f
u
s
io
n
u
s
in
g
f
u
zz
y
lo
g
ic
is
g
iv
en
as
f
o
llo
ws
[1
1
]
.
a.
R
ea
d
f
ir
s
t im
ag
e
in
v
ar
iab
le
I
1
an
d
f
in
d
its
s
ize
(
r
o
ws:
r
1
,
co
l
u
m
n
s
: c
1
)
.
b.
R
ea
d
s
ec
o
n
d
im
ag
e
in
v
ar
iab
le
I
2
an
d
f
in
d
its
s
ize
(
r
o
ws:
r
2
,
co
lu
m
n
s
: c
2
)
.
c.
Var
iab
les
I
1
an
d
I
2
ar
e
im
a
g
es
in
m
atr
ix
f
o
r
m
wh
e
r
e
e
ac
h
p
ix
el
g
r
ay
lev
el
v
al
u
e
i
s
in
th
e
r
a
n
g
e
f
r
o
m
0
to
2
5
5
.
d.
C
o
m
p
ar
e
r
o
ws
an
d
co
lu
m
n
s
o
f
b
o
th
i
n
p
u
t
im
a
g
es.
I
f
th
ese
t
wo
im
ag
es
ar
e
n
o
t
o
f
th
e
s
am
e
s
ize,
s
elec
t
th
e
p
o
r
tio
n
,
wh
ich
ar
e
o
f
s
am
e
s
iz
e.
e.
C
o
n
v
er
t th
e
im
ag
es in
c
o
lu
m
n
f
o
r
m
wh
ic
h
h
as C
=
r
1
×c
1
en
t
r
ies.
f.
Ma
k
e
a
f
u
zz
y
in
f
er
e
n
ce
s
y
s
tem
f
ile,
wh
ich
h
as two
i
n
p
u
t im
ag
es.
g.
Dec
id
e
n
u
m
b
er
an
d
ty
p
e
o
f
m
em
b
er
s
h
ip
f
u
n
ctio
n
s
f
o
r
b
o
th
th
e
in
p
u
t
im
ag
es
b
y
tu
n
in
g
th
e
m
em
b
er
s
h
i
p
f
u
n
ctio
n
s
.
h.
I
n
p
u
t im
a
g
es in
an
tece
d
e
n
t a
r
e
r
eso
lv
ed
to
a
d
eg
r
ee
o
f
m
em
b
er
s
h
ip
r
an
g
i
n
g
0
to
2
5
5
.
i.
Ma
k
e
f
u
zz
y
i
f
-
th
en
r
u
les
f
o
r
i
n
p
u
t
im
ag
es,
wh
ic
h
r
eso
lv
e
th
o
s
e
two
an
tece
d
en
ts
to
a
s
in
g
l
e
n
u
m
b
e
r
f
r
o
m
0
to
2
5
5
.
I
n
th
e
p
r
o
p
o
s
ed
m
eth
o
d
an
i
m
ag
e
s
et
o
f
1
0
d
if
f
er
e
n
t
im
a
g
es
ar
e
u
s
ed
to
tr
ain
th
e
n
eu
r
al
n
etwo
r
k
.
E
v
er
y
im
ag
e
is
f
ir
s
t
d
iv
id
ed
in
to
n
u
m
b
e
r
o
f
b
lo
ck
s
an
d
f
e
atu
r
es
ar
e
ca
lcu
lated
as
s
h
o
w
n
in
Fig
u
r
e
2
.
T
h
e
b
lo
ck
s
ize
p
lay
s
an
im
p
o
r
tan
t
r
o
le
in
d
is
tin
g
u
is
h
in
g
th
e
b
lu
r
r
ed
an
d
u
n
-
b
lu
r
r
ed
r
eg
i
o
n
s
f
r
o
m
ea
ch
o
th
er
.
Af
ter
d
iv
id
in
g
th
e
im
ag
es
in
to
b
lo
c
k
s
,
th
e
f
ea
tu
r
e
v
al
u
es
o
f
ev
er
y
b
lo
c
k
o
f
all
th
e
im
ag
es
ar
e
ca
lcu
lated
an
d
a
f
ea
tu
r
es
f
ile
is
cr
ea
ted
.
A
s
u
f
f
icien
t
n
u
m
b
er
o
f
f
ea
tu
r
e
v
e
cto
r
s
ar
e
u
s
ed
to
tr
ain
th
e
n
eu
r
al
n
etwo
r
k
.
T
h
e
tr
ain
ed
n
eu
r
al
n
etwo
r
k
is
th
en
u
s
ed
to
f
u
s
e
an
y
s
et
o
f
m
u
lti
-
f
o
cu
s
im
ag
es.
Fig
u
r
e
2
.
B
lo
ck
d
iag
r
am
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
2
.
2
.
Neuro
f
uzzy
ba
s
ed
im
a
g
e
f
us
io
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
9
,
No
.
3
,
Dec
em
b
e
r
2
0
2
0
:
19
5
–
20
4
198
I
n
th
e
f
ield
o
f
a
r
tific
ial
in
tellig
en
ce
,
Neu
r
o
-
Fu
zz
y
r
e
f
er
s
to
co
m
b
in
atio
n
s
o
f
ar
tific
ial
n
eu
r
al
n
etwo
r
k
s
an
d
f
u
zz
y
lo
g
ic.
Ne
u
r
o
-
Fu
zz
y
co
m
p
o
s
ite
r
esu
lts
i
n
a
h
y
b
r
id
in
telli
g
en
t
s
y
s
tem
th
at
s
y
n
er
g
izes
th
ese
two
tech
n
iq
u
es
b
y
c
o
m
b
in
in
g
th
e
h
u
m
a
n
-
lik
e
r
ea
s
o
n
i
n
g
s
ty
le
o
f
f
u
zz
y
s
y
s
tem
s
with
th
e
lear
n
in
g
an
d
co
n
n
ec
tio
n
is
t
s
tr
u
ctu
r
e
o
f
n
e
u
r
al
n
etwo
r
k
s
.
Ne
u
r
o
-
Fu
zz
y
h
y
b
r
id
izatio
n
is
wid
el
y
ter
m
e
d
as
Fu
zz
y
Neu
r
al
Netwo
r
k
(
FNN)
o
r
Neu
r
o
-
Fu
zz
y
Sy
s
tem
(
NFS)
in
t
h
e
l
iter
atu
r
e.
Neu
r
o
-
Fu
zz
y
s
y
s
tem
in
co
r
p
o
r
ates
th
e
h
u
m
an
-
lik
e
r
ea
s
o
n
in
g
s
ty
le
o
f
f
u
zz
y
s
y
s
tem
s
th
r
o
u
g
h
th
e
u
s
e
o
f
f
u
zz
y
s
ets
an
d
a
lin
g
u
is
ti
c
m
o
d
el
co
n
s
is
tin
g
o
f
a
s
et
o
f
I
F
-
T
HE
N
f
u
zz
y
r
u
les.
T
h
e
s
tr
en
g
th
o
f
n
eu
r
o
-
f
u
zz
y
s
y
s
tem
s
in
v
o
lv
es
two
co
n
tr
ad
icto
r
y
r
eq
u
ir
em
e
n
ts
in
f
u
zz
y
Mo
d
elin
g
in
ter
p
r
etab
ilit
y
v
er
s
u
s
ac
cu
r
ac
y
.
I
n
p
r
ac
tice,
o
n
e
o
f
th
e
two
p
r
o
p
e
r
ties
p
r
ev
ails
.
T
h
e
Neu
r
o
-
Fu
zz
y
in
f
u
zz
y
m
o
d
eli
n
g
r
esear
ch
f
ield
is
d
iv
id
ed
i
n
to
two
ar
ea
s
:
lin
g
u
is
tic
f
u
zz
y
m
o
d
elin
g
th
at
is
f
o
cu
s
ed
o
n
i
n
ter
p
r
etab
ilit
y
,
m
ain
ly
th
e
Ma
m
d
an
i
m
o
d
el;
an
d
p
r
ec
is
e
f
u
zz
y
m
o
d
elin
g
th
at
is
f
o
c
u
s
ed
o
n
ac
cu
r
ac
y
,
m
ai
n
ly
th
e
T
a
k
ag
i
-
S
u
g
en
o
-
Kan
g
(
T
SK)
m
o
d
el.
3.
AL
G
O
RI
T
H
M
F
O
R
NE
UR
O
F
UZ
Z
Y
B
AS
E
D
I
M
AG
E
F
USI
O
N
T
h
e
alg
o
r
ith
m
f
o
r
p
ix
el
-
le
v
el
im
ag
e
f
u
s
io
n
u
s
in
g
n
e
u
r
o
f
u
zz
y
lo
g
ic
is
g
iv
en
as f
o
llo
ws.
−
R
ea
d
f
ir
s
t im
ag
e
in
v
ar
iab
le
I
1
an
d
f
in
d
its
s
ize
(
r
o
ws:
zl,
co
lu
m
n
s
: sl).
−
R
ea
d
s
ec
o
n
d
im
ag
e
in
v
ar
iab
le
I
2
an
d
f
in
d
its
s
ize
(
r
o
ws:
z2
.
co
lu
m
n
s
: s2
)
.
−
Var
iab
les
I
1
an
d
I
2
ar
e
im
a
g
es
in
m
atr
ix
f
o
r
m
wh
er
e
ea
c
h
p
ix
el
v
alu
e
is
in
th
e
r
an
g
e
f
r
o
m
0
-
2
5
5
.
