T
E
L
KO
MNIK
A
, V
ol
.
17
,
No.
6,
Dec
e
mb
er
20
1
9,
p
p.
2
94
8
~
2958
IS
S
N: 1
69
3
-
6
93
0
,
accr
ed
ited
F
irst
Gr
ad
e b
y K
em
en
r
istekdikti,
Decr
ee
No: 2
1/E/
K
P
T
/20
18
DOI:
10.12928/TE
LK
OM
N
IK
A
.v
1
7
i
6
.
12408
◼
29
48
Rec
ei
v
ed
Dec
e
mb
er
13
,
20
1
8
; R
ev
i
s
ed
F
eb
r
ua
r
y
8
,
20
1
9
;
A
c
c
ep
ted
Ma
r
c
h
12
,
20
1
9
Pr
og
ress
i
on
a
p
proac
h f
or i
ma
g
e den
oisin
g
Bilal
Ch
ar
mo
u
t
i
*
1
, Ahm
ad
Kadri
Jun
o
h
2
, M
o
h
d
Y
u
s
o
f
f
Ma
sho
r
3
,
Naja
h
G
h
az
a
li
4
,
Ma
h
yun
Ab
W
ah
ab
5
,
W
an
Z
u
ki
Az
man
W
an Muh
a
m
ad
6
,
Z
ainab Y
ahya
7
,
Ab
d
es
se
l
am Be
r
o
u
al
8
1
,2
,4
,6
,7
In
s
t
i
tu
t
e
o
f
E
n
g
i
n
e
e
r
i
n
g
M
a
th
e
m
a
ti
c
s
,
Un
i
v
e
r
s
i
t
i
M
a
l
a
y
s
i
a
P
e
rl
i
s
(Un
i
M
AP),
Pe
r
l
i
s
,
M
a
l
a
y
s
i
a
3
Sc
h
o
o
l
o
f
M
e
c
h
a
tro
n
i
c
,
U
n
i
v
e
rs
i
ti
M
a
l
a
y
s
i
a
Pe
rl
i
s
(Un
i
M
AP),
Pe
rl
i
s
,
M
a
l
a
y
s
i
a
5
Sc
h
o
o
l
o
f
En
v
i
ro
n
m
e
n
ta
l
E
n
g
i
n
e
e
ri
n
g
,
Un
i
v
e
r
s
i
t
i
M
a
l
a
y
s
i
a
P
e
rl
i
s
(Un
i
M
AP),
Pe
r
l
i
s
,
M
a
l
a
y
s
i
a
8
Dep
a
rt
m
e
n
t
o
f
Co
m
p
u
t
e
r Sc
i
e
n
c
e
,
In
te
r
n
a
ti
o
n
a
l
I
s
l
a
m
i
c
Un
i
v
e
rs
i
t
y
M
a
l
a
y
s
i
a
,
M
a
l
a
y
s
i
a
*C
o
rre
s
p
o
n
d
i
n
g
a
u
th
o
r,
e
-
m
a
i
l
:
d
y
b
i
l
e
l
@y
a
h
o
o
.f
r
Ab
strac
t
Rem
o
v
i
n
g
n
o
i
s
e
fro
m
th
e
i
m
a
g
e
b
y
re
t
a
i
n
i
n
g
th
e
d
e
t
a
i
l
s
a
n
d
f
e
a
tu
r
e
s
o
f
t
h
i
s
tre
a
t
e
d
i
m
a
g
e
re
m
a
i
n
s
a
s
ta
n
d
i
n
g
c
h
a
l
l
e
n
g
e
fo
r
th
e
r
e
s
e
a
rc
h
e
rs
i
n
th
i
s
fi
e
l
d
.
T
h
e
re
f
o
re
,
th
i
s
s
tu
d
y
i
s
c
a
rri
e
d
o
u
t
t
o
p
ro
p
o
s
e
a
n
d
i
m
p
l
e
m
e
n
t
a
n
e
w
d
e
n
o
i
s
i
n
g
te
c
h
n
i
q
u
e
fo
r
re
m
o
v
i
n
g
i
m
p
u
l
s
e
n
o
i
s
e
fro
m
t
h
e
d
i
g
i
ta
l
i
m
a
g
e
,
u
s
i
n
g
a
n
e
w
wa
y
.
Th
i
s
t
e
c
h
n
i
q
u
e
p
e
rm
i
t
s
th
e
n
a
rro
w
i
n
g
o
f
t
h
e
g
a
p
b
e
twe
e
n
th
e
o
ri
g
i
n
a
l
a
n
d
th
e
re
s
to
r
e
d
i
m
a
g
e
s
,
v
i
s
u
a
l
l
y
a
n
d
q
u
a
n
ti
t
a
ti
v
e
l
y
b
y
a
d
o
p
ti
n
g
th
e
m
a
th
e
m
a
ti
c
a
l
c
o
n
c
e
p
t
'
'
a
ri
th
m
e
ti
c
p
ro
g
r
e
s
s
i
o
n
'
'
.
Th
ro
u
g
h
th
i
s
p
a
p
e
r,
th
i
s
c
o
n
c
e
p
t
i
s
i
n
t
e
g
ra
t
e
d
i
n
t
o
th
e
i
m
a
g
e
d
e
n
o
i
s
i
n
g
,
d
u
e
to
i
ts
a
b
i
l
i
ty
i
n
m
o
d
e
l
l
i
n
g
th
e
v
a
ri
a
ti
o
n
o
f
p
i
x
e
l
s
’
i
n
t
e
n
s
i
t
y
i
n
th
e
i
m
a
g
e
.
T
h
e
p
r
i
n
c
i
p
l
e
o
f
th
e
p
ro
p
o
s
e
d
d
e
n
o
i
s
i
n
g
te
c
h
n
i
q
u
e
re
l
i
e
s
o
n
th
e
p
re
c
i
s
i
o
n
,
whe
r
e
i
t
k
e
e
p
s
t
h
e
u
n
c
o
rru
p
te
d
p
i
x
e
l
s
b
y
u
s
i
n
g
e
ff
e
c
ti
v
e
n
o
i
s
e
d
e
te
c
t
i
o
n
a
n
d
c
o
n
v
e
rt
s
th
e
c
o
rru
p
te
d
p
i
x
e
l
s
b
y
r
e
p
l
a
c
i
n
g
t
h
e
m
wit
h
o
th
e
r
c
l
o
s
e
s
t
p
i
x
e
l
s
fro
m
th
e
o
ri
g
i
n
a
l
i
m
a
g
e
a
t
l
o
wer
c
o
s
t
a
n
d
wit
h
m
o
re
s
i
m
p
l
i
c
i
t
y
.
Key
w
ords
:
a
ri
th
m
e
ti
c
p
ro
g
re
s
s
i
o
n
,
d
e
n
o
i
s
i
n
g
t
e
c
h
n
i
q
u
e
,
i
m
a
g
e
,
i
m
a
g
e
p
ro
c
e
s
s
i
n
g
,
i
m
p
u
l
s
e
n
o
i
s
e
,
n
o
i
s
e
Copy
righ
t
©
2
0
1
9
Uni
v
e
rsi
t
a
s
Ahm
a
d
D
a
hl
a
n.
All
rig
ht
s
r
e
s
e
rve
d
.
1.
Int
r
o
d
u
ctio
n
Ima
g
e
i
s
c
on
s
i
d
ered
as
a
p
owerfu
l
pl
atf
orm
t
o
c
arr
y
an
d
to
tr
an
s
m
i
t
i
nf
orma
ti
o
n
be
twee
n
pe
op
l
e,
w
he
r
e
i
t
i
s
v
ery
i
mp
orta
nt
i
n
a
l
ot
of
f
i
e
l
ds
s
uc
h
as
bi
ol
o
gy
,
as
tr
on
o
my
,
i
nd
us
tr
i
al
,
me
d
i
c
al
an
d
s
urv
ei
l
l
an
c
e
[
1].
T
h
us
,
i
t
att
r
ac
t
s
th
e
att
en
ti
on
of
a
l
ot
of
r
es
ea
r
c
h
ers
i
n
r
es
tori
n
g
the
u
nk
no
w
n
or
i
gi
na
l
i
ma
g
e
fr
om
the
de
grade
d
i
ma
ge
c
au
s
ed
by
an
y
f
ac
tors
tha
t
ma
y
d
eg
r
ad
e
or
r
ed
uc
e
th
e
i
ma
ge
qu
al
i
ty
(
e.g
.
b
l
ur)
.
O
ne
of
the
fac
t
ors
i
s
th
e
no
i
s
e
th
at
ma
y
b
e
i
ntrod
uc
ed
i
n
the
i
ma
ge
wi
t
h m
an
y
f
orms
(
ad
di
t
i
v
e,
mu
l
ti
pl
i
c
ati
v
e a
nd
i
m
pu
l
s
i
v
e), thro
ug
h o
ne
of
the
s
e
ph
as
es
:
i
ma
ge
ac
qu
i
s
i
ti
on
,
tr
an
s
m
i
s
s
i
on
or
s
torag
e
[2]
.
T
he
no
i
s
e
i
s
a
pa
r
as
i
ti
c
or
wei
r
d
i
n
format
i
on
th
at
aff
ec
ts
th
e
v
i
s
ua
l
as
p
ec
t
o
f
the
i
m
ag
e
by
c
ha
n
gi
ng
t
he
p
i
x
el
v
a
l
u
es
an
d
i
t
ma
k
es
the
ot
he
r
s
ub
s
eq
ue
nt
i
m
ag
e
proc
es
s
i
ng
s
uc
h
as
s
eg
m
en
ta
ti
o
n,
c
om
pres
s
i
on
,
a
na
l
y
s
es
,
ex
tr
ac
ti
o
n
of
i
nfo
r
ma
t
i
on
,
c
l
as
s
i
f
i
c
ati
on
an
d
etc
mo
r
e
d
i
ffi
c
u
l
t
.
T
he
i
ma
ge
i
s
c
orr
up
te
d
d
ue
t
o
the
fac
t
t
ha
t
the
r
e
are
v
ario
us
ty
pe
s
o
f
n
oi
s
e
s
uc
h
as
the
G
au
s
s
i
a
n
no
i
s
e,
P
oi
s
s
on
no
i
s
e,
S
pe
c
k
l
e
no
i
s
e,
S
a
l
t
an
d
P
ep
pe
r
n
oi
s
e
an
d m
a
n
y
mo
r
e f
u
nd
a
me
nta
l
no
i
s
e t
y
pe
s
i
n t
he
c
as
e
of
d
i
g
i
ta
l
i
ma
ge
s
[
3].
It
s
ho
ul
d
b
e
no
t
ed
tha
t
th
e
i
m
ag
e
r
es
torat
i
o
n
c
on
c
ep
t
v
ari
es
ac
c
ord
i
ng
t
o
the
d
eg
r
ad
ati
on
f
ac
tor
wh
ere
i
t
mi
g
ht
b
e
de
bl
urr
i
ng
[4]
,
i
np
a
i
nt
i
n
g
[5]
an
d
etc
.
In
the
c
as
e
of
de
gra
da
t
i
on
c
a
us
ed
by
t
h
e
no
i
s
e,
the
i
ma
g
e
r
es
torat
i
o
n
c
an
al
s
o
be
c
a
l
l
ed
as
i
m
ag
e
de
n
oi
s
i
ng
,
no
i
s
e
r
em
ov
al
,
or
no
i
s
e
r
ed
uc
ti
on
,
w
he
r
e
i
t
w
as
fi
r
s
t
l
y
i
ntrodu
c
e
d
by
W
i
e
ne
r
a
n
d
K
ol
mo
go
r
ov
i
n
the
19
4
0s
[2
].
Henc
e,
De
n
oi
s
i
n
g
i
m
ag
e
i
s
a
c
r
i
t
i
c
a
l
a
n
d
pri
ma
r
y
ph
as
e
i
n
t
he
i
ma
ge
proc
es
s
i
ng
(
prepr
oc
es
s
i
ng
p
ha
s
e),
a
i
mi
n
g
to
r
em
ov
e
or
r
ed
uc
e
the
n
oi
s
e
fr
o
m
th
e
no
i
s
y
i
ma
ge
by
pres
erv
i
ng
the
i
ma
ge
fea
t
ures
,
us
i
ng
the
v
ario
us
t
ec
hn
i
qu
es
(
f
i
l
ters
)
.
T
he
o
bs
erv
ed
i
ma
ge
(
e.g
.
ph
ot
og
r
a
ph
,
c
h
art)
w
as
i
n
i
ti
al
l
y
d
i
g
i
ti
z
e
d
a
nd
s
tored
i
n
the
d
i
gi
tal
m
em
ory
as
a
ma
tr
i
x
of
bi
n
ary
nu
m
be
r
s
, wh
ere th
i
s
di
g
i
ta
l
i
ma
ge
c
a
n b
e
proc
e
s
s
ed
[6
].
T
he
f
un
c
ti
on
pri
nc
i
p
l
e
of
t
he
s
p
ati
al
d
om
a
i
n
fi
l
t
erin
g
i
s
to
r
ep
l
ac
e
the
c
orr
up
t
e
d
p
i
x
el
(
no
i
s
e)
by
an
oth
er
v
al
ue
fr
om
i
ts
n
ei
gh
b
ors
i
n
th
e
no
i
s
y
i
ma
ge
,
s
uc
h
as
the
m
e
di
a
n,
th
e
me
an
an
d
s
o
o
n.
T
he
ma
n
ne
r
of
s
el
ec
ti
ng
th
i
s
v
al
u
e,
i
s
on
e
of
the
mo
s
t
i
mp
ort
an
t
p
oi
n
t
s
whi
c
h
aff
ec
t
the
eff
ec
ti
v
en
es
s
an
d
eff
i
c
i
en
c
y
of
the
i
m
ag
e
d
en
o
i
s
i
n
g
me
th
od
.
T
h
i
s
s
tu
dy
ai
ms
to
tr
ea
t
th
i
s
by
propos
i
ng
a
n
ew
tec
h
ni
qu
e
for
d
ete
c
t
i
ng
an
d
r
ep
l
a
c
i
ng
t
he
c
orr
up
ted
pi
x
e
l
,
by
i
nt
eg
r
at
i
ng
the
ma
t
he
ma
t
i
c
al
c
on
c
ep
t
''
A
r
i
thm
eti
c
P
r
o
gres
s
i
on
''
i
n
i
ma
ge
de
no
i
s
i
ng
.
T
h
i
s
i
s
i
n
order
to
ge
t
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
MNIK
A
IS
S
N: 1
69
3
-
6
93
0
◼
P
r
og
r
es
s
i
o
n a
pp
r
oa
c
h
fo
r
i
ma
ge
de
no
i
s
i
ng
(
B
i
l
a
l
C
ha
r
mo
ut
i
)
2949
the
be
s
t
s
i
m
i
l
arit
y
wi
t
h
the
orig
i
na
l
i
m
ag
e.
T
he
pe
r
f
ormanc
e
of
t
he
de
no
i
s
i
ng
t
ec
hn
i
qu
es
v
arie
s
fr
om
o
ne
f
i
l
t
er to
an
ot
he
r
f
i
l
t
er, and
i
t c
a
n b
e e
v
a
l
ua
ted
ac
c
ordi
ng
to
v
ari
ou
s
c
r
i
ter
i
a
, whi
c
h
are:
−
T
he
no
i
s
ed
i
ma
g
e
(
i
np
ut):
the
ty
pe
of
n
oi
s
e
(
ad
di
t
i
v
e,
mu
l
t
i
p
l
i
c
at
i
v
e,
i
mp
u
l
s
e
no
i
s
e,
mi
x
ed
no
i
s
e),
l
ev
e
l
of
no
i
s
e (h
i
gh
,
l
ow)
, s
tr
uc
ture
of
i
ma
ge
(
t
e
x
ture, s
mo
o
th,
ed
g
e), p
i
x
el
i
nte
ns
i
ty
.
−
F
i
l
ter
(
de
no
i
s
i
ng
to
ol
)
:
c
om
pu
tat
i
on
al
c
os
t
(
ac
c
ep
tab
l
e,
hi
gh
)
,
fi
l
ter
i
m
pl
e
me
nt
ati
on
(
s
i
mp
l
e,
c
om
pl
ex
)
.
−
Res
tored
i
ma
g
e
(
ou
tpu
t)
:
me
as
ure
me
n
t
of
n
oi
s
e
(
S
NR,
M
S
E
,
P
S
NR
,
ec
t.),
v
i
s
ua
l
qu
a
l
i
ty
(
bl
ur, ar
t
i
fac
ts
,
i
nfo
r
ma
t
i
on
l
os
s
an
d e
tc
)
.
i
n
brie
f,
th
i
s
s
ec
ti
o
n
prov
i
de
s
the
n
ec
es
s
ary
i
nf
ormat
i
o
n
r
es
tr
i
c
ted
i
n
th
e
pr
i
ma
r
y
p
ha
s
e
o
f
i
ma
g
e
proc
es
s
i
ng
(
i
ma
ge
pr
e
-
proc
es
s
i
ng
)
, w
hi
c
h
i
s
c
al
l
ed
Ima
ge
d
en
o
i
s
i
n
g.
2.
L
it
e
r
atu
r
e
Re
vie
w
2.1.
Imp
u
l
se
No
i
se
In
c
as
e
of
th
e
gr
ay
s
c
al
e
i
ma
ge
,
t
he
i
mp
u
l
s
e
no
i
s
e
ma
y
b
e
r
e
pres
en
t
ed
by
r
a
nd
om
v
al
ue
s
(
R
V
)
of
pi
x
e
l
s
(
v
al
u
e
b
etwe
en
0
to
25
5)
i
n
t
he
c
orr
up
te
d
i
ma
ge
,
or
by
f
i
x
ed
v
a
l
ue
s
(
F
V
)
al
s
o
c
al
l
e
d
"
s
al
t
&
pe
p
pe
r
"
no
i
s
e
produc
e
d
by
r
an
d
o
m
pa
r
t
i
al
di
s
tr
i
bu
t
i
on
of
wh
i
te
pi
x
e
l
s
(
v
al
u
e
25
5)
an
d
bl
ac
k
pi
x
e
l
s
(
v
al
u
e
0)
i
nto
th
e
i
m
ag
e
[
7],
as
s
ho
wn
i
n
F
i
gu
r
e
1
,
un
l
i
k
e
g
au
s
s
i
an
n
oi
s
e
wi
th
t
he
en
t
i
r
e
d
i
s
tr
i
bu
ti
o
n
(
al
l
i
m
ag
e
pi
x
e
l
s
)
[8]
.
T
he
i
m
ag
e
(
,
)
c
orr
up
ted
by
RV
an
d
F
V
no
i
s
e
i
s
de
s
c
r
i
be
d b
y
(
,
)
(
2) and
(
,
)
(
1), r
es
pe
c
ti
v
e
l
y
[9
].
(
,
)
=
{
(
,
)
∈
[
0
,
255
]
,
wi
th
pr
ob
a
b
il
ity
p
(
,
)
,
wi
th
pr
ob
a
b
i
l
ity
1
−
p
(
1)
(
,
)
=
{
0
255
,
wi
th
pr
ob
a
b
il
ity
p
(
,
)
,
wi
th
pr
ob
a
b
il
ity
1
−
p
(
2)
F
i
gu
r
e
1.
I
ma
g
e w
i
th
s
a
l
t
a
nd
p
ep
p
er no
i
s
e
2.2
.
E
f
f
ici
ent
Me
t
h
o
d
s
f
o
r
Remo
v
ing
Imp
u
lse
No
i
se
f
r
o
m
Im
age
2.2
.1
.
Me
d
ian
-
r
el
ated
Filt
e
r
s
Me
d
i
an
fi
l
ter
[
2]
be
l
on
gs
to
the
f
am
i
l
y
o
f
no
n
-
l
i
ne
ar
fi
l
te
r
s
,
i
t
i
s
a
s
i
m
pl
e
f
i
l
t
er
[8]
,
w
hi
c
h
i
s
ba
s
ed
on
the
r
an
k
order
i
n
g
of
p
i
x
el
v
a
l
u
es
fr
om
th
e
proc
es
s
ed
area.
T
he
c
orr
up
te
d
p
i
x
el
i
s
r
ep
l
ac
e
d
by
on
e
(
me
di
an
)
tak
en
fr
om
a
l
l
p
i
x
el
s
i
n
t
he
an
a
l
y
s
ed
wi
n
do
w
c
en
t
ered
on
th
at
p
i
x
el
,
i
ns
tea
d
of
the
m
ea
n
v
a
l
ue
whi
c
h
i
s
de
r
i
v
e
d
fr
om
a
c
al
c
ul
ate
d
v
a
l
u
e,
an
d
t
hi
s
i
s
a
n
a
dv
an
t
ag
e
for
the
me
di
an
fi
l
ter
[1
0].
T
he
me
d
i
an
f
i
l
t
er
i
s
r
ob
us
t
to
d
i
f
ferent
ty
pe
s
of
n
oi
s
e,
w
he
r
e
i
t
y
i
e
l
ds
great
r
es
ul
ts
wi
th
i
mp
ul
s
e
no
i
s
e
[
11
]
a
nd
ou
t
pe
r
form
s
t
he
l
i
n
ea
r
fi
l
ter
i
n
pres
erv
i
n
g
i
ma
ge
e
dg
es
[
12
].