Use
Gr
ay
C
o
lo
r
m
ap
.
−
C
o
m
p
ar
e
r
o
ws
an
d
co
lu
m
n
s
o
f
b
o
t
h
in
p
u
t
im
a
g
es.
I
f
th
e
tw
o
im
ag
es
ar
e
n
o
t
o
f
th
e
s
am
e
s
ize,
s
elec
t
th
e
p
o
r
tio
n
.
W
h
ich
ar
e
o
f
s
am
e
s
ize.
−
C
o
n
v
er
t th
e
im
ag
es
in
c
o
lu
m
n
f
o
r
m
wh
ic
h
h
as C
=
zl*
s
l e
n
tr
ies.
−
Fo
r
m
a
tr
ain
in
g
d
ata,
wh
ich
is
a
m
atr
ix
with
th
r
ee
co
lu
m
n
s
an
d
en
tr
ies
in
ea
ch
c
o
lu
m
n
ar
e
f
o
r
m
0
t
o
2
5
5
in
s
tep
s
o
f
1
.
−
Fo
r
m
a
ch
ec
k
d
ata.
W
h
ich
is
a
m
atr
ix
o
f
Pix
els o
f
two
in
p
u
t
im
ag
es in
co
lu
m
n
f
o
r
m
at.
I
n
f
ea
tu
r
e
-
lev
el
im
ag
e
f
u
s
io
n
,
th
e
s
elec
tio
n
o
f
d
if
f
er
en
t
f
ea
tu
r
es
is
an
im
p
o
r
tan
t
task
.
I
n
m
u
lti
-
f
o
cu
s
im
ag
es,
s
o
m
e
o
f
th
e
o
b
jects
ar
e
clea
r
(
in
f
o
cu
s
)
an
d
s
o
m
e
o
b
jects
ar
e
b
lu
r
r
ed
(
o
u
t
o
f
f
o
cu
s
)
.
T
h
e
au
th
o
r
h
as
u
s
ed
f
iv
e
d
if
f
e
r
en
t
f
ea
tu
r
es
to
ch
ar
ac
ter
iz
e
th
e
in
f
o
r
m
atio
n
l
ev
el
co
n
tain
ed
i
n
a
s
p
ec
if
ic
p
o
r
tio
n
o
f
th
e
im
ag
e.
T
h
is
f
ea
tu
r
e
s
et
in
clu
d
es
Var
ian
ce
,
E
n
er
g
y
o
f
Gr
a
d
ien
t,
C
o
n
tr
ast
Vis
ib
ilit
y
,
Sp
atial
Fre
q
u
en
cy
a
n
d
C
an
n
y
E
d
g
e
in
f
o
r
m
atio
n
.
3
.
1
.
F
e
a
t
ures
s
elec
t
io
n
C
o
n
tr
ast
Vis
ib
ilit
y
:
I
t
ca
lcu
la
tes
th
e
d
ev
iatio
n
o
f
a
b
lo
ck
o
f
p
ix
els
f
r
o
m
th
e
b
lo
ck
’
s
m
ea
n
v
alu
e.
T
h
er
ef
o
r
e,
it r
elate
s
to
th
e
clea
r
n
ess
lev
el
o
f
th
e
b
lo
ck
.
T
h
e
v
is
ib
ilit
y
o
f
th
e
im
ag
e
b
l
o
ck
is
o
b
tain
ed
u
s
in
g
(
1
).
V
=
1
×
∑
∑
)
2
=
1
=
1
(
1
)
o
f
wh
er
e,
VI
,
,
m
×
n,
I
(
i,j)
r
e
f
e
r
s
C
o
n
tr
ast
Vi
s
ib
ilit
y
,
m
ea
n
,
s
ize
o
f
th
e
b
lo
ck
B
k
an
d
r
o
ws
an
d
co
lu
m
n
s
o
f
th
e
im
ag
e
r
esp
ec
tiv
ely
.
Var
ian
ce
:
Var
ian
ce
is
u
s
ed
to
m
ea
s
u
r
e
th
e
ex
ten
t
o
f
f
o
c
u
s
in
an
im
a
g
e
b
lo
ck
.
I
t
is
a
m
ath
em
atica
l
ex
p
ec
tatio
n
o
f
th
e
a
v
er
ag
e
s
q
u
ar
ed
d
ev
iatio
n
s
f
r
o
m
th
e
m
ea
n
.
A
p
s
eu
d
o
ce
n
ter
weig
h
ted
lo
ca
l
v
ar
ian
ce
in
t
h
e
n
eig
h
b
o
r
h
o
o
d
o
f
a
n
im
ag
e
p
i
x
el
d
eter
m
in
es
th
e
am
p
lific
atio
n
f
ac
to
r
m
u
ltip
ly
i
n
g
th
e
d
if
f
er
en
ce
b
etwe
en
th
e
im
ag
e
p
ix
el
a
n
d
its
b
lu
r
r
ed
co
u
n
ter
p
a
r
t
b
ef
o
r
e
it
is
co
m
b
in
ed
with
th
e
o
r
i
g
in
al
im
a
g
e.
I
t
is
ca
lcu
lated
u
s
in
g
(
2
)
.
VI
=
1
×
−
)
,
(
)
,
(
j
i
k
k
j
i
I
(
2
)
wh
er
e,
V
is
Va
r
ian
ce
,
is
th
e
m
ea
n
v
al
u
e
o
f
th
e
b
lo
ck
im
a
g
e,
I
(
i,
j)
is
r
o
ws
an
d
co
l
u
m
n
s
o
f
th
e
im
ag
e
an
d
m
×
n
is
th
e
im
ag
e
s
ize.
A
h
ig
h
v
alu
e
o
f
v
ar
ian
ce
s
h
o
ws th
e
g
r
ea
ter
ex
ten
t o
f
f
o
cu
s
in
th
e
im
ag
e
b
lo
ck
.
Sp
atial
Fre
q
u
en
cy
:
Sp
atial
f
r
eq
u
en
cy
m
ea
s
u
r
es
th
e
ac
tiv
ity
lev
el
in
an
im
ag
e.
I
t
is
u
s
ed
t
o
ca
lcu
late
th
e
f
r
eq
u
e
n
cy
c
h
an
g
es
alo
n
g
r
o
ws
an
d
co
lu
m
n
s
o
f
th
e
i
m
a
g
e.
Sp
atial
f
r
eq
u
en
c
y
r
ef
e
r
s
to
th
e
n
u
m
b
er
o
f
p
air
s
On
e
-
th
ir
d
o
f
a
m
illi
m
etr
e
is
a
co
n
v
en
ie
n
t
u
n
it
o
f
r
etin
al
d
is
tan
ce
b
ec
au
s
e
a
n
im
ag
e
th
is
s
ize
is
s
aid
to
s
u
b
ten
d
o
n
e
d
e
g
r
ee
o
f
v
is
u
al
an
g
le
o
n
th
e
r
etin
a.
T
o
g
iv
e
an
ex
a
m
p
le,
in
d
e
x
f
in
g
er
n
ail
ca
s
ts
an
i
m
ag
e
o
f
th
is
s
ize
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
A
n
eu
r
o
fu
z
z
y
ima
g
e
fu
s
io
n
u
s
in
g
b
lo
ck
b
a
s
ed
fea
t
u
r
e
leve
l m
eth
o
d
(
S
.
Ma
r
y
P
r
a
ve
en
a
)
199
wh
en
th
at
n
ail
is
v
iewe
d
at
ar
m
'
s
len
g
th
,
a
ty
p
ical
h
u
m
an
th
u
m
b
,
n
o
t
ju
s
t
th
e
n
ail,
b
u
t
th
e
en
tire
wid
th
,
ca
s
ts
an
im
ag
e
ab
o
u
t
twice
as
b
ig
,
two
d
eg
r
ee
s
o
f
v
is
u
al
a
n
g
le.
T
h
e
s
ize
o
f
th
e
r
etin
al
im
ag
e
ca
s
t
b
y
s
o
m
e
o
b
ject
d
ep
en
d
s
o
n
th
e
d
is
tan
ce
o
f
th
a
t
o
b
ject
f
r
o
m
th
e
ey
e,
as
th
e
d
is
tan
ce
b
etwe
en
th
e
ey
e
an
d
a
n
o
b
ject
d
ec
r
ea
s
es,
th
e
o
b
ject'
s
im
ag
e
s
u
b
ten
d
s
a
g
r
ea
ter
v
is
u
al
a
n
g
le.
T
h
e
u
n
it
em
p
lo
y
e
d
to
ex
p
r
ess
s
p
atial
f
r
eq
u
en
cy
is
th
e
n
u
m
b
er
o
f
c
y
cles th
at
f
all
with
in
o
n
e
d
eg
r
ee
o
f
v
is
u
al
an
g
le.
Sp
atial
f
r
eq
u
en
c
y
is
m
ea
s
u
r
ed
u
s
in
g
(
3
)
.
SF
=
√
(
RF
)
2
+
(
CF
)
2
(
3
)
wh
er
e
,
RF
=
√
1
m
×
n
=
m
i
1
=
n
j
2
[
(
,
)
−
(
,
−
1
)
]
2
CF
=
√
1
×
=
=
n
j
m
i
1
2
[
(
,
)
−
(
,
−
1
)
]
2
wh
er
e,
SF
is
Sp
atial
Fre
q
u
en
cy
,
R
F
is
R
o
w
Fre
q
u
en
cy
,
C
F
i
s
C
o
lu
m
n
Fre
q
u
en
cy
,
m
×
n
is
s
ize
o
f
im
ag
e,
I
(
i,
j
)
is
th
e
r
o
ws an
d
co
lu
m
n
s
o
f
t
h
e
im
ag
e.