Nev
erthe
l
es
s
,
i
t
s
ho
ws
a
l
i
mi
ta
ti
o
n
i
n
c
as
e
of
h
i
g
h
de
ns
i
ty
of
no
i
s
e
by
r
em
ov
i
n
g
s
om
e
i
m
po
r
ta
nt
i
nfo
r
ma
t
i
on
fr
om
t
he
i
m
ag
e
[8]
.
In
order
t
o
ex
c
ee
d
th
i
s
l
i
mi
t
ati
o
n,
s
ev
era
l
ex
te
ns
i
on
te
c
hn
i
q
ue
s
(
de
r
i
v
e
d
f
r
om
MF
)
ha
v
e
be
en
prop
os
ed
s
uc
h
as
:
W
e
i
gh
t
ed
me
d
i
a
n
f
i
l
t
er
(
W
MF)
[13
]
wh
i
c
h
at
tac
he
s
hi
g
he
r
we
i
g
hts
(
c
oe
ffi
c
i
e
nts
)
t
o
t
he
p
i
x
el
s
tha
t
are
c
l
os
er
to
th
e
c
e
ntr
al
pi
x
e
l
,
k
no
w
i
ng
t
ha
t
i
n
t
h
e
c
as
e
of
MF
Evaluation Warning : The document was created with Spire.PDF for Python.
◼
IS
S
N: 16
93
-
6
93
0
T
E
L
KO
MNIK
A
V
ol
.
17
,
No
.
6,
D
ec
em
b
er 20
19
:
29
4
8
-
2958
2950
the
w
ei
g
hts
are
eq
u
al
.
Me
a
nwhi
l
e,
i
n
the
c
as
e
th
at
th
i
s
a
dd
i
ti
on
a
l
w
ei
g
ht
g
oe
s
o
nl
y
to
the
c
e
ntra
l
pi
x
el
o
f
tr
e
ate
d
w
i
n
do
w,
th
e
f
i
l
t
er
w
i
l
l
be
c
al
l
ed
t
he
c
e
nte
r
w
ei
gh
te
d
me
di
a
n
fi
l
ter
s
(
CW
MF)
[14
],
di
r
ec
ti
on
a
l
wei
gh
t
ed
me
di
a
n
f
i
l
t
er
(
DW
MF)
[
15
],
s
wi
tc
hi
n
g
Me
d
i
an
F
i
l
ter
[2]
,
r
ec
u
r
s
i
v
e
we
i
gh
t
ed
me
d
i
an
fi
l
ter (
R
W
MF)
[4
] a
n
d o
th
ers
[1
6].
2.2
.2
.
Ad
apt
iv
e
F
ilte
r
ing
O
v
er
ma
ny
y
ea
r
s
the
ad
ap
t
i
v
e
fi
l
teri
ng
tec
h
ni
q
ue
s
tak
e
a
s
i
gn
i
f
i
c
an
t
p
art
of
att
en
t
i
o
n
an
d
us
a
ge
[
17
]
t
ha
t
c
arr
y
gre
at
i
mp
orta
nc
e
i
n
m
an
y
a
pp
l
i
c
ati
on
s
of
s
i
gn
al
proc
es
s
i
ng
.
A
s
a
s
tat
i
s
ti
c
a
l
ap
pro
ac
h,
t
hi
s
ty
pe
of
a
l
g
orit
h
ms
c
ha
ng
es
i
ts
c
h
arac
teri
s
t
i
c
s
au
t
om
at
i
c
al
l
y
a
nd
r
ec
urs
i
v
el
y
to
b
e
a
da
p
ted
to
th
e
s
tat
i
s
ti
c
al
pa
r
a
me
t
ers
of
the
tr
ea
t
i
n
g
s
i
g
na
l
wi
t
h
no
r
eq
u
i
r
ed
prio
r
i
nfo
r
m
ati
on
,
i
n
order
t
o o
pt
i
m
i
s
e t
h
e
i
nn
er c
oe
ff
i
c
i
en
ts
of
t
he
f
i
l
ter [1
8].
T
he
o
pe
r
at
i
on
m
ec
ha
n
i
s
m
of
th
e
a
da
p
ti
v
e
fi
l
ter
i
s
i
l
l
us
tr
ate
d
i
n
F
i
g
ure
2,
w
hi
c
h
ai
ms
to
mi
n
i
m
i
z
e
t
he
err
or
pro
du
c
e
d
by
th
e
s
ub
s
tr
ac
t
i
on
of
̂
(
o
utp
ut
of
the
fl
ex
i
b
l
e
al
g
orit
h
m)
fr
om
the
orig
i
na
l
i
ma
g
e
(
fr
ee
of
n
oi
s
e)
as
r
efe
r
e
nc
e.
T
he
n,
thi
s
err
or
i
s
us
ed
wi
t
h
th
e
no
i
s
y
i
m
ag
e
(
i
np
ut
of
al
go
r
i
thm
)
to
up
d
ate
th
e
fi
l
ter
pa
r
a
me
t
ers
,
wi
th
att
en
t
i
o
n
to
th
e
mi
ni
mi
s
ed
c
r
i
teri
a
i
n
order
to
ac
hi
ev
e
an
o
pti
m
al
a
l
go
r
i
thm
at
the
e
nd
of
t
he
proc
es
s
[17
].
T
h
ere
are
a
l
ot
of
b
as
i
c
fi
l
t
erin
g
tec
hn
i
q
ue
s
ad
op
te
d
for
thi
s
c
on
c
e
pt
to
b
e
an
ad
ap
ti
v
e
tec
h
ni
qu
e
s
uc
h
a
s
me
di
an
fi
l
ter,
bi
l
ate
r
a
l
f
i
l
t
er,
W
i
e
ne
r
f
i
l
t
e
r
,
fuz
z
y
fi
l
ter,
mo
r
p
ho
l
og
i
c
al
f
i
l
t
er,
an
d
s
o
on
.
S
o
m
e
of
t
he
m
are
pres
en
te
d i
n
[7
,
9].
F
i
gu
r
e
2
.
A
da
pti
v
e
fi
l
ter
a.
A
da
pti
v
e
Me
di
a
n Fi
l
ter (
A
M
F
)
T
he
ad
a
pti
v
e
m
ed
i
an
f
i
l
t
er
was
b
ei
n
g
us
ed
mo
r
e
f
am
i
l
i
arly
tha
n
t
he
c
l
as
s
i
c
me
d
i
a
n
f
i
l
t
er
s
i
nc
e
i
t
i
s
m
ore
eff
ec
ti
v
e
c
o
mp
are
d
to
th
i
s
l
as
t.
T
h
e
A
M
F
wor
k
s
on
the
de
t
ec
ti
o
n
of
the
c
orr
up
te
d
pi
x
el
c
o
mp
are
d
to
i
ts
ne
i
gh
bo
r
ho
od
i
n
the
tr
e
ate
d
w
i
n
d
ow
to
be
l
ab
el
e
d
as
a
no
i
s
y
pi
x
el
;
t
he
s
i
z
e
of
th
i
s
wi
nd
ow
ma
y
be
v
ar
i
ed
ac
c
ordi
n
g
to
the
c
om
p
aris
on
c
r
i
t
eria
.
T
he
n,
th
i
s
l
ab
e
l
ed
pi
x
e
l
i
s
r
ep
l
ac
e
d
by
t
he
me
di
a
n
p
i
x
el
o
f
th
e
t
es
ted
ne
i
gh
bo
urh
oo
d
[
19
].
A
MF
g
i
v
es
a
mu
c
h
be
t
ter
r
es
u
l
t
i
n
r
e
mo
v
i
n
g
th
e
s
a
l
t
a
nd
pe
pp
er
no
i
s
e,
c
o
mp
are
d
wi
th
oth
er
me
di
a
n
f
i
l
t
er
ty
pe
s
,
wh
eth
er
i
n
the
v
i
s
u
al
qu
a
l
i
ty
or No
i
s
e r
ati
o
c
r
i
teri
a [
1
6].
b.
A
da
pti
v
e
W
e
i
g
hte
d M
e
di
an
F
i
l
ter (
A
W
MF)
T
he
A
W
MF
i
s
c
on
s
i
de
r
e
d
as
an
ad
v
an
c
e
d
tec
hn
i
qu
e
c
om
pa
r
ed
wi
th
the
c
l
as
s
i
c
wei
g
hte
d
m
ed
i
an
fi
l
teri
ng
.
A
W
MF
i
s
t
he
proc
es
s
by
w
hi
c
h
t
he
w
ei
gh
te
d
m
ed
i
an
f
i
l
t
er
ha
s
b
ee
n
ap
p
l
i
e
d
ad
a
pti
v
e
l
y
t
o t
he
no
i
s
y
i
m
ag
e
, b
y
way
of
ad
j
us
t
i
ng
t
he
fi
l
ter
pa
r
am
et
ers
an
d
th
os
e
w
ei
g
ht
c
oe
ffi
c
i
en
ts
by
t
he
l
oc
a
l
s
tat
i
s
t
i
c
s
of
the
tr
ea
t
ed
area.
T
h
e
A
W
MF
gi
v
es
t
he
po
s
s
i
b
i
l
i
ty
of
r
em
ov
i
n
g n
oi
s
e
by
pres
erv
i
ng
e
dg
es
a
nd
i
m
ag
e
fea
t
ur
es
[1
2].
2.2
.3
.
F
u
z
z
y
F
ilteri
n
g
T
he
c
on
c
e
pt
o
f
th
e
f
uz
z
y
s
et
th
eo
r
y
,
h
ad
be
e
n
us
e
d
t
he
f
i
r
s
t
ti
me
f
or
i
ma
g
e
pr
oc
es
s
i
ng
i
n
[20
].
T
he
n
,
s
ev
eral
fi
l
ter
i
ng
tec
h
ni
qu
es
ha
s
e
i
th
er
be
en
g
en
erated
or
de
v
el
o
pe
d
gra
du
a
l
l
y
fr
om
t
hi
s
l
og
i
c
i
n
t
he
s
e
l
as
t
de
c
ad
es
[2
1].
T
he
s
e
f
uz
z
y
tec
hn
i
qu
es
tr
ea
t
th
e
v
ari
et
y
i
n
t
he
no
i
s
y
da
ta
whe
n i
t
c
om
es
fro
m a
mb
i
gu
i
ty
i
ns
t
ea
d
of
r
an
do
m
ne
s
s
[2
1].
F
uz
z
y
i
ma
g
e
pr
oc
es
s
i
ng
ho
l
ds
t
hree
ma
i
n
ph
a
s
es
[22
],
th
e
f
i
r
s
t
on
e
c
al
l
ed
''
fuz
z
i
fi
c
at
i
on
''
,
t
ha
t
tr
a
ns
for
ms
the
i
np
u
t
da
t
a
to
t
he
m
em
be
r
s
h
i
p
p
l
a
ne
to
de
a
l
w
i
th
me
mb
ers
h
i
p
v
al
ue
s
,
wh
i
l
e
th
e
s
ec
o
nd
an
d
the
m
os
t
c
r
uc
i
a
l
on
e
i
s
th
e
m
od
i
fi
c
a
ti
o
n
of
t
he
s
e
m
em
b
e
r
s
hi
p
v
al
ue
s
us
i
ng
s
ui
tab
l
e
fuz
z
y
tec
hn
i
qu
es
,
a
nd
the
l
as
t
on
e
i
s
th
e
''
d
efu
z
z
i
f
i
c
ati
on
''
,
pe
r
for
me
d
to
ge
t
the
o
utp
u
t d
ata
i
n
the
or
i
gi
na
l
pl
a
ne
.
F
uz
z
y
l
og
i
c
i
s
c
ha
r
ac
teri
z
ed
by
the
i
nh
er
e
nt
u
nc
erta
i
nty
,
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
MNIK
A
IS
S
N: 1
69
3
-
6
93
0
◼
P
r
og
r
es
s
i
o
n a
pp
r
oa
c
h
fo
r
i
ma
ge
de
no
i
s
i
ng
(
B
i
l
a
l
C
ha
r
mo
ut
i
)
2951
tha
t
ma
k
es
i
t
mo
r
e
fi
t
t
o
th
e
i
m
ag
e
c
orr
up
te
d
by
th
e
i
mp
u
l
s
e
no
i
s
e
[
8].
T
h
erefor
e,
It
d
ea
l
s
mu
c
h
mo
r
e
w
i
t
h
th
e
me
d
i
a
n
f
i
l
t
ers
[23
],
as
l
on
g
as
i
t
i
s
a
g
o
od
t
oo
l
for
no
i
s
e
de
tec
t
i
o
n.
T
he
f
uz
z
y
fi
l
t
er
i
s
ba
s
ed
on
t
wo
m
ai
n
fe
at
ures
;
the
f
i
r
s
t
on
e
i
s
t
o
es
t
i
ma
t
e
a
"
fuz
z
y
de
r
i
v
ati
v
e"
t
o
be
s
of
t
wi
th
the
f
i
ne
d
et
ai
l
s
of
i
ma
ge
s
u
c
h
as
the
e
dg
e
,
ac
c
ordi
ng
t
o
the
l
oc
a
l
v
aria
ti
o
n
of
i
m
ag
e;
the
s
ec
on
d,
i
s
th
at
th
e
me
mb
ers
h
i
p f
un
c
ti
on
s
are
ad
ap
t
i
v
e t
o
the
n
oi
s
e
l
ev
el
to
y
i
el
d "
fuz
z
y
s
mo
oth
i
ng
"
[
24
].
T
he
r
e
are
s
ev
era
l
ty
p
es
of
fuz
z
y
fi
l
ters
th
at
h
av
e
be
en
i
ntro
du
c
ed
b
y
r
es
ea
r
c
he
r
s
s
uc
h
as
,
the
w
el
l
-
k
no
wn,
F
IRE
fi
l
ters
[25
].
T
he
a
da
pt
i
v
e
wei
g
hte
d
fuz
z
y
me
a
n
fi
l
ter
(
A
W
F
M)
[
26
],
ad
ap
ti
v
e
fuz
z
y
s
wi
tc
h
i
ng
fi
l
ter
(
A
F
S
F
)
[
27
],
th
e
i
tera
t
i
v
e
f
uz
z
y
c
on
tr
ol
ba
s
e
d
f
i
l
ter
fr
om
[2
2],
the
fuz
z
y
bi
l
ate
r
a
l
fi
l
ter
i
n
g
(
F
B
F
)
[5]
,
the
f
uz
z
y
r
an
do
m
i
mp
u
l
s
e
no
i
s
e
r
ed
uc
ti
on
me
th
od
(
F
RINRM)
[2
8],
t
he
fuz
z
y
s
i
mi
l
ar
i
ty
fi
l
ter
(
F
S
B
)
[
6]
,
A
F
uz
z
y
Noi
s
e
Re
du
c
ti
on
Me
tho
d
for
C
ol
or
Ima
g
es
[2
8]
an
d o
t
he
r
s
i
n
[
7,
9,
16].
2.3.
E
va
luat
ion
P
ar
a
met
er
s
of
Imag
e Qu
a
lity
T
he
i
ma
g
e
q
ua
l
i
ty
i
s
on
e
of
t
he
ev
al
ua
t
i
on
c
r
i
ter
i
a
o
f
th
e
de
no
i
s
i
n
g
tec
h
ni
q
ue
s
pe
r
forma
nc
e
(
c
i
t
ed
i
n
t
he
l
as
t
s
ec
ti
o
n),
he
nc
e
th
e
qu
e
s
ti
on
po
s
e
d
i
s
h
ow
t
o
as
s
es
s
thi
s
q
ua
l
i
ty
.
T
he
w
ay
by
w
hi
c
h
th
e
i
ma
ge
qu
a
l
i
ty
i
s
ev
al
u
ate
d
(
ev
al
u
ati
on
c
r
i
teri
a)
ma
y
s
pl
i
t
i
nto
tw
o
w
ay
s
the
fi
r
s
t
o
ne
i
s
th
e
v
i
s
ua
l
e
v
al
ua
t
i
o
n
de
t
ermi
ne
d
by
th
e
ob
s
erv
er,
wher
e
the
hu
m
an
j
ud
g
me
nt
i
s
i
nte
r
es
t
ed
i
n
the
i
ma
ge
c
o
mp
on
en
ts
ap
pe
aranc
e,
wh
eth
er
or
no
t
i
t
c
on
t
ai
ns
a
n
y
de
gradat
i
o
n
fac
tors
s
uc
h
as
arti
f
ac
ts
,
di
s
c
on
t
i
nu
i
ti
es
an
d
b
l
ur
[2
9].
T
he
s
ec
o
nd
on
e
i
s
th
e
qu
a
n
t
i
tat
i
v
e
ev
al
u
ati
on
by
us
i
ng
t
he
m
e
as
ureme
nt
pa
r
am
et
ers
,
am
on
g
t
he
m
an
d
t
he
mo
s
t
us
ed
are:
S
i
gn
a
l
to
Noi
s
e
Rat
i
o
(
S
N
R)
,
wh
i
c
h
me
as
ures
th
e
am
ou
n
t
of
no
i
s
e
i
n
the
n
oi
s
y
i
ma
g
e
(
,
)
us
i
ng
the
s
tan
da
r
d
d
ev
i
at
i
on
of
the
no
i
s
e
(
)
an
d
i
m
ag
e
(
)
(
(
)
=
60
i
nd
i
c
at
i
n
g
go
o
d
i
ma
g
e
qu
a
l
i
ty
)
[30
]
,
as
gi
v
en
by
(
3);
Me
an
S
qu
ar
ed
E
r
r
or
(
MS
E
)
,
th
at
me
as
ures
th
e
di
s
s
i
m
i
l
arit
y
be
twe
en
th
e
r
es
tored
i
ma
g
e
̂
(
,
)
an
d
th
e
orig
i
n
al
o
ne
(
,
)
as
s
ho
wn
i
n
(
4),
th
us
when
ev
er
the
M
S
E
i
s
l
ower,
t
he
i
ma
ge
de
no
i
s
i
n
g
w
i
l
l
ac
hi
ev
e
mo
r
e
s
uc
c
es
s
[31
];
P
e
ak
S
i
g
na
l
to
No
i
s
e
Rati
o
(
P
S
NR)
[32
]
,
a
we
l
l
k
no
wn
pa
r
am
e
ter
wh
i
c
h
h
as
an
i
nv
ers
e
r
e
l
at
i
on
s
h
i
p
wi
th
MS
E
,
as
de
no
ted
i
n (
5).
It
i
s
no
t
ne
c
es
s
ary
to
be
an
e
nta
i
l
m
en
t
r
el
a
ti
o
n
b
et
ween
t
he
v
i
s
ua
l
an
d
qu
an
ti
tat
i
v
e
as
s
es
s
me
nts
,
be
c
au
s
e
s
o
me
ti
me
s
an
i
m
ag
e
,
ev
en
wi
th
h
i
g
h
P
S
NR
or
l
ow
M
S
E
,
do
es
n
ot
ap
pe
ar to
be
c
l
ea
n
.
=
(
)
(
)
(
3)
M
SE
=
1
∑
(
[
,
]
−
̂
[
,
]
)
2
,
(
4)
PS
N
R
=
10
l
og
10
(
255
2
)
(
5)
:
T
he
n
um
b
er of
pi
x
e
l
s
i
n
i
m
ag
e.
2.4.
A
r
it
h
met
ic
P
r
o
g
r
es
si
o
n
(
AP
)
A
s
eq
u
en
c
e
i
s
a
s
et
o
f
nu
mb
ers
wr
i
tt
en
i
n
a
pa
r
t
i
c
ul
ar
order
,
s
uc
h
as
the
s
e
n
u
mb
ers
whi
c
h
are
g
i
v
en
by
th
e
for
m
2
4
6
8
10
1
2
.
.
.
A
n
arit
hm
eti
c
pr
og
r
es
s
i
on
(
A
P
)
i
s
a
pa
r
ti
c
u
l
ar
c
as
e
of
s
eq
u
en
c
e
where
a
ny
term
i
n
thi
s
s
eq
ue
nc
e
i
s
ob
t
ai
ne
d
by
ad
di
n
g
a
c
on
s
tan
t
v
al
ue
to
t
he
prev
i
ou
s
t
erm,
w
he
r
e
th
e
c
o
ns
tan
t
i
s
c
al
l
ed
t
he
c
om
m
on
d
i
ffe
r
en
c
e
.
F
or
ex
am
p
l
e,
th
e
t
erms
o
f
an
ari
thm
eti
c
progr
es
s
i
on
s
tarte
d
by
th
e
t
erm
a,
are
as
fol
l
ows
:
,
+
,
+
2
,
+
3
,
.
.
.
2.4.
1
.
F
ind
ing
a
Mis
s
ing
(
U
n
kno
w
n
)
T
er
m
o
f
an
A
r
it
h
met
ic
P
r
o
g
r
e
ss
ion
T
he
ge
n
eral
ter
m
of
an
ari
thm
et
i
c
progr
es
s
i
o
n
(
A
P
)
,
[
33
]
c
an
be
pres
e
nte
d
by
t
he
(
6).
T
hu
s
,
to
f
i
nd
th
e
mi
s
s
i
n
g
(
de
s
i
r
e
d)
ter
m
o
f
a
n
arit
hm
et
i
c
pro
gres
s
i
on
(
A
P
)
we
s
i
mp
l
y
us
e
the
f
ol
l
ow
i
ng
formu
l
a:
=
+
(
−
1
)
(
6)
where
,
n
i
s
th
e
nu
mb
er
of
d
es
i
r
ed
t
erms
i
n
t
hi
s
s
eq
ue
n
c
e,
the
c
o
mm
o
n
d
i
ffe
r
e
nc
e
of
A
P
an
d
a
i
s
the
f
i
r
s
t te
r
m.