E
n
er
g
y
o
f
Gr
a
d
ien
t
(
E
OG)
:
I
t
is
also
u
s
ed
to
m
ea
s
u
r
e
th
e
am
o
u
n
t
o
f
f
o
c
u
s
in
an
.
I
t
is
ca
lcu
lated
u
s
in
g
(
4
)
.
EO
G
=
−
=
−
=
1
1
1
1
m
i
n
j
f
f
j
i
2
2
(
+
)
(
4
)
wh
er
e,
)
,
(
)
,
1
(
j
i
f
j
i
f
f
i
−
+
=
)
,
(
)
1
,
(
j
i
f
j
i
f
f
j
−
+
=
E
OG
is
E
n
er
g
y
o
f
Gr
ad
ien
t
f
i
E
n
er
g
y
o
f
th
e
r
o
w,
f
j
is
th
e
E
n
e
r
g
y
o
f
th
e
co
lu
m
n
,
m
×
n
is
t
h
e
s
ize
o
f
th
e
im
ag
e.
E
d
g
e
I
n
f
o
r
m
atio
n
:
T
h
e
ed
g
e
p
ix
els ca
n
b
e
f
o
u
n
d
in
th
e
im
ag
e
b
lo
ck
b
y
u
s
in
g
C
an
n
y
ed
g
e
d
etec
to
r
.
I
t
r
etu
r
n
s
1
if
th
e
cu
r
r
en
t p
ix
el
b
elo
n
g
s
to
s
o
m
e
ed
g
e
in
th
e
im
ag
e
o
th
er
wis
e
it r
etu
r
n
s
0
.
T
h
e
e
d
g
e
f
ea
tu
r
e
is
ju
s
t
th
e
n
u
m
b
er
o
f
ed
g
e
p
ix
els
co
n
tain
ed
with
in
th
e
im
ag
e
b
lo
ck
.
E
d
g
e
d
etec
tio
n
is
a
f
u
n
d
am
en
tal
to
o
l
in
im
ag
e
p
r
o
ce
s
s
in
g
an
d
co
m
p
u
ter
v
is
i
o
n
,
p
ar
ticu
lar
l
y
in
th
e
ar
ea
s
o
f
f
ea
tu
r
e
d
etec
tio
n
an
d
f
ea
tu
r
e
ex
tr
ac
tio
n
,
wh
ich
aim
at
id
en
tify
i
n
g
p
o
in
ts
in
a
d
ig
ital
im
ag
e
at
w
h
ich
th
e
im
ag
e
b
r
ig
h
tn
es
s
ch
an
g
es
s
h
ar
p
l
y
o
r
m
o
r
e
f
o
r
m
ally
h
as d
is
co
n
tin
u
ities
.
T
h
e
p
u
r
p
o
s
e
o
f
d
etec
tin
g
s
h
ar
p
ch
an
g
es in
im
ag
e
b
r
ig
h
tn
ess
is
to
ca
p
tu
r
e
im
p
o
r
tan
t e
v
en
ts
an
d
ch
an
g
es
in
p
r
o
p
er
ties
o
f
th
e
wo
r
ld
.
E
d
g
es
ex
tr
ac
ted
f
r
o
m
n
o
n
-
tr
iv
ial
im
ag
es
ar
e
o
f
ten
h
am
p
e
r
ed
b
y
f
r
ag
m
en
tatio
n
,
m
ea
n
in
g
t
h
at
t
h
e
ed
g
e
cu
r
v
es
ar
e
n
o
t
co
n
n
ec
ted
,
m
is
s
in
g
ed
g
e
s
eg
m
en
ts
as
well
as
f
alse
ed
g
es
n
o
t
co
r
r
esp
o
n
d
in
g
to
in
ter
es
tin
g
p
h
en
o
m
en
a
in
t
h
e
im
a
g
e,
th
u
s
co
m
p
licatin
g
t
h
e
s
u
b
s
eq
u
en
t
task
o
f
in
ter
p
r
etin
g
t
h
e
im
ag
e
d
ata.
E
d
g
e
d
etec
tio
n
is
o
n
e
o
f
th
e
f
u
n
d
a
m
en
tal
s
tep
s
in
im
ag
e
p
r
o
ce
s
s
in
g
,
im
a
g
e
an
aly
s
is
,
im
ag
e
p
atter
n
r
ec
o
g
n
itio
n
,
an
d
c
o
m
p
u
ter
v
is
io
n
tec
h
n
iq
u
es.
3
.
2
.
Art
if
ici
a
l neura
l net
wo
rk
Ar
tific
ial
n
eu
r
al
n
etwo
r
k
s
(
A
NNs)
h
av
e
p
r
o
v
en
to
b
e
a
m
o
r
e
p
o
wer
f
u
l
an
d
s
elf
-
a
d
ap
tiv
e
m
eth
o
d
o
f
p
atter
n
r
ec
o
g
n
itio
n
as
c
o
m
p
ar
ed
to
tr
ad
itio
n
al
lin
ea
r
a
n
d
s
im
p
le
n
o
n
lin
ea
r
an
aly
s
es.
T
h
e
ANN
-
b
ased
m
eth
o
d
em
p
lo
y
s
a
n
o
n
lin
ea
r
r
esp
o
n
s
e
f
u
n
ctio
n
th
at
iter
ates m
an
y
ti
m
es in
a
s
p
ec
ial
n
etwo
r
k
s
tr
u
ctu
r
e
in
o
r
d
e
r
to
le
ar
n
th
e
co
m
p
le
x
f
u
n
ctio
n
al
r
elatio
n
s
h
ip
b
etwe
en
in
p
u
t
an
d
o
u
tp
u
t
tr
ain
i
n
g
d
ata.
T
h
e
in
p
u
t
lay
er
h
as
s
ev
e
r
al
n
eu
r
o
n
s
,
wh
ich
r
e
p
r
esen
t
th
e
f
ea
tu
r
e
f
ac
t
o
r
s
ex
tr
ac
ted
a
n
d
n
o
r
m
alize
d
f
r
o
m
im
ag
e
A
an
d
im
ag
e
B
.
T
h
e
h
id
d
en
lay
er
h
as
s
ev
e
r
al
n
eu
r
o
n
s
an
d
th
e
o
u
t
p
u
t
la
y
er
h
as
o
n
e
n
eu
r
o
n
(
o
r
m
o
r
e
n
eu
r
o
n
)
.
T
h
e
i
th
n
eu
r
o
n
o
f
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
9
,
No
.
3
,
Dec
em
b
e
r
2
0
2
0
:
19
5
–
20
4
200
in
p
u
t la
y
er
c
o
n
n
ec
ts
with
th
e
j
th
n
eu
r
o
n
o
f
th
e
h
i
d
d
en
la
y
er
b
y
weig
h
t Wi
j,
an
d
weig
h
t b
et
wee
n
th
e
jth
n
eu
r
o
n
o
f
th
e
h
id
d
e
n
lay
er
an
d
t
h
e
t
t
h
n
eu
r
o
n
o
f
o
u
tp
u
t
lay
er
is
V
jt
(
in
th
is
ca
s
e
t
=
1
)
.
T
h
e
weig
h
tin
g
f
u
n
ctio
n
is
u
s
ed
to
s
im
u
late
an
d
r
ec
o
g
n
iz
e
th
e
r
esp
o
n
s
e
r
elatio
n
s
h
ip
b
etwe
en
f
ea
tu
r
es
o
f
f
u
s
ed
im
ag
e
an
d
co
r
r
esp
o
n
d
in
g
f
ea
tu
r
e
f
r
o
m
o
r
ig
in
al
im
a
g
es
(
im
ag
e
A
an
d
im
ag
e
B
)
.
A
s
th
e
f
ir
s
t
s
tep
o
f
ANN
-
b
ase
d
d
ata
f
u
s
io
n
,
two
r
eg
is
ter
ed
im
ag
es
ar
e
d
e
co
m
p
o
s
ed
in
to
s
ev
er
al
b
lo
ck
s
with
s
ize
o
f
M
an
d
N
.
T
h
en
,
f
ea
t
u
r
es
o
f
th
e
co
r
r
esp
o
n
d
in
g
b
lo
c
k
s
in
th
e
t
wo
o
r
ig
i
n
al
im
ag
es
a
r
e
ex
tr
ac
ted
,
an
d
th
e
n
o
r
m
alize
d
f
ea
t
u
r
e
v
ec
to
r
in
cid
en
t
to
n
eu
r
al
n
etwo
r
k
s
ca
n
b
e
co
n
s
tr
u
cted
.
T
h
e
f
ea
tu
r
es
u
s
ed
h
er
e
to
ev
alu
ate
th
e
f
u
s
io
n
ef
f
ec
t
a
r
e
n
o
r
m
ally
s
p
atial
f
r
eq
u
e
n
cy
,
v
is
ib
ilit
y
,
an
d
ed
g
e.
T
h
e
n
ex
t
s
tep
is
to
s
elec
t
s
o
m
e
v
ec
to
r
s
am
p
les
to
tr
ain
n
eu
r
al
n
etwo
r
k
s
.
An
ANN
is
a
u
n
iv
er
s
al
f
u
n
ctio
n
ap
p
r
o
x
i
m
ato
r
th
at
d
ir
ec
tly
ad
ap
ts
to
an
y
n
o
n
lin
ea
r
f
u
n
ctio
n
d
ef
in
ed
b
y
a
r
ep
r
esen
ta
tiv
e
s
et
o
f
tr
ain
i
n
g
d
ata.
On
ce
t
r
ain
ed
,
th
e
ANN
m
o
d
el
ca
n
r
em
e
m
b
er
a
f
u
n
cti
o
n
al
r
elatio
n
s
h
i
p
an
d
b
e
u
s
ed
f
o
r
f
u
r
th
er
ca
lcu
latio
n
s
.
Fo
r
t
h
ese
r
ea
s
o
n
s
,
th
e
AN
N
co
n
ce
p
t
h
as
b
ee
n
ad
o
p
ted
t
o
d
e
v
elo
p
s
tr
o
n
g
l
y
n
o
n
lin
ea
r
m
o
d
els f
o
r
m
u
ltip
le
s
en
s
o
r
s
d
ata
f
u
s
io
n
.
3
.
3
.
F
ee
d f
o
rwa
rd
neura
l net
wo
rk
A
n
eu
r
o
n
ca
n
h
a
v
e
a
n
y
n
u
m
b
er
o
f
in
p
u
ts
f
r
o
m
o
n
e
to
n
,
w
h
er
e
n
is
th
e
t
o
tal
n
u
m
b
er
o
f
i
n
p
u
ts
.
T
h
e
in
p
u
ts
m
ay
b
e
r
ep
r
esen
ted
th
e
r
ef
o
r
e
as
x
1
,
x
2
,
x
3
…
x
n
.
An
d
th
e
co
r
r
esp
o
n
d
i
n
g
weig
h
ts
f
o
r
th
e
in
p
u
ts
as
w
1
,
w
2
,
w
3
…
w
n.
,
th
e
s
u
m
m
atio
n
o
f
th
e
weig
h
ts
m
u
ltip
lied
b
y
t
h
e
in
p
u
ts
is
x
1
w
1
+
x
2
w
2
+
x
3
w
3
….
+
x
n
w
n
.
Hen
ce
,
a
=
x
1
w
1
+x
2
w
2
+x
3
w
3
.
.
.
+x
n
w
n
.
Ass
u
m
in
g
an
ar
r
ay
o
f
in
p
u
ts
an
d
weig
h
ts
ar
e
alr
ea
d
y
in
itialized
as
x
[
n
]
an
d
w[
n
]
th
en
double activation = 0;
for (int i=0; i<n; i++)
{
activation += x[i] * w[i];
}
if
th
e
ac
tiv
atio
n
>
t
h
r
esh
o
ld
,
o
u
tp
u
t is 1
an
d
if
ac
tiv
atio
n
<
t
h
r
esh
o
ld
o
u
tp
u
t is 0
.
O
n
e
way
o
f
is
b
y
o
r
g
an
is
in
g
th
e
n
eu
r
o
n
s
in
to
a
d
esig
n
ca
lled
a
f
ee
d
f
o
r
war
d
n
etwo
r
k
.
I
t
g
ets
its
n
am
e
f
r
o
m
t
h
e
way
th
e
n
eu
r
o
n
s
in
ea
ch
lay
er
f
ee
d
th
eir
o
u
t
p
u
t
f
o
r
war
d
to
th
e
n
e
x
t
lay
er
u
n
til
we
g
et
th
e
f
in
al
o
u
tp
u
t
f
r
o
m
th
e
n
eu
r
al
n
etwo
r
k
.
A
f
ee
d
f
o
r
war
d
n
e
u
r
al
n
etw
o
r
k
is
f
ir
s
t
tr
ai
n
ed
with
t
h
e
b
l
o
ck
f
ea
t
u
r
es
o
f
ten
p
air
s
o
f
m
u
lti
-
f
o
cu
s
im
ag
es.
A
f
ea
tu
r
e
s
et
in
clu
d
in
g
s
p
atial
f
r
eq
u
en
c
y
,
co
n
tr
ast
v
is
ib
ilit
y
,
ed
g
es,
v
ar
ian
ce
an
d
en
er
g
y
o
f
g
r
a
d
ien
t
is
u
s
ed
to
d
ef
in
e
th
e
clar
ity
o
f
th
e
im
ag
e
b
lo
ck
.
B
lo
ck
s
ize
is
d
eter
m
in
ed
ad
a
p
tiv
ely
f
o
r
ea
ch
im
ag
e.
T
h
e
tr
ain
ed
n
eu
r
a
l n
etwo
r
k
is
th
en
u
s
ed
to
f
u
s
e
an
y
p
air
o
f
m
u
lti
-
f
o
cu
s
im
ag
es
.
3
.
4
.
Q
ua
ntit
a
t
iv
e
m
ea
s
ures
T
h
er
e
ar
e
d
if
f
e
r
en
t
q
u
an
titativ
e
m
ea
s
u
r
es
wh
ich
a
r
e
u
s
ed
to
ev
alu
ate
th
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
f
u
s
io
n
tech
n
iq
u
es.
W
e
u
s
ed
th
r
ee
m
e
asu
r
es
R
o
o
t
Me
an
Sq
u
ar
e
E
r
r
o
r
(
R
MSE
)
,
Peak
Sig
n
al
to
No
is
e
R
atio
(
PS
NR
)
an
d
en
tr
o
p
y
(
He)
.
3
.
4
.
1
.
Ro
o
t
m
e
a
n sq
ua
re
er
ro
r
T
h
e
an
aly
tical
p
er
f
o
r
m
an
ce
s
tu
d
ies
wer
e
aim
ed
to
q
u
an
titativ
ely
ass
ess
im
ag
e
f
u
s
io
n
p
er
f
o
r
m
an
ce
i
n
a
s
tr
aig
h
tf
o
r
war
d
m
an
n
er
.
T
h
e
r
o
o
t
m
ea
n
s
q
u
ar
e
er
r
o
r
(
R
MSE
)
,
d
ef
in
ed
b
y
th
e
d
ev
i
atio
n
s
b
etwe
en
th
e
r
ef
er
en
ce
im
a
g
e
p
ix
el
v
alu
e
R
(
i,
j)
an
d
th
e
f
u
s
ed
im
ag
e
p
i
x
el
v
alu
e
F(i,
j)
,
is
co
m
p
u
ted
as
=
√
∑
∑
[
(
,
)
−
(
,
)
]
2
=
1
=
1
×
(
6
)
wh
er
e
m
×
n
is
th
e
in
p
u
t
im
a
g
e
s
ize.
I
f
t
h
e
v
alu
e
o
f
0
c
o
r
r
esp
o
n
d
to
th
e
c
o
m
p
lete
im
a
g
e
r
ec
o
n
s
tr
u
ctio
n
f
o
r
b
lo
ck
m
×
n
,
it
is
a
p
er
f
ec
t
im
ag
e,
wh
ich
h
as
b
ee
n
ac
h
iev
e
d
th
r
o
u
g
h
ac
cu
r
ate
r
ec
o
n
s
t
r
u
ctio
n
o
f
m
u
lti
f
o
cu
s
to
th
e
r
ef
er
en
ce
im
ag
e.
3
.
4
.
2
.
E
ntr
o
py
E
n
tr
o
p
y
is
k
n
o
wn
to
b
e
a
m
ea
s
u
r
e
o
f
th
e
a
m
o
u
n
t o
f
u
n
ce
r
tai
n
ly
ab
o
u
t th
e
im
ag
e
.
I
t is th
en
g
iv
en
b
y
=
−
∑
2
−
1
=
0
(
5
)
wh
er
e
L
is
th
e
n
u
m
b
er
o
f
g
r
ay
lev
els.
3
.
4
.
3
.
P
ea
k
s
ig
na
l t
o
no
is
e
ra
t
io
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
A
n
eu
r
o
fu
z
z
y
ima
g
e
fu
s
io
n
u
s
in
g
b
lo
ck
b
a
s
ed
fea
t
u
r
e
leve
l m
eth
o
d
(
S
.
Ma
r
y
P
r
a
ve
en
a
)
201
I
t
d
eter
m
in
es
th
e
d
eg
r
ee
o
f
r
e
s
em
b
lan
ce
b
etwe
en
r
ef
er
e
n
ce
an
d
f
u
s
ed
im
ag
e.
A
b
i
g
g
er
v
alu
e
s
h
o
w
g
o
o
d
f
u
s
io
n
r
esu
lt.
=
20
10
[
2
1
×
∑
∑
[
(
,
)
−
(
,
)
]
2
=
1
=
1
]
(
7
)
4.
P
E
RF
O
RM
A
NCE
E
VA
L
U
AT
I
O
N
T
h
e
in
p
u
t
im
ag
e
is
d
iv
id
ed
i
n
t
o
b
lo
ck
s
an
d
th
e
f
i
v
e
f
ea
tu
r
es
ar
e
ex
tr
ac
ted
u
s
in
g
f
ee
d
f
o
r
wa
r
d
n
eu
r
al
n
etwo
r
k
.