Evaluation Warning : The document was created with Spire.PDF for Python.
◼
IS
S
N: 16
93
-
6
93
0
T
E
L
KO
MNIK
A
V
ol
.
17
,
No
.
6,
D
ec
em
b
er 20
19
:
29
4
8
-
2958
2952
2.4
.2
.
Ari
t
h
m
etic
P
r
o
g
r
es
sion
in
Imag
e P
r
o
ce
ss
ing
Ma
ny
as
pe
c
ts
a
nd
s
i
tua
ti
o
n
s
of
l
i
v
es
c
a
n
b
e
mo
d
el
ed
b
y
the
arit
hm
et
i
c
pr
og
r
es
s
i
o
n,
a
nd
a
l
ot
of
ph
en
o
me
n
on
c
an
b
e
i
nte
r
pre
ted
to
a
s
eq
ue
nc
e
of
ter
ms
.
W
e
j
us
t
n
ee
d
to
k
no
w
ho
w
c
an
us
e
i
t
i
n
r
e
al
l
i
f
e
a
nd
h
ow
c
an
ex
pl
oi
t
i
t
t
o
h
av
e
v
ari
ou
s
v
i
ew
po
i
nt
a
bo
u
t
ho
w
t
hi
n
gs
oc
c
ur
i
n
the
l
i
fet
i
me
.
C
on
c
ern
i
ng
t
he
i
m
ag
e
proc
es
s
i
n
g,
the
arit
hm
e
ti
c
progr
es
s
i
o
n
ha
s
a
p
orti
on
i
n
tr
ea
ti
ng
the
i
ma
g
e.
F
or
e
x
am
pl
e,
order
i
ng
the
p
i
x
el
s
of
t
he
i
m
ag
e
b
as
ed
on
the
ari
thm
eti
c
s
eq
ue
nc
e
as
do
ne
i
n
[3
4]
.
F
urther
mo
r
e,
i
n
[
35
],
th
e
arit
h
me
t
i
c
pro
gres
s
i
on
i
s
u
s
ed
for
i
ma
g
e
wate
r
m
ark
i
ng
.
3
. Re
se
a
r
ch
Me
t
h
o
d
T
he
pr
op
os
e
d
me
th
od
o
l
o
g
y
of
t
hi
s
de
n
oi
s
i
ng
me
tho
d
i
n
tr
od
uc
es
i
n
d
eta
i
l
s
t
he
s
ev
eral
s
tep
s
tha
t
are
a
do
pt
ed
t
o
r
es
tore
pe
r
f
ec
tl
y
th
e
ori
gi
na
l
i
ma
g
e
fr
om
the
no
i
s
y
i
ma
ge
i
n
ord
er
to
s
ol
v
e
the
pro
bl
em
of
n
oi
s
e
whi
c
h
r
ed
uc
es
a
nd
th
r
ea
ts
the
i
ma
g
e
qu
a
l
i
ty
,
us
i
ng
a
s
i
mp
l
e
ma
th
em
a
ti
c
a
l
c
on
c
ep
t
c
al
l
e
d
''
A
r
i
t
hm
e
ti
c
P
r
og
r
es
s
i
o
n
(
A
P
)
''
.
F
i
gu
r
e
3
d
es
c
r
i
be
s
th
e
fl
ow
c
ha
r
t
of
tho
s
e
s
tep
s
wh
i
c
h
are
i
n
a
l
l
,
t
hree
s
t
eps
-
th
e
fi
r
s
t
s
tep
i
s
to
c
r
ea
te
th
e
no
i
s
y
i
ma
ge
by
ad
di
n
g
no
i
s
e
to
t
he
orig
i
n
al
or
no
i
s
e
fr
e
e
i
m
ag
e
(
i
np
u
t
i
ma
ge
)
,
t
he
n
t
hi
s
no
i
s
y
i
ma
g
e
i
s
pa
s
s
ed
to
the
s
ec
o
nd
s
te
p
w
hi
c
h
i
s
the
f
i
l
teri
ng
ph
as
e
to
r
em
ov
e
no
i
s
e
fr
om
t
he
i
m
ag
e
an
d
i
t
i
s
al
s
o
di
v
i
d
ed
i
nto
t
wo
s
ub
-
p
ha
s
es
.
T
he
fi
r
s
t
on
e
i
s
the
no
i
s
e
de
t
ec
ti
o
n,
wh
ere
the
c
orr
up
ted
pi
x
e
l
s
s
ho
ul
d
be
d
i
s
ti
n
gu
i
s
he
d
fr
om
the
un
c
orr
up
t
ed
p
i
x
el
s
,
an
d
th
en
th
es
e
c
orr
up
ted
pi
x
el
s
wi
l
l
be
r
ep
l
ac
e
d
by
t
he
c
orr
ec
t
v
al
u
es
i
n
the
s
ec
on
d
s
ub
-
ph
as
e
(
pi
x
e
l
r
es
to
r
ati
on
)
,
wh
ere
the
un
c
orr
up
te
d
pi
x
e
l
s
r
es
t
i
nta
ng
i
bl
e.
T
h
e
ex
tr
ac
ti
o
n
o
f
the
a
l
tern
ate
v
a
l
ue
of
ea
c
h
pi
x
e
l
(
c
orr
ec
t
v
al
ue
)
w
i
l
l
b
e
d
on
e
to
t
h
e
no
i
s
y
i
ma
ge
.
T
he
s
e
f
ormer
s
tep
s
(
fi
l
ter
i
ng
)
y
i
el
d
the
es
t
i
ma
ti
o
n
(
r
es
torati
o
n)
of
th
e
or
i
gi
na
l
i
ma
ge
(
o
utp
u
t
i
ma
g
e).
F
i
n
al
l
y
,
i
n
t
he
l
as
t
s
tep
,
M
S
E
an
d
P
S
N
R
are
c
al
c
ul
a
ted
us
i
n
g
th
e
i
np
ut
a
nd
o
utp
u
t
i
ma
ge
s
,
i
n
ord
er
to
ev
a
l
ua
t
e
t
he
pe
r
form
an
c
e
of
propos
e
d
me
th
od
by
c
o
mp
ar
i
ng
th
es
e l
as
t v
a
l
u
es
wi
t
h o
t
he
r
v
al
ue
s
r
el
a
ted
to
oth
er
de
n
oi
s
i
ng
m
eth
o
ds
.
F
i
gu
r
e
3
.
M
eth
od
o
l
o
gy
fl
o
w
c
ha
r
t
3.1.
Ad
d
No
ise
T
hi
s
p
ha
s
e
i
s
c
o
nc
erne
d
wi
t
h
c
r
e
ati
ng
the
n
oi
s
y
i
ma
g
e
w
hi
c
h
i
s
the
n
u
s
ed
i
n
the
de
n
oi
s
i
ng
tec
h
ni
q
ue
.
T
hi
s
n
oi
s
y
i
ma
ge
i
s
ob
ta
i
n
e
d
by
ad
d
i
n
g
n
oi
s
e
to
th
e
orig
i
na
l
i
ma
ge
(
no
i
s
e
fr
e
e
i
ma
g
e)
ac
c
ordi
ng
to
two
pa
r
am
ete
r
s
l
i
nk
e
d
wi
t
h
th
e
no
i
s
e
w
hi
c
h
ar
e
the
ty
p
e
an
d
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
MNIK
A
IS
S
N: 1
69
3
-
6
93
0
◼
P
r
og
r
es
s
i
o
n a
pp
r
oa
c
h
fo
r
i
ma
ge
de
no
i
s
i
ng
(
B
i
l
a
l
C
ha
r
mo
ut
i
)
2953
the
i
nte
ns
i
ty
of
no
i
s
e.
In
t
h
e
c
as
e
of
i
m
pu
l
s
e
no
i
s
e,
a
nu
mb
er
of
p
i
x
el
s
i
n
the
or
i
g
i
na
l
i
ma
g
e
are
c
ho
s
en
r
an
do
ml
y
t
o b
e
c
ha
ng
ed
by
r
an
do
m v
al
ue
s
of
pi
x
el
s
(
v
a
l
ue
s
r
a
ng
i
ng
from
0 t
o
25
5).
That
ac
ti
on
y
i
el
ds
a
no
i
s
y
i
ma
g
e
,
wh
ere
the
i
nte
ns
i
ty
of
no
i
s
e
i
s
c
o
ntro
l
l
a
bl
e
an
d
i
t
i
s
r
ep
r
es
en
t
ed
by
the
pe
r
c
en
tag
e
of
the
s
e
r
a
nd
om
v
a
l
u
es
(
no
i
s
e)
i
n
th
e
no
i
s
y
i
m
ag
e.
A
s
me
nt
i
o
ne
d
prev
i
ou
s
l
y
,
the
i
mp
ul
s
e
no
i
s
e
i
s
t
he
ty
p
e
of
no
i
s
e
c
ho
s
en
to
be
tr
e
ate
d
i
n
thi
s
s
t
ud
y
,
d
ue
t
o
i
ts
s
tr
uc
ture
tha
t
ma
k
es
t
he
ad
op
t
i
on
of
th
e
propos
e
d
ma
th
em
ati
c
a
l
c
o
nc
ep
t
''
arit
h
me
t
i
c
pr
og
r
es
s
i
on
''
mo
r
e
ab
l
e
i
n s
u
c
c
es
s
ful
l
y
s
ol
v
i
n
g t
hi
s
k
i
nd
o
f p
r
ob
l
e
ms
.
3.2.
F
ilte
r
ing
T
hi
s
s
tep
i
s
t
he
o
pp
os
i
te
o
f
the
f
i
r
s
t
on
e.
T
he
m
ai
n
c
on
c
ern
of
t
he
f
i
l
t
erin
g
ph
as
e
i
s
to
de
tec
t
t
he
c
orr
up
t
ed
pi
x
e
l
s
(
no
i
s
e)
an
d
th
en
c
ha
ng
e
i
ts
v
al
ue
by
the
c
orr
ec
t
on
e,
whi
c
h
i
s
the
ne
ares
t
v
al
u
e
to
the
o
r
i
gi
n
al
.
T
h
i
s
e
na
b
l
es
t
he
r
em
ov
a
l
o
f
t
he
no
i
s
e
by
pr
es
erv
i
ng
oth
er
i
ma
ge
fe
atu
r
es
.
T
he
d
en
o
i
s
i
ng
me
tho
d
(
fi
l
t
erin
g)
or
r
es
torati
on
's
op
era
ti
o
n
i
n
t
h
e
no
i
s
y
i
m
ag
e
c
om
es
fr
o
m
t
he
un
c
orr
u
pte
d
p
i
x
el
s
[
36
].