T
h
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
e
x
is
tin
g
Av
er
a
g
e,
Ma
x
i
m
u
m
,
Mi
n
im
u
m
an
d
PC
A
b
ased
tech
n
iq
u
es
ar
e
co
m
p
ar
ed
with
th
e
r
esu
lts
o
f
t
h
e
p
r
o
p
o
s
ed
tech
n
iq
u
e.
T
h
e
e
x
p
er
im
en
tatio
n
r
esu
lts
ar
e
o
b
tai
n
ed
to
ev
alu
ate
th
e
p
er
f
o
r
m
an
ce
o
f
th
e
p
r
o
p
o
s
ed
tech
n
iq
u
e.
T
h
e
s
im
u
latio
n
is
ca
r
r
ied
o
u
t
b
y
u
s
in
g
Ma
t
lab
7
.
5
,
th
e
s
im
u
latio
n
win
d
o
w
is
as
s
h
o
wn
in
th
e
Fig
u
r
e
3
(
h
)
.
T
h
e
Fig
u
r
e
3
(
a,
b
,
c,
d
,
e
an
d
f
)
ar
e
th
e
r
esu
lts
o
f
th
e
f
u
s
ed
im
ag
e
b
y
v
ar
io
u
s
f
u
s
io
n
tech
n
iq
u
es
s
u
ch
as
av
er
a
g
in
g
,
m
in
im
u
m
,
m
ax
im
u
m
,
PC
A
an
d
b
lo
ck
-
b
ased
f
ea
tu
r
e
lev
el
m
eth
o
d
r
esp
ec
tiv
el
y
.
T
ab
le
I
s
h
o
ws
th
e
en
tr
o
p
y
o
f
th
e
v
ar
io
u
s
m
eth
o
d
s
an
d
f
ig
u
r
e
5
s
h
o
ws
th
e
g
r
ap
h
ical
p
er
f
o
r
m
an
ce
o
f
t
h
e
s
am
e
f
r
o
m
wh
ich
it
ca
n
b
e
s
ee
n
th
at
t
h
e
p
r
o
p
o
s
ed
tech
n
iq
u
e
g
iv
es
b
etter
r
esu
lts
th
an
th
e
p
r
ev
io
u
s
m
et
h
o
d
s
.
E
v
alu
atio
n
m
ea
s
u
r
es
ar
e
u
s
ed
to
ev
alu
ate
th
e
q
u
ality
o
f
th
e
f
u
s
ed
im
a
g
e.
T
h
e
f
u
s
ed
im
ag
es
ar
e
ev
alu
ated
,
ta
k
in
g
t
h
e
f
o
llo
win
g
p
ar
am
eter
s
in
to
co
n
s
i
d
er
atio
n
.
(
a)
(
b
)
(
c)
(
d
)
(
e)
(f)
(
g
)
(
h
)
Fig
u
r
e
3
.
Me
d
ical
im
a
g
es (
C
T
an
d
MRI)
f
u
s
ed
b
y
f
u
zz
y
lo
g
i
c
an
d
n
e
u
r
o
f
u
zz
y
lo
g
ic
.
(
a
)
I
n
p
u
t CT
im
ag
e,
(
b
)
I
n
p
u
t M
R
I
im
ag
e
,
(
c)
Fu
s
ed
b
y
av
er
ag
i
n
g
,
(
d
)
Fu
s
ed
b
y
m
in
im
u
m
,
(
e)
Fu
s
ed
b
y
m
ax
im
u
m
,
(
f
)
Fu
s
ed
b
y
PC
A,
(
g
)
Fu
s
ed
b
y
p
r
o
p
o
s
ed
m
eth
o
d
,
(
h
)
Simu
latio
n
win
d
o
w
T
ab
le
1
.
R
esu
lts
o
f
q
u
a
n
titativ
e
m
ea
s
u
r
es o
f
m
e
d
ical
im
ag
es
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
9
,
No
.
3
,
Dec
em
b
e
r
2
0
2
0
:
19
5
–
20
4
202
Q
u
a
n
t
i
t
a
t
i
v
e
M
e
a
s
u
r
e
s
A
v
e
r
a
g
e
M
i
n
i
m
u
m
M
a
x
i
m
u
m
P
C
A
B
l
o
c
k
B
a
se
d
M
e
t
h
o
d
En
t
r
o
p
y
(
d
B
)
6
.
3
8
4
.
2
5
5
.
8
3
8
.
8
6
2
0
.
7
8
R
M
S
E
6
2
.
6
7
7
4
.
7
6
7
4
.
7
6
2
1
.
0
2
3
2
.
2
3
P
S
N
R
(
d
B
)
1
6
.
5
8
1
6
.
2
0
1
6
.
2
0
1
7
.
4
3
4
.
5
3
4
.
1
.
I
ma
g
e
qu
a
lity
ind
ex
I
m
ag
e
q
u
ality
in
d
e
x
(
I
QI
)
m
ea
s
u
r
es
th
e
s
im
ilar
ity
b
etwe
e
n
two
im
ag
es
(
1
1
&
I
2
)
an
d
its
v
alu
e
r
an
g
es f
r
o
m
-
1
to
1
.
I
QI
is
eq
u
al
to
1
if
b
o
th
im
ag
es
ar
e
id
en
t
ical
.
4
.
2
.
M
utua
l
info
rma
t
io
n m
e
a
s
ure
Mu
tu
al
in
f
o
r
m
atio
n
m
ea
s
u
r
e
(
MI
M)
f
u
r
n
is
h
es
th
e
am
o
u
n
t
o
f
in
f
o
r
m
atio
n
o
f
o
n
e
im
ag
e
in
an
o
th
er
.
T
h
is
g
iv
es th
e
g
u
id
elin
es f
o
r
s
elec
tin
g
th
e
b
est f
u
s
io
n
m
eth
o
d
.
Giv
en
two
im
a
g
es
M (
i,
j)
a
n
d
N
(
i,
j)
.
W
h
er
e,
P
M
(
x
)
an
d
P
N
(
y
)
ar
e
th
e
p
r
o
b
ab
ilit
y
d
en
s
ity
f
u
n
cti
o
n
s
in
th
e
in
d
iv
i
d
u
al
im
ag
es,
an
d
P
MN
(
x,
y)
is
jo
in
t p
r
o
b
ab
ilit
y
d
en
s
ity
f
u
n
cti
o
n
.
4
.
3
.
F
us
io
n
f
a
ct
o
r
Giv
en
two
im
ag
es A
an
d
B
,
an
d
th
eir
f
u
s
ed
im
ag
e
F,
t
h
e
Fu
s
io
n
f
ac
to
r
(
FF
)
.
FF
=
I
AF
+
I
BF
(
8)
W
h
er
e
I
AF
an
d
I
BF
ar
e
th
e
MI
M
v
alu
es
b
etwe
en
in
p
u
t
im
ag
es
an
d
f
u
s
ed
im
a
g
e.
A
h
ig
h
er
v
alu
e
o
f
FF
in
d
icate
s
th
at
f
u
s
ed
im
ag
e
c
o
n
tain
s
m
o
d
er
ately
g
o
o
d
am
o
u
n
t
o
f
in
f
o
r
m
atio
n
p
r
esen
t
i
n
b
o
t
h
th
e
im
ag
es.
Ho
wev
er
,
a
h
i
g
h
v
al
u
e
o
f
FF
d
o
es n
o
t im
p
ly
t
h
at
th
e
in
f
o
r
m
a
tio
n
f
r
o
m
b
o
t
h
im
ag
es is
s
y
m
m
etr
ically
f
u
s
ed
.
4
.
3
.
1
.
F
us
io
n
s
y
m
m
et
r
y
Fu
s
io
n
s
y
m
m
etr
y
(
FS
)
is
an
i
n
d
icatio
n
o
f
th
e
d
e
g
r
ee
o
f
s
y
m
m
etr
y
in
th
e
in
f
o
r
m
atio
n
co
n
ten
t
f
r
o
m
b
o
th
th
e
im
ag
es.
T
h
e
q
u
ality
o
f
f
u
s
io
n
tech
n
i
q
u
e
d
ep
en
d
s
o
n
th
e
d
e
g
r
ee
o
f
Fu
s
io
n
s
y
m
m
e
tr
y
.
Sin
ce
FS
is
th
e
s
y
m
m
etr
y
f
ac
to
r
,
wh
e
n
th
e
s
e
n
s
o
r
s
ar
e
o
f
g
o
o
d
q
u
ality
,
FS
s
h
o
u
ld
b
e
as
lo
w
as
p
o
s
s
ib
le
s
o
th
at
th
e
f
u
s
ed
im
ag
e
d
er
iv
es
f
ea
tu
r
es
f
r
o
m
b
o
th
in
p
u
t
im
ag
es
[
11
-
1
4
]
.
I
f
a
n
y
o
f
th
e
s
en
s
o
r
s
is
o
f
lo
w
q
u
ality
th
en
it
is
b
etter
to
m
ax
im
ize
FS
th
an
m
in
im
izi
n
g
it.
4
.
3
.
2
.
F
us
io
n
ind
ex
T
h
e
s
tu
d
y
p
r
o
p
o
s
es
a
p
ar
am
et
er
ca
lled
Fu
s
io
n
in
d
e
x
f
r
o
m
t
h
e
f
ac
to
r
s
Fu
s
io
n
s
y
m
m
etr
y
a
n
d
Fu
s
io
n
f
ac
to
r
.