T
hu
s
,
i
t
i
s
v
ery
i
mp
ortant
to
o
b
s
erv
e
(
mo
de
l
)
the
c
on
n
ec
ti
o
n
be
twee
n
th
es
e
no
i
s
e
-
fr
e
e
p
i
x
el
s
,
an
d
ho
w
th
e
pi
x
e
l
i
nte
ns
i
ty
c
ha
ng
es
thro
ug
h
the
i
m
ag
e.
T
h
i
s
g
i
v
es
t
he
op
po
r
t
un
i
ty
to
tr
ea
t
t
he
two
ma
i
n
po
i
nts
i
n
th
i
s
p
ha
s
e
(
de
tec
ti
on
an
d
s
up
r
es
s
i
on
of
no
i
s
e)
be
c
au
s
e
an
y
pi
x
e
l
t
ha
t
do
es
no
t
s
ub
j
ec
t
to
th
i
s
s
tat
us
(
mo
de
l
)
i
n
the
i
ma
g
e,
c
an
be
c
on
s
i
dr
ed
as
a
no
i
s
e,
an
d
s
ho
ul
d
r
ep
l
ac
e
i
ts
v
al
u
e
by
an
ot
he
r
v
al
u
e
de
r
i
v
ed
fr
om
ot
he
r
un
c
orr
up
te
d
p
i
x
el
s
an
d
s
u
bj
ec
t
to
the
i
nte
ns
i
ty
be
ha
v
i
or.
F
i
na
l
l
y
,
t
he
o
bta
i
ne
d
r
es
tored
i
ma
g
e
c
arr
i
es
th
e s
a
me
f
ea
t
ures
o
f th
e
orig
i
na
l
i
ma
ge
.
T
he
p
i
x
el
s
are
l
oc
a
ted
i
n
di
ff
erent
areas
i
n
th
e
i
m
a
ge
w
i
th
a
pa
r
t
i
c
ul
ar
ord
er
du
e
to
the
i
nt
en
s
i
ty
be
h
av
i
or.
T
he
i
nte
ns
i
ty
o
f
th
e
pi
x
e
l
v
ari
es
i
n
t
he
i
m
ag
e
gra
du
a
l
l
y
(
gr
a
di
e
nt
c
o
l
or)
or
c
on
s
tan
t
l
y
(
on
e
c
o
l
or)
.
T
h
es
e
i
nte
ns
i
ty
v
a
l
ue
s
ma
y
be
mo
de
l
ed
as
t
erms
o
f
an
arit
hm
eti
c
progr
es
s
i
on
w
i
th
a
c
om
mo
n
d
i
ffe
r
en
c
e
:
tak
e
a
c
on
s
ta
nt
v
a
l
ue
d
i
fere
nt
fr
om
0
i
n
the
c
as
e
o
f
gradi
en
t c
o
l
or a
nd
i
t
i
s
eq
u
al
t
o 0
i
n t
he
c
as
e
of
o
ne
c
o
l
or, as
i
t i
s
s
h
own
i
n F
i
gu
r
e
4.
T
he
prop
os
ed
de
n
oi
s
i
ng
m
eth
o
d
be
l
on
gs
to
t
he
s
pa
t
i
al
d
om
a
i
n
f
i
l
teri
ng
a
nd
i
t
e
x
pl
oi
ts
thi
s
ex
i
s
ti
n
g
c
h
arac
teri
s
t
i
c
i
n
th
e
i
m
ag
e
i
l
l
us
tr
ate
d
i
n
F
i
gu
r
e
5,
where
e
ac
h
p
i
x
e
l
b
el
o
ng
s
t
o
o
ne
or
ma
ny
s
eq
u
en
c
es
of
p
i
x
el
v
al
u
es
.
T
h
en
,
th
i
s
c
ha
r
ac
teri
s
t
i
c
c
an
be
mo
d
el
e
d
us
i
ng
a
ma
th
em
a
ti
c
a
l
c
on
c
e
pt
c
a
l
l
e
d
“
arit
hm
e
ti
c
progr
es
s
i
o
n
”
t
o
g
et
t
he
al
terna
te
pi
x
e
l
of
the
c
orr
u
pte
d
on
e,
whi
c
h
s
ho
u
l
d
be
l
on
g t
o t
h
i
s
s
eq
ue
nc
e.
T
he
F
i
gu
r
e
5
i
l
l
us
tr
ate
s
h
o
w
the
i
nt
en
s
i
ty
of
p
i
x
el
s
c
h
an
ge
s
i
n
the
i
ma
g
e
wi
t
ho
ut
no
i
s
e,
where
e
ac
h
pi
x
e
l
th
at
b
el
o
ng
s
to
on
e
or
mo
r
e
pa
r
t
i
c
ul
ar
order
s
(
s
eq
ue
nc
es
)
of
i
n
teg
er
n
um
b
ers
go
fr
om
0
to
25
5
(
0,
1,2
,.
.,2
55
)
,
for
ex
am
p
l
e
t
he
p
i
x
el
wi
th
v
al
ue
4
2
be
l
on
gs
to
m
an
y
s
eq
ue
nc
es
(
arit
hm
eti
c
pro
gres
s
i
on
)
wh
i
c
h
are
i
nd
i
c
a
ted
i
n
th
e
fi
gu
r
e
by
bl
ue
arr
ows
,
where
e
ac
h
s
eq
ue
nc
e
i
s
c
ha
r
ac
teri
z
ed
by
i
ts
pa
r
t
i
c
ul
ar c
om
mo
n d
i
ff
erenc
e
.
A
s
i
t
me
nt
i
on
ed
i
n
th
e
prev
i
ou
s
c
ha
p
ter,
the
ar
i
thm
et
i
c
progr
es
s
i
o
n
i
s
c
ha
r
ac
teri
s
ed
by
a
term
a
nd
a
c
o
ns
tan
t
,
wh
i
l
e
i
n
the
r
ea
l
i
ma
ge
the
d
i
ffe
r
en
c
e
b
etwe
en
pi
x
el
s
i
n
a
s
eq
u
en
c
e
of
p
i
x
el
s
i
s
n
ot
us
a
l
l
y
c
on
s
t
an
t
i
n
m
os
t
c
as
e,
the
r
efo
r
e t
o
mo
d
el
th
es
e s
e
qu
en
c
e
as
an
ar
i
th
me
t
i
c
pr
og
r
es
s
i
on
w
e n
ee
d t
o
es
ti
m
ate
a c
o
mm
on
di
ff
erenc
e
fr
om
the
s
e
d
i
f
ferenc
es
by
ma
k
i
ng
i
ts
a
v
erage
as
the
es
ti
ma
t
or
of
.
F
or
ex
am
p
l
e,
i
n
t
he
c
as
e
of
a
s
eq
ue
nc
e
of
pi
x
e
l
s
wi
t
h
the
f
ol
l
ow
i
n
g
v
al
ue
s
a
nd
d
i
ffe
r
e
nc
es
(
1
,
2
,
3
,
4
)
∶
where
:
1
=
50
−
44
=
6
(
7)
2
=
57
−
50
=
7
(
8)
3
=
65
−
57
=
8
(9
)
4
=
69
−
65
=
4
(
10
)
T
he
es
ti
ma
t
or
̂
of
i
s
c
al
c
u
l
ate
d
fr
om
(
7),
(
8),
(
9),
(
1
0)
by
the
av
erage
of
1
,
2
,
3
,
4
.
44
50
57
6
5
69
1
2
3
4
Evaluation Warning : The document was created with Spire.PDF for Python.
◼
IS
S
N: 16
93
-
6
93
0
T
E
L
KO
MNIK
A
V
ol
.
17
,
No
.
6,
D
ec
em
b
er 20
19
:
29
4
8
-
2958
2954
̂
=
(
1
+
2
+
3
+
4
)
4
⁄
≃
6
(
11
)
i
n
th
i
s
c
as
e,
the
pi
x
e
l
c
an
be
c
on
s
i
d
ered
as
a
term
o
f
arit
hm
eti
c
progr
es
s
i
on
,
a
n
d
to
r
es
tore
i
ts
v
al
ue
w
e
ap
pl
y
th
e
r
ul
es
of
c
al
c
ul
a
ti
o
n
of
un
k
no
w
n
A
P
's
term
to
ge
t
t
he
m
os
t
c
orr
ec
t
v
al
ue
f
or
tha
t
pi
x
e
l
.
A
c
c
ordi
ng
to
th
at
,
th
i
s
m
ath
e
ma
t
i
c
al
c
on
c
e
pt
(
arit
h
me
t
i
c
pro
gres
s
i
on
)
c
a
n
b
e
us
e
d
as
a t
oo
l
f
or r
em
ov
i
ng
the
no
i
s
e a
n
d res
tori
ng
t
he
d
eg
r
a
d
ed
i
ma
g
e.
F
i
gu
r
e
4
.
I
nte
ns
i
ty
v
aria
t
i
o
n
i
n
i
ma
ge
F
i
gu
r
e
5
.
T
h
e a
p
pe
aranc
e
of
p
i
x
el
s
v
al
ue
s
i
n
no
i
s
e
-
fr
e
e i
ma
g
e
3.2
.1
.
No
is
e
Det
e
ctio
n
T
he
eff
i
c
i
en
c
y
of
the
pro
po
s
ed
d
en
o
i
s
i
n
g
me
t
ho
d
r
e
l
i
es
on
tw
o
c
r
uc
i
al
ph
as
es
,
on
e
of
the
m
i
s
th
e
d
ete
c
t
i
on
no
i
s
e
whi
c
h
ai
ms
to
ma
k
e
the
tr
ea
tm
en
t
op
er
ati
on
(
th
e
c
ha
ng
e)
c
on
fi
n
ed
on
l
y
t
o
th
e
c
orr
up
t
ed
pi
x
e
l
s
.
T
hu
s
,
t
hi
s
s
ec
ti
on
s
h
ows
ho
w
t
o
d
i
s
ti
n
gu
i
s
h
b
etwe
en
the
c
orr
u
pte
d
(
no
i
s
e)
a
nd
un
c
orr
up
t
ed
pi
x
el
s
.
Us
ua
l
l
y
as
i
n
the
c
as
e
of
oth
er
i
m
ag
e
fi
l
ter
i
ng
me
th
od
s
wh
i
c
h
ad
op
t
the
no
i
s
e
de
te
c
t
i
o
n
p
ha
s
e
i
n
i
ts
proc
es
s
i
n
g,
the
f
i
l
t
er
i
s
fi
r
s
t
l
o
ok
i
ng
f
or
the
c
orr
up
ted
pi
x
e
l
(
de
tec
ti
on
)
an
d
the
n
l
oo
k
i
n
g
for
the
wi
tc
h
v
al
u
e
(
al
tern
ate
pi
x
e
l
)
tha
t
s
h
ou
l
d
r
ep
l
a
c
e
thi
s
pi
x
e
l
,
i
n
order
to
tr
ea
t
o
nl
y
the
c
orr
u
pte
d
p
i
x
el
s
an
d
k
ee
p
ot
he
r
s
wi
th
the
s
am
e
v
al
u
es
(
ob
j
ec
ti
v
e
of
no
i
s
e
detec
t
i
on
ph
as
e).
I
n
ou
r
a
pp
r
oa
c
h
we
s
ee
k
t
o
ac
hi
e
v
e
tha
t
o
bj
ec
ti
v
e
ba
s
e
d
o
n
the
A
P
,
wh
ere
the
pi
x
e
l
w
hi
c
h
do
es
n
ot
be
l
on
g
to
an
y
s
eq
u
en
c
es
of
n
um
b
ers
wi
th
a
de
ter
m
i
ne
d
c
om
mo
n
di
ff
erenc
e
, wi
l
l
be
c
o
ns
i
de
r
ed
as
a
no
i
s
e (c
orr
up
te
d p
i
x
el
)
, a
s
i
t
i
s
c
l
e
arly
s
ee
n i
n
F
i
gu
r
e
6.
F
i
gu
r
e
6
.
T
h
e a
p
pe
aranc
e
of
p
i
x
el
s
v
al
ue
s
i
n n
oi
s
y
Im
ag
e (s
al
t &
pe
pp
er)
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
MNIK
A
IS
S
N: 1
69
3
-
6
93
0
◼
P
r
og
r
es
s
i
o
n a
pp
r
oa
c
h
fo
r
i
ma
ge
de
no
i
s
i
ng
(
B
i
l
a
l
C
ha
r
mo
ut
i
)
2955
3.2
.2
.
Re
sto
r
ing
t
h
e
P
ixe
l
s (No
ise
S
u
p
p
r
e
ss
ion
)
T
hi
s
s
tep
i
s
th
e
s
ec
on
d
c
r
uc
i
al
ph
as
e
i
n
t
he
f
i
l
teri
ng
proc
es
s
.
B
as
e
d
o
n
t
he
prev
i
ou
s
s
ub
-
ph
as
e,
ev
ery
pi
x
el
w
hi
c
h
i
s
c
on
s
i
de
r
e
d
as
no
i
s
e
wi
l
l
be
r
e
pl
ac
ed
by
th
e
c
orr
ec
t
ex
tr
ac
ted
v
al
ue
an
d
k
e
ep
s
ot
he
r
p
i
x
el
s
wi
th
t
he
s
am
e
v
al
u
es
to
ob
tai
n
th
e
de
s
i
r
ed
i
ma
g
e.
T
hu
s
,
the
q
ue
s
ti
on
he
r
e
i
s
ho
w
to
ob
ta
i
n
t
hi
s
c
orr
ec
t
v
al
ue
,
a
nd
to
an
s
wer
t
hi
s
q
ue
s
ti
on
we
us
e
as
an
ex
am
p
l
e,
F
i
gu
r
e
5,
wh
i
c
h
de
no
tes
the
no
i
s
e
-
fr
ee
i
ma
ge
,
an
d
F
i
gu
r
e
6
be
l
ow,
w
hi
c
h
i
l
l
us
tr
ate
s
the
di
s
tr
i
b
ut
i
on
of
c
orr
u
pte
d
p
i
x
el
s
v
a
l
ue
s
(
as
s
i
gn
e
d
by
r
ed
c
i
r
c
l
es
)
i
n
the
no
i
s
y
i
ma
ge
of
t
he
s
am
e o
r
i
gi
na
l
i
ma
g
e i
n Fi
gu
r
e 5
.
It
i
s
c
l
ea
r
l
y
s
ee
n
th
at,
fr
o
m
the
fi
r
s
t
c
as
e
(
no
i
s
e
-
fr
ee
i
ma
ge
)
t
o
the
s
ec
o
nd
o
ne
(
no
i
s
y
i
ma
ge
)
,
th
e
pi
x
e
l
wi
th
th
e
v
al
u
e
42
as
s
i
gn
e
d
by
gree
n
c
i
r
c
l
e
i
n
F
i
gu
r
e
5
,
was
c
ha
ng
ed
t
o
be
on
the
v
al
ue
25
5
(
s
al
t
n
oi
s
e)
as
s
i
gn
e
d
by
r
ed
c
i
r
c
l
e
i
n
F
i
gu
r
e
6
.
In
order
t
o
r
es
to
r
e
the
m
i
s
s
i
ng
(
orig
i
na
l
)
v
al
ue
of
the
c
o
r
r
up
ted
pi
x
e
l
,
th
i
s
l
as
t
i
s
tak
en
as
an
un
k
n
own
t
erm
o
f
o
ne
of
the
ex
i
s
ti
n
g a
r
i
thm
eti
c
pr
og
r
es
s
i
on
s
s
urr
ou
nd
i
n
g t
h
i
s
pi
x
el
.
In
thi
s
c
as
e
th
e
c
orr
up
te
d
pi
x
e
l
i
s
s
urr
ou
nd
ed
by
two
s
eq
u
en
c
es
,
th
e
fi
r
s
t
on
e
i
s
i
nd
i
c
ate
d
by
a
gree
n
arr
ow
an
d
de
no
t
ed
by
t
h
e
terms
:
44
50
57
65
6
9
.
T
he
s
ec
on
d
s
eq
ue
nc
e
i
s
i
nd
i
c
ate
d
by
a
bl
ue
arr
o
w
an
d
de
no
te
d
by
t
he
t
erms
:
91
12
3
13
9
.
B
y
c
om
pa
r
i
ng
the
s
e
s
eq
ue
nc
e
s
,
we
s
e
e
th
at
t
h
e
fi
r
s
t
o
ne
c
o
nta
i
ns
a
hi
gh
er
nu
m
be
r
o
f
ter
ms
,
an
d
w
i
th
di
ff
erenc
es
c
l
os
er
to
e
ac
h
ot
he
r
.
T
hu
s
,
the
f
i
r
s
t
s
eq
ue
nc
e
i
s
m
o
r
e
ab
l
e
to
be
m
od
e
l
e
d
as
an
ari
thm
eti
c
progr
es
s
i
on
.
T
he
c
om
mo
n
di
ffe
r
e
nc
e
i
s
es
ti
m
ate
d
us
i
ng
(
11
)
.
W
i
t
h
r
eg
ard
t
o
the
tr
ea
te
d
i
ma
ge
i
n
F
i
gu
r
e
6,
th
e
c
orr
up
te
d
p
i
x
el
P
by
th
e
v
al
ue
"
25
5"
i
s
c
on
s
i
de
r
e
d
as
a
n
u
nk
no
wn
t
erm
o
f
thi
s
A
P
,
an
d
i
t c
a
n b
e res
to
r
ed
us
i
n
g
i
ts
ne
i
gh
bo
r
"
44
"
by
the
ne
x
t (12)
.
=
44
−
6
=
38
(
12
)
Refe
r
r
i
n
g
t
o
t
he
orig
i
na
l
i
m
ag
e
i
n
F
i
g
ure
5,
t
he
orig
i
n
a
l
v
al
ue
of
the
p
i
x
el
i
s
"
4
2"
,
an
d
the
ob
tai
ne
d
v
a
l
u
e
by
the
propos
e
d
me
t
ho
d
i
s
"
38
"
,
whi
l
e
i
n
the
c
as
e
t
ha
t
we
u
s
e
for
ex
am
p
l
e
the
me
di
an
fi
l
ter,
t
he
r
es
tor
ed
pi
x
e
l
tak
es
the
v
a
l
ue
"
6
5"
.
A
s
i
t
i
s
,
the
pr
op
os
e
d
me
th
od
g
i
v
es
a
r
es
tored v
al
ue
w
hi
c
h
i
s
c
l
os
er to th
e o
r
i
gi
na
l
, c
om
pa
r
e
d
to
th
e m
ed
i
an
f
i
l
teri
ng
.
3.3.
E
va
luat
ion
of
F
ilteri
n
g
Ap
p
r
o
ac
h
In
t
hi
s
l
as
t
s
tep
,
t
he
i
n
pu
t
(
no
i
s
y
i
ma
ge
)
an
d
o
utp
u
t
(
r
es
tored
i
m
ag
e)
of
th
e
pro
po
s
ed
fi
l
t
er
ar
e
us
e
d
to
c
al
c
u
l
ate
t
he
ev
al
ua
t
i
on
p
arame
ters
whi
c
h
are
M
S
E
a
nd
P
S
NR
for
c
om
pa
r
i
s
o
n
wi
th
o
th
er de
no
i
s
i
ng
me
t
ho
ds
, i
n
order
t
o e
v
a
l
ua
t
e t
h
e
pe
r
forma
nc
e o
f th
i
s
fi
l
ter.
4
. Re
sult
s
a
n
d
An
aly
s
is
In
th
i
s
s
tud
y
a
no
th
er
way
i
s
tak
en
to
d
ev
el
op
a
ne
w
de
n
oi
s
i
ng
me
th
od
t
ha
t
ai
ms
to
r
em
ov
e
the
i
mp
u
l
s
e
no
i
s
e
f
r
om
i
ma
ge
by
pres
erv
i
ng
t
he
f
i
n
e
d
eta
i
l
s
of
i
m
ag
e
fe
atu
r
es
s
uc
h
as
the
e
dg
es
,
an
d
av
o
i
di
ng
ot
he
r
eff
ec
ts
of
f
i
l
t
erin
g,
f
or
ex
am
p
l
e
t
he
bl
ur.
T
hi
s
de
n
oi
s
i
n
g
m
eth
od
tr
ea
ts
the
i
m
pu
l
s
e
no
i
s
e
an
d
i
t
i
s
ex
pe
c
ted
to
g
i
v
e
g
oo
d
qu
a
l
i
ty
of
r
es
torat
i
o
n
whe
the
r
v
i
s
ua
l
l
y
or
c
om
pu
t
ati
on
a
l
,
w
he
r
e
as
i
t
i
s
al
s
o
ex
p
ec
ted
t
o
b
e
ex
te
nd
ed
i
n
the
fut
ure
to
an
o
th
er
ty
pe
of
no
i
s
e
s
uc
h a
s
ad
d
i
t
i
v
e o
r
m
ul
t
i
p
l
i
c
ati
v
e
no
i
s
e
.
In
ord
er
to
tes
t
the
p
erfor
m
an
c
e
o
f
t
he
propos
e
d
de
n
o
i
s
i
ng
tec
h
ni
qu
e,
e
i
gh
t
i
ma
g
es
as
s
am
pl
i
n
g h
av
e b
e
en
us
ed
a
nd
th
ey
are
pres
en
t
ed
i
n
Fi
gu
r
e
7
(
B
o
ats
,
P
ep
pe
r
s
, Ho
us
e,
M
an
dr
i
l
l
)
.