T
h
e
f
u
s
io
n
i
n
d
ex
(
FI)
i
s
d
ef
in
ed
as
FI
=
I
AF
/
I
BF
(
9
)
I
s
th
e
m
u
tu
al
i
n
f
o
r
m
atio
n
in
d
ex
b
etwe
en
m
u
ltis
p
ec
tr
al
im
a
g
e
an
d
f
u
s
ed
im
a
g
e
a
n
d
I
BF
is
th
e
m
u
tu
al
in
f
o
r
m
atio
n
in
d
e
x
b
etwe
en
p
a
n
ch
r
o
m
atic
im
a
g
e
an
d
f
u
s
ed
i
m
ag
e.
T
h
e
q
u
ality
o
f
f
u
s
io
n
te
ch
n
iq
u
e
d
ep
en
d
s
o
n
th
e
d
eg
r
ee
o
f
f
u
s
io
n
in
d
ex
.
W
h
er
e
p
c
o
n
tain
s
th
e
h
is
to
g
r
am
,
co
u
n
ts
r
etu
r
n
ed
f
r
o
m
im
h
is
t.
5.
R
E
SU
L
T
S AN
D
D
I
SCU
SS
I
O
NS
T
h
er
e
ar
e
m
an
y
t
y
p
ical
ap
p
licatio
n
s
f
o
r
im
a
g
e
f
u
s
io
n
.
Mo
d
e
r
n
s
p
ec
tr
al
s
ca
n
n
er
s
g
ath
er
u
p
to
s
ev
er
al
h
u
n
d
r
ed
o
f
s
p
ec
tr
al
b
a
n
d
s
wh
ich
ca
n
b
e
b
o
th
v
is
u
alize
d
an
d
p
r
o
ce
s
s
ed
in
d
iv
id
u
ally
,
o
r
wh
ich
ca
n
b
e
f
u
s
ed
in
to
a
s
in
g
le
im
ag
e,
d
ep
e
n
d
in
g
o
n
t
h
e
im
ag
e
a
n
aly
s
is
task
.
I
n
th
is
s
ec
tio
n
,
in
p
u
t
im
ag
es
ar
e
f
u
s
ed
u
s
in
g
f
u
zz
y
lo
g
ic
ap
p
r
o
ac
h
[
15
-
1
7
]
.
So
,
it
is
co
n
clu
d
ed
th
at
r
esu
lts
o
b
tain
ed
f
r
o
m
th
e
im
p
lem
en
tati
o
n
o
f
n
e
u
r
o
f
u
zz
y
lo
g
ic
-
b
ased
im
ag
e
f
u
s
io
n
ap
p
r
o
ac
h
p
er
f
o
r
m
s
b
etter
f
o
r
f
ir
s
t
two
tes
t
ca
s
e
s
an
d
f
u
zz
y
b
ased
im
ag
e
f
u
s
io
n
s
h
o
ws
b
etter
p
er
f
o
r
m
a
n
ce
f
o
r
th
ir
d
test
ca
s
e.
So
f
u
r
th
er
in
v
esti
g
atio
n
is
n
ee
d
ed
to
r
eso
l
v
e
th
is
is
s
u
e.
Ou
r
ex
p
er
im
en
tal
r
esu
lts
s
h
o
w
th
a
t
n
eu
r
o
f
u
zz
y
l
o
g
ic
-
b
ased
im
a
g
e
f
u
s
io
n
a
p
p
r
o
ac
h
p
r
o
v
id
es
b
etter
p
er
f
o
r
m
an
ce
wh
en
co
m
p
ar
e
d
to
f
u
zz
y
b
ased
im
ag
e
f
u
s
io
n
f
o
r
f
ir
s
t
two
ex
am
p
les.
I
m
ag
e
q
u
ality
in
d
ex
(
I
QI
)
,
th
e
s
im
ilar
ity
b
etwe
en
r
ef
e
r
en
ce
a
n
d
f
u
s
ed
im
ag
e
(
0
.
9
9
9
9
,
0
.
9
8
2
9
,
an
d
0
.
3
1
8
2
)
ar
e
h
ig
h
er
f
o
r
f
ir
s
t
two
ca
s
es
wh
e
n
co
m
p
ar
ed
to
v
al
u
es
o
b
tain
ed
f
r
o
m
f
u
zz
y
b
ased
f
u
s
io
n
tech
n
iq
u
e
(
0
.
9
7
5
8
,
0
.
9
8
2
4
,
an
d
0
.
8
8
7
1
)
.
T
h
e
h
ig
h
e
r
v
alu
es
f
o
r
f
u
s
io
n
f
ac
t
o
r
(
FF
)
f
r
o
m
f
ir
s
t
two
ex
am
p
les
(
2
.
8
1
1
5
,
3
.
3
4
3
8
,
1
.
0
1
0
9
)
o
b
tain
ed
f
r
o
m
th
e
n
eu
r
o
f
u
zz
y
b
ased
f
u
s
io
n
ap
p
r
o
ac
h
in
d
icat
es
th
at
f
u
s
ed
i
m
ag
e
c
o
n
tain
s
m
o
d
er
ately
g
o
o
d
a
m
o
u
n
t
o
f
in
f
o
r
m
atio
n
p
r
esen
t
in
b
o
th
th
e
im
ag
es
co
m
p
ar
ed
to
FF
v
alu
es
(
1
.
0
9
6
5
,
2
.
1
3
2
9
,
1
.
9
8
6
4
)
o
b
tain
e
d
f
r
o
m
f
u
zz
y
b
ased
f
u
s
io
n
ap
p
r
o
ac
h
.
T
h
e
am
o
u
n
t
o
f
in
f
o
r
m
atio
n
o
f
o
n
e
im
ag
e
in
an
o
t
h
er
,
m
u
tu
al
in
f
o
r
m
atio
n
m
ea
s
u
r
e
(
MI
M)
v
al
u
es
(
1
.
4
6
5
6
,
1
.
5
0
7
9
,
0
.
7
6
3
4
)
ar
e
a
ls
o
s
ig
n
if
ican
tly
b
etter
wh
ic
h
s
h
o
ws
th
at
n
eu
r
o
f
u
zz
y
b
ased
f
u
s
io
n
m
eth
o
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
A
n
eu
r
o
fu
z
z
y
ima
g
e
fu
s
io
n
u
s
in
g
b
lo
ck
b
a
s
ed
fea
t
u
r
e
leve
l m
eth
o
d
(
S
.
Ma
r
y
P
r
a
ve
en
a
)
203
p
r
eser
v
es
m
o
r
e
in
f
o
r
m
atio
n
c
o
m
p
ar
ed
t
o
f
u
zz
y
b
ased
im
ag
e
f
u
s
io
n
.
T
h
e
o
th
er
ev
alu
atio
n
m
ea
s
u
r
es
lik
e
r
o
o
t
m
ea
n
s
q
u
ar
e
er
r
o
r
(
R
MSE
)
with
lo
wer
an
d
p
ea
k
s
ig
n
al
to
n
o
is
e
r
atio
(
PS
NR
)
,
C
o
r
r
elatio
n
C
o
ef
f
icien
t
(
C
C
)
with
h
ig
h
er
v
alu
es
(
0
.
9
4
5
9
,
0
.
8
9
7
9
,
0
.
1
2
6
5
)
o
b
tain
ed
f
o
r
m
n
eu
r
o
f
u
zz
y
b
ased
f
u
s
io
n
ap
p
r
o
ac
h
a
r
e
also
co
m
p
ar
ativ
ely
b
etter
f
o
r
f
ir
s
t
two
ca
s
es.
T
h
e
en
tr
o
p
y
,
t
h
e
am
o
u
n
t
o
f
i
n
f
o
r
m
atio
n
t
h
at
ca
n
b
e
u
s
e
d
to
ch
ar
ac
ter
ize
th
e
in
p
u
t
im
a
g
e
(
7
.
2
7
5
7
,
7
.
3
2
0
2
,
4
.
4
8
9
4
)
ar
e
b
etter
f
o
r
two
ex
am
p
les
o
b
tai
n
e
d
f
r
o
m
n
eu
r
o
f
u
zz
y
b
ased
im
ag
e
f
u
s
io
n
tech
n
iq
u
e.
Fig
u
r
e
4
s
h
o
ws
th
e
Me
d
ica
l
im
ag
es
(
C
T
a
n
d
MRI
B
r
ain
)
f
u
s
ed
b
y
d
if
f
er
e
n
t
im
a
g
es
f
u
s
io
n
tech
n
iq
u
es
an
d
t
h
e
p
r
o
p
o
s
ed
m
eth
o
d
.
Fig
u
r
e
a
a
n
d
b
a
r
e
th
e
in
p
u
t
C
T
an
d
MRI
im
ag
es
r
esp
ec
tiv
ely
.
T
ab
le
2
s
h
o
ws
th
e
r
esu
lts
o
f
q
u
an
tita
tiv
e
m
ea
s
u
r
es
o
f
m
ed
ical
im
ag
es
s
u
ch
as
b
r
ain
.
T
h
e
v
alu
e
o
f
ea
c
h
q
u
ality
ass
es
s
m
en
t p
ar
am
eter
s
o
f
all
m
en
tio
n
ed
f
u
s
io
n
a
p
p
r
o
ac
h
es
ar
e
d
ep
icted
in
T
ab
le
3
.
(
a)
(
b
)
(
c)
(
d
)
(
e)
(f)
(
g
)
Fig
u
r
e
4
.
Me
d
ical
im
a
g
es (
C
T
an
d
MRI
B
r
ain
)
f
u
s
ed
b
y
d
if
f
er
en
t im
ag
e
f
u
s
io
n
tech
n
i
q
u
es
an
d
th
e
p
r
o
p
o
s
ed
m
eth
o
d
.
(
a)
I
n
p
u
t CT
im
ag
e,
(
b
)
I
n
p
u
t M
R
I
im
ag
e,
(
c)
Fu
s
ed
b
y
av
e
r
ag
in
g
,
(
d
)
Fu
s
ed
b
y
m
in
im
u
m
,
(
e
)
Fu
s
ed
b
y
m
ax
im
u
m
,
(
f
)
Fu
s
ed
b
y
PC
A,
(
g
)
Fu
s
ed
b
y
co
n
tr
ast
T
ab
le
2
.
R
esu
lts
o
f
q
u
a
n
titativ
e
m
ea
s
u
r
es o
f
m
e
d
ical
im
ag
es
(
b
r
ain
)
S
.
N
O
M
ETH
O
D
S
EN
TR
O
P
Y
(
d
B
)
1.
A
v
e
r
a
g
i
n
g
6
.
8
9
8
1
2.
M
i
n
i
m
u
m
2
.
0
9
0
6
3.
max
i
mu
m
6
.
7
5
8
2
4.
P
C
A
8
.
4
2
7
1
5.
B
l
o
c
k
b
a
s
e
d
f
e
a
t
u
r
e
l
e
v
e
l
me
t
h
o
d
(
P
r
o
p
o
se
d
)
8
.
9
4
2
7
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
9
,
No
.
3
,
Dec
em
b
e
r
2
0
2
0
:
19
5
–
20
4
204
T
ab
le
3
.
E
v
alu
atio
n
in
d
ices f
o
r
im
ag
e
f
u
s
io
n
b
ased
o
n
f
u
zz
y
an
d
n
e
u
r
o
f
u
zz
y
l
o
g
ic
ap
p
r
o
ac
h
es
M
ETH
O
D
I
Q
I
M
I
M
FF
FS
FI
R
M
S
E
P
S
N
R
E
N
T
R
O
P
Y
CC
SF
F
U
ZZY
F
U
S
I
O
N
(
EX
1
)
0
.
8
7
5
8
0
.
4
3
2
8
1
.
0
9
6
5
0
.
0
7
7
9
0
.
7
3
0
3
4
4
.
8
4
4
8
1
5
.
0
9
6
6
4.
4
8
8
1
0
.
5
8
3
3
1
0
.
4
5
6
7
(
EX
1
)
(
EX
2
)
0
.
8
8
2
4
0
.
8
5
8
2
2
.
1
3
7
9
0
.
0
5
1
8
0
.
8
1
2
2
4
8
.
9
9
3
5
1
4
.
3
2
8
0
5.
3
9
8
1
0
.
7
5
9
8
1
6
.
9
7
4
9
(
EX
2
)
(
EX
3
)
0
.
8
8
7
1
1
.
5
5
7
6
1
.
9
8
6
4
0
.
2
8
4
1
3
.
6
3
2
5
4
2
.
4
8
5
0
1
5
.
5
6
6
1
5.
8
2
0
5
0
.
7
7
6
1
2
5
.
4
7
1
1
(
EX
3
)
N
EU
R
O
F
U
ZZY
F
U
S
I
O
N
(
EX
1
)
0
.
9
9
9
9
1
.
3
6
5
6
2
.
8
1
1
5
0
.
0
2
1
3
1
.
0
8
3
7
1
4
.
9
5
5
4
2
4
.
6
3
4
8
7.
2
7
5
7
0
.
9
4
5
9
2
5
.
5
6
9
8
0
.
9
9
9
9
0
.
9
7
2
9
1
.
3
6
5
6
1
.
5
0
7
9
2
.
8
1
1
5
3
.
3
4
3
8
0
.
0
2
1
3
0
.
0
4
9
0
1
.
0
8
3
7
0
.
8
2
1
3
1
4
.
9
5
5
4
3
4
.
4
6
6
2
2
4
.
6
3
4
8
1
7
.
3
8
2
9
7.
2
7
5
7
7.
3
2
0
2
0
.
9
4
5
9
0
.
8
9
7
9
2
5
.
5
6
9
8
3
7
.
1
1
6
9
(
EX
2
)
0
.
9
7
2
9
0
.
2
1
8
2
1
.
5
0
7
9
0
.
7
6
3
4
3
.
3
4
3
8
1
.
0
1
0
9
0
.
0
4
9
0
0
.
2
4
3
1
0
.
8
2
1
3
2
.
8
9
2
6
3
4
.
4
6
6
2
6
8
.
5
3
1
9
1
7
.
3
8
2
9
1
1
.
4
1
2
9
7.
3
2
0
2
4.
4
8
9
4
0
.
8
9
7
9
0
.
1
2
6
5
3
7
.
1
1
6
9
1
6
.
9
9
2
6
(
EX
3
)
6.
CO
NCLU
SI
O
N
In
t
h
is
p
a
p
e
r,
b
l
o
c
k
-
b
a
se
d
fe
a
tu
re
-
lev
e
l
m
u
lt
i
-
f
o
c
u
s
ima
g
e
f
u
sio
n
t
e
c
h
n
iq
u
e
is
p
r
o
p
o
se
d
f
o
r
f
u
sin
g
i
m
a
g
e
s
th
a
t
a
re
n
o
t
in
fo
c
u
s.
A
fe
e
d
fo
rwa
rd
n
e
u
ra
l
n
e
two
r
k
is
first
train
e
d
wit
h
th
e
b
l
o
c
k
fe
a
tu
re
s
o
f
a
p
a
ir
o
f
m
u
lt
i
-
fo
c
u
s
ima
g
e
s.
A
fe
a
tu
re
se
t
in
c
lu
d
in
g
sp
a
ti
a
l
fre
q
u
e
n
c
y
,
c
o
n
tras
t
v
isi
b
il
it
y
,
e
d
g
e
s,
v
a
rian
c
e
a
n
d
e
n
e
rg
y
o
f
g
ra
d
ien
t
is
u
se
d
to
d
e
fi
n
e
th
e
c
larity
o
f
th
e
ima
g
e
b
lo
c
k
.
Bl
o
c
k
siz
e
is
d
eter
m
in
ed
a
d
a
p
ti
v
e
ly
f
o
r
e
a
c
h
ima
g
e
.
T
h
e
train
e
d
n
e
u
ra
l
n
e
two
rk
is
t
h
e
n
u
se
d
to
f
u
se
a
n
y
p
a
ir
o
f
m
u
lt
i
-
fo
c
u
s
i
m
a
g
e
s.
Th
e
p
e
rf
o
rm
a
n
c
e
o
f
t
h
e
e
x
isti
n
g
A
v
e
ra
g
e
,
M
a
x
imu
m
,
M
in
imu
m
a
n
d
P
CA
b
a
se
d
tec
h
n
iq
u
e
s a
re
c
o
m
p
a
re
d
with
th
e
re
su
lt
s o
f
th
e
p
ro
p
o
se
d
tec
h
n
iq
u
e
s.
Th
e
e
x
p
e
rime
n
tatio
n
re
su
lt
s a
re
o
b
tain
e
d
to
e
v
a
lu
a
te
th
e
p
e
rf
o
rm
a
n
c
e
o
f
t
h
e
p
ro
p
o
se
d
tec
h
n
i
q
u
e
.
Ex
p
e
rime
n
tatio
n
re
su
lt
s
sh
o
w
th
a
t
th
e
p
ro
p
o
se
d
tec
h
n
iq
u
e
p
e
rfo
rm
s
b
e
t
ter
th
a
n
th
e
e
x
isti
n
g
tec
h
n
iq
u
e
s.
T
h
e
e
x
p
e
rime
n
tal
re
su
lt
s
c
lea
rly
sh
o
w
t
h
a
t
t
h
e
p
ro
p
o
se
d
ima
g
e
f
u
si
o
n
u
si
n
g
fu
z
z
y
l
o
g
ic
g
iv
e
s a
c
o
n
si
d
e
ra
b
le i
m
p
ro
v
e
m
e
n
t
o
n
t
h
e
q
u
a
li
ty
o
f
th
e
f
u
sio
n
sy
ste
m
a
n
d
n
e
u
r
o
fu
z
z
y
b
a
se
d
ima
g
e
fu
sio
n
p
re
se
rv
e
s m
o
re
tex
tu
re
i
n
fo
rm
a
ti
o
n
.
RE
F
E
R
E
NC
E
S
[1
]
H.
Li
,
S
.
M
a
n
j
u
n
a
t
h
a
n
d
S
.
K.
M
it
ra
.
,
“
M
u
lt
i
-
se
n
s
o
r
ima
g
e
f
u
si
o
n
u
sin
g
t
h
e
wa
v
e
let
tra
n
sfo
rm
,
”
in
Gr
a
p
h
ica
l
M
o
d
e
ls a
n
d
Im
a
g
e
Pro
c
e
ss
in
g
,
v
o
l.
5
7
,
n
o
.
3
,
p
p
.
2
3
5
-
2
4
5
,
1
9
9
5
.
[2
]
P
.
J.
Bu
rt
,
R.
J.
Lo
lez
y
n
sk
i
.
,
“
E
n
h
a
n
c
e
d
ima
g
e
c
a
p
tu
re
th
r
o
u
g
h
fu
sio
n
,
”
in
:
P
r
o
c
e
e
d
i
n
g
s
o
f
t
h
e
F
o
u
rth
I
n
tern
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Co
m
p
u
ter
Vi
si
o
n
,
Be
rli
n
,
G
e
rm
a
n
y
,
v
o
l.
3
,
p
p
.
1
7
3
-
1
8
2
,
1
9
9
3
.
[3
]
Yu
fe
n
g
Zh
e
n
g
,
Ed
wa
rd
A.
Esso
c
k
,
Bru
c
e
C.
,
“
Ha
n
se
n
.
An
A
d
v
a
n
c
e
d
Im
a
g
e
F
u
si
o
n
Alg
o
rit
h
m
Ba
se
d
o
n
Wav
e
let
Tran
sfo
rm
–
In
c
o
rp
o
ra
ti
o
n
wit
h
P
CA
a
n
d
m
o
rp
h
o
l
o
g
ica
l
P
ro
c
e
ss
in
g
,
”
Pro
c
e
e
d
in
g
s
o
f
t
h
e
S
PIE
,
v
o
l.
5
2
9
8
,
p
p
.
1
7
7
-
1
8
7
,
2
0
0
4
.
[4
]
I.
Blo
c
h
,
“
I
n
fo
rm
a
ti
o
n
c
o
m
b
in
a
ti
o
n
o
p
e
ra
t
o
rs
fo
r
d
a
ta
f
u
sio
n
.
a
re
v
iew
with
c
las
sifica
ti
o
n
,
”
IEE
E
T
ra
n
s.
S
M
C:
Pa
rt
A
,
v
o
l.
2
6
,
p
p
.
5
2
-
6
7
,
1
9
9
6
.
[5
]
H.
A.
El
to
u
k
h
y
,
S
.
Ka
v
u
si,
“
A
c
o
m
p
u
tat
io
n
a
ll
y
e
fficie
n
t
a
l
g
o
rit
h
m
fo
r
m
u
lt
i
-
f
o
c
u
s
ima
g
e
re
c
o
n
stru
c
ti
o
n
,
”
P
ro
c
e
e
d
in
g
s o
f
S
PIE
El
e
c
tro
n
ic I
ma
g
in
g
,
v
o
l.
3
3
,
p
p
.
4
5
-
5
1
,
2
0
0
3
.
[6
]
Ab
d
u
l
Ba
sit
S
id
d
iq
u
i,
M
.
Arfa
n
Ja
ffa
r.
,
“
Ay
y
a
z
Hu
ss
a
in
a
n
d
An
wa
r
M
.
M
irza
,
Blo
c
k
-
b
a
se
d
F
e
a
tu
re
-
lev
e
l
M
u
lt
i
-
fo
c
u
s Im
a
g
e
F
u
si
o
n
,
”
IEE
E
in
si
g
n
a
l
p
ro
c
e
ss
in
g
,
v
o
l.
2
,
p
p
.
1
3
4
-
1
3
9
,
2
0
1
0
.
[7
]
Li
,
B
.
M
a
n
ju
n
a
th
,
S
.
M
it
ra
,
“
M
u
l
ti
se
n
so
r
ima
g
e
fu
si
o
n
u
sin
g
t
h
e
w
a
v
e
let
tran
sfo
rm
,
G
ra
p
h
,
”
M
o
d
e
ls
Ima
g
e
Pr
o
c
e
ss
,
v
o
l.
5
7
,
n
o
.
3
,
p
p
.
2
3
5
-
2
4
5
,
1
9
9
5
.
[8
]
Ish
it
a
De
,
Bh
a
b
a
t
o
sh
Ch
a
n
d
a
.
,
“
A
sim
p
le
a
n
d
e
ffi
c
ien
t
a
lg
o
rit
h
m
fo
r
m
u
l
ti
fo
c
u
s
ima
g
e
fu
sio
n
u
si
n
g
m
o
rp
h
o
lo
g
ica
l
wa
v
e
lets,”
IEE
E
in
S
i
g
n
a
l
Pr
o
c
e
ss
in
g
,
v
o
l.
3
1
,
p
p
.
9
2
4
-
9
3
6
,
2
0
0
6
.
[9
]
G
o
n
z
a
lo
P
a
jare
s
a
n
d
Je
su
s
M
a
n
u
e
l
d
e
la
Cr
u
z
,
“
A
wa
v
e
let
-
b
a
se
d
Im
a
g
e
F
u
sio
n
Tu
t
o
rial,
”
I
EE
E
i
n
P
a
tt
e
r
n
Rec
o
g
n
it
io
n
,
v
o
l.
3
7
,
n
o
.
9
,
p
p
.
1
8
5
5
-
1
8
7
2
,
2
0
0
4
.
[1
0
]
H.
J.
He
ij
m
a
n
s
a
n
d
J.
G
o
u
tsias
,
“
No
n
li
n
e
a
r
m
u
lt
ires
o
l
u
ti
o
n
si
g
n
a
l
d
e
c
o
m
p
o
siti
o
n
sc
h
e
m
e
s.
II.
M
o
rp
h
o
lo
g
ica
l
wa
v
e
lets,”
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Ima
g
e
Pro
c
e
ss
in
g
,
v
o
l.
9
,
p
p
.
1
8
9
7
-
1
9
1
3
,
2
0
0
0
.
[1
1
]
Ya
n
g
,
X.H.,
Hu
a
n
g
,
F
.
Z.
,
Li
u
,
G
.
“
Urb
a
n
Re
m
o
te
Im
a
g
e
F
u
sio
n
Us
in
g
F
u
z
z
y
Ru
les
,
”
IEE
E
Pr
o
c
e
e
d
in
g
s
o
f
t
h
e
Ei
g
h
t
h
I
n
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
M
a
c
h
in
e
L
e
a
rn
i
n
g
a
n
d
Cy
b
e
rn
e
ti
c
s
,
p
p
.
1
0
1
-
1
0
9
,
2
0
0
9
.
[1
2
]
A.
To
e
t,
“
Im
a
g
e
fu
sio
n
b
y
a
ra
ti
o
o
f
lo
w
p
a
ss
p
y
ra
m
id
,
”
Pa
tt
e
rn
Rec
o
g
n
it
i
o
n
L
e
tt
e
rs
,
v
o
l.
9
,
n
o
.
4
,
p
p
.
2
4
5
-
2
5
3
,
1
9
8
9
.
[1
3
]
Xin
m
a
n
Zh
a
n
g
,
Jiu
q
lan
g
Ha
n
a
n
d
P
e
ifei
Li
u
,
“
Re
sto
ra
ti
o
n
a
n
d
fu
sio
n
o
p
ti
m
iza
ti
o
n
sc
h
e
m
e
o
f
m
u
lt
ifo
c
u
s
ima
g
e
u
sin
g
g
e
n
e
ti
c
se
a
rc
h
stra
teg
ies
,
”
Op
ti
c
a
A
p
p
l
ica
ta
io
n
,
v
o
l
.
XX
XV
,
n
o
.
4
,
2
0
0
5
.
[1
4
]
X.
Ya
n
g
,
W
.
Ya
n
g
,
J.
P
e
i,
“
Diffe
re
n
t
fo
c
u
s
p
o
in
ts
ima
g
e
s
fu
si
o
n
b
a
se
d
o
n
wa
v
e
let
d
e
c
o
m
p
o
siti
o
n
,
”
Pro
c
e
e
d
in
g
s
o
f
T
h
ird
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
In
f
o
rm
a
ti
o
n
Fu
si
o
n
,
v
o
l.
1
,
p
p
.
3
-
8
,
2
0
0
0
.
[1
5
]
S
.
M
u
k
h
o
p
a
d
h
y
a
y
,
B.
Ch
a
n
d
a
,
“
F
u
sio
n
o
f
2
d
g
ra
y
sc
a
le
ima
g
e
s
u
sin
g
m
u
lt
isc
a
le
m
o
rp
h
o
lo
g
y
,
”
P
a
t
tern
Re
c
o
g
n
it
i
o
n
,
v
o
l.
3
4
,
p
p
.
1
9
3
9
-
1
9
4
9
,
2
0
0
1
.
[1
6
]
V.
P
.
S
.
Na
id
u
a
n
d
J
.
R.
Ra
o
l
,
“
P
ix
e
l
-
lev
e
l
Im
a
g
e
F
u
sio
n
u
si
n
g
Wav
e
lets
a
n
d
P
r
in
c
ip
a
l
C
o
m
p
o
n
e
n
t
An
a
ly
sis,”
i
n
De
fen
c
e
S
c
ien
c
e
J
o
u
rn
a
l
,
v
o
l.
5
8
,
n
o
.
3
,
p
p
.
3
3
8
-
3
5
2
,
M
a
y
2
0
0
8
.
[1
7
]
Oth
m
a
n
Kh
a
li
fa
,
“
Wav
e
let
Co
d
i
n
g
De
sig
n
fo
r
Im
a
g
e
Da
ta Co
m
p
re
ss
io
n
,
”
In
ter
n
a
ti
o
n
a
l
Ara
b
J
o
u
rn
a
l
o
f
In
f
o
rm
a
ti
o
n
T
e
c
h
n
o
l
o
g
y
,
v
o
l.
2
,
n
o
.
2
,
p
p
.
1
1
8
-
1
2
8
,
2
0
0
5
.
Evaluation Warning : The document was created with Spire.PDF for Python.