T
he
r
es
ul
t
of
th
os
e
fi
l
ter
i
ng
me
th
od
s
i
s
i
nd
i
c
a
ted
v
i
s
u
al
l
y
(
pi
c
tures
)
i
n
F
i
g
ure
8
an
d
qu
an
ti
t
ati
v
e
l
y
P
S
N
R
i
n
T
ab
l
e
1,
w
i
th
th
e
prop
os
ed
t
ec
hn
i
qu
e,
i
n
s
ev
eral
v
al
ue
s
of
no
i
s
e
a
mo
un
t
as
s
i
g
ne
d
by
pe
r
c
en
ta
ge
.
T
he
pr
i
ma
r
y
i
mp
l
e
me
nt
ati
o
n
of
thi
s
fi
l
ter
t
o
r
e
mo
v
e
t
h
e
s
al
t
an
d
p
ep
p
er
no
i
s
e,
t
o
g
i
v
e
ac
c
ep
tab
l
e
r
es
ul
ts
c
o
mp
ar
ed
t
o
s
om
e
m
eth
od
s
w
hi
c
h
are
c
on
s
i
d
ered
as
ef
fi
c
i
e
nt
m
eth
o
ds
for
r
em
ov
i
n
g
i
m
pu
l
s
e
no
i
s
e
fr
om
i
m
ag
es
:
S
tan
da
r
d
m
e
di
a
n
fi
l
ter
(
S
MF)
,
we
i
g
hte
d
me
d
i
an
f
i
l
t
er
(W
MF)
,
di
r
ec
ti
o
na
l
we
i
gh
t
ed
me
di
an
f
i
l
t
er
(
DW
MF)
.
T
he
r
es
ul
ts
pres
e
nte
d
i
n
F
i
gu
r
e
8
an
d
the
T
a
bl
e
1,
i
l
l
us
tr
at
e
t
ha
t
t
he
pro
po
s
ed
de
no
i
s
i
ng
tec
hn
i
qu
e
gi
v
es
an
ac
c
ep
tab
l
e
pe
r
f
ormanc
e
c
om
pa
r
e
d
t
o
th
e
ex
i
s
ti
n
g
me
th
od
s
w
he
th
er
v
i
s
u
al
l
y
or
qu
an
t
i
tat
i
v
el
y
wi
t
h
P
S
N
R.
T
he
s
e
r
es
ul
ts
i
l
l
us
tr
ate
tha
t
the
prop
os
ed
me
th
od
h
as
s
uc
c
ee
de
d
to
ge
t
a
c
l
os
er
r
es
tore
d
i
ma
g
e
to
th
e
ori
gi
na
l
i
ma
ge
,
by
way
of
r
ep
l
ac
i
n
g
the
c
orr
up
t
ed
pi
x
e
l
by
th
e
mo
s
t c
orr
ec
t o
n
e.
Evaluation Warning : The document was created with Spire.PDF for Python.
◼
IS
S
N: 16
93
-
6
93
0
T
E
L
KO
MNIK
A
V
ol
.
17
,
No
.
6,
D
ec
em
b
er 20
19
:
29
4
8
-
2958
2956
(
a)
(
b)
(
e)
(
f)
(
c
)
(
d)
(
g)
(
h)
F
i
gu
r
e
7.
T
h
e
s
am
pl
i
ng
i
m
a
ge
s
(a
-
d)
wi
th
ou
t
no
i
s
e
(
orig
i
n
al
i
ma
ge
s
)
an
d (e
-
h)
w
i
th
20
% s
al
t
a
nd
pe
p
pe
r
n
oi
s
e
(
a)
(
b)
(
e)
(
f)
(
c
)
(
d)
(
g)
(
h)
F
i
gu
r
e
8.
C
om
p
aris
on
of
r
e
s
torati
o
n res
ul
ts
of
t
he
S
MF
wi
th
th
e
prop
os
ed
t
ec
hn
i
qu
e f
or i
ma
ge
s
c
orr
up
ted
by
fi
x
ed
-
v
a
l
ue
d i
mp
u
l
s
e n
oi
s
e
(a
-
d)
fi
l
t
ered
wi
t
h S
MF
(e
-
h)
fi
l
t
ered
wi
t
h t
h
e p
r
o
po
s
ed
me
th
od
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
MNIK
A
IS
S
N: 1
69
3
-
6
93
0
◼
P
r
og
r
es
s
i
o
n a
pp
r
oa
c
h
fo
r
i
ma
ge
de
no
i
s
i
ng
(
B
i
l
a
l
C
ha
r
mo
ut
i
)
2957
T
ab
l
e
1.
C
om
pa
r
i
s
o
n o
f
Re
s
torati
o
n Re
s
u
l
ts
i
n P
S
NR
f
or Ima
ge
s
Cor
r
up
ted
by
Fix
ed
-
v
al
ue
d
Imp
ul
s
e
N
oi
s
e (S
al
t
&
P
e
pp
er)
I
mag
e
s
\
Me
t
h
o
d
s
S
MF
W
MF
D
W
MF
P
r
o
p
o
s
e
d
B
o
a
t
s
20%
60%
33
.
18
22
.
40
35
.
55
24
.
74
37
.
11
26
.
34
38
.
01
27
.
95
P
e
p
p
e
r
s
20%
60%
35
.
25
25
.
45
37
.
12
27
.
20
39
.
11
30
.
34
39
.
82
31
.
55
H
o
u
s
e
20%
60%
33
.
66
22
.
13
35
.
85
24
.
53
37
.
24
27
.
54
38
.
91
28
.
42
Man
d
r
il
l
2
0
%
60%
31
.
21
22
.
64
34
.
32
25
.
76
36
.
11
26
.
34
37
.
63
27
.
98
5
. Con
clus
ion
T
hi
s
work
i
s
c
arr
i
ed
ou
t to
t
r
ea
t th
e
probl
em
of
i
m
pu
l
s
e
no
i
s
e i
n t
he
i
m
ag
e
, wh
i
c
h t
hreats
i
ts
v
i
s
ua
l
qu
al
i
ty
an
d
m
ak
e
oth
er
i
ma
g
e
proc
es
s
i
n
g
s
uc
h
as
s
eg
me
nta
t
i
o
n
a
nd
c
om
pres
s
i
o
n
mo
r
e
di
ff
i
c
ul
t.
T
h
e
c
ho
s
en
way
for
t
ha
t
i
s
t
he
s
p
ati
al
fi
l
teri
ng
i
n
wh
i
c
h
t
he
proc
es
s
i
s
s
i
mp
l
e
a
nd
de
a
l
s
d
i
r
ec
tl
y
w
i
th
th
e
pi
x
e
l
s
as
do
ne
i
n
th
e
me
d
i
a
n
an
d
m
ea
n
f
i
l
t
ers
.
T
he
ad
op
t
ed
ma
t
he
ma
t
i
c
al
c
on
c
ep
t
i
n
t
hi
s
s
tu
dy
i
s
t
he
A
r
i
t
hm
e
ti
c
P
r
og
r
es
s
i
o
n,
w
hi
c
h
a
l
l
ows
th
e
mo
d
el
i
ng
of
the
be
h
av
i
o
ur
or
v
aria
t
i
on
of
pi
x
e
l
s
i
nte
ns
i
ty
i
n
t
he
i
m
ag
e
wi
t
h
h
i
g
h
pr
ec
i
s
i
on
.
In
order
to
de
r
i
v
e
n
ew
v
al
ue
s
for
the
c
orr
up
te
d
pi
x
e
l
s
th
at
are
s
ub
j
ec
t
to
th
e
s
a
me
f
ea
tures
an
d
gi
v
e
a
p
erfec
t
r
es
torat
i
o
n
of
the
i
ma
g
e
by
pres
erv
i
n
g
th
e
fi
n
e
de
t
ai
l
s
(
orig
i
na
l
i
ty
o
f
i
ma
ge
)
.
F
i
na
l
l
y
,
th
i
s
de
n
oi
s
i
ng
me
th
od
i
s
prov
en
as
an
eff
i
c
i
e
nt
wa
y
to
r
es
tore
the
i
ma
ge
c
o
mp
are
d
to
the
c
urr
en
t
f
a
mi
l
i
ar
fi
l
ters
i
n
the
as
p
ec
t o
f
ev
al
ua
t
i
on
c
r
i
t
eria
.
B
es
i
d
es
w
ha
t
th
i
s
d
en
oi
s
i
n
g
m
eth
od
prov
i
de
s
as
a
n
e
x
ten
s
i
on
o
f
t
he
r
an
g
e
of
s
ol
uti
o
ns
ag
a
i
ns
t
th
i
s
pro
bl
e
m
o
f
no
i
s
e
i
n
t
he
i
m
ag
e,
i
t
al
s
o
ho
l
ds
l
i
mi
t
ati
on
s
a
bo
ut
the
th
i
ng
s
tha
t
s
ho
u
l
d
be
ex
po
s
e
d.
O
ne
l
i
mi
tat
i
on
i
s
tha
t
t
he
s
tu
dy
i
s
c
on
s
tr
ai
n
ed
by
th
e
nu
mb
er
of
s
a
mp
l
i
n
g
i
m
ag
es
whi
c
h
t
ota
l
fou
r
i
ma
ge
s
(
B
oa
ts
,
P
ep
pe
r
s
,
Ho
us
e,
Ma
nd
r
i
l
l
)
.
W
he
n
t
he
n
um
b
er
of
s
a
mp
l
i
ng
i
ma
ge
s
i
s
hi
gh
er,
t
he
m
et
ho
d
w
i
l
l
prov
e
i
ts
eff
i
c
i
en
c
y
i
n
r
es
tori
n
g
al
l
i
ma
g
e
m
od
e
l
s
.
A
n
oth
er
l
i
m
i
ta
ti
o
n
i
s
a
bo
ut
t
he
p
erf
ormanc
e
o
f
the
f
i
l
ter
wh
i
c
h
de
c
r
ea
s
es
wi
th
th
e
r
i
s
e
of
the
n
oi
s
e's
am
ou
nt
i
n
th
e
n
oi
s
y
i
m
ag
e,
wh
i
l
e
th
i
s
drop
s
h
ou
l
d
be
v
ery
s
l
ow
t
o
pres
erv
e
the
s
tab
i
l
i
ty
i
n
the
f
i
l
t
erin
g
pe
r
f
orma
nc
e.
F
urthermor
e
,
th
e
s
tud
y
pro
po
s
es
a
d
en
o
i
s
i
n
g
tec
h
ni
q
ue
c
on
f
i
n
ed
i
n
r
em
ov
i
n
g
on
e
ty
pe
of
n
oi
s
e
wh
i
c
h
i
s
i
mp
ul
s
e
no
i
s
e.
Thi
s
i
s
wh
en
a
grea
t n
um
be
r
o
f
i
ma
g
es
us
e
d
i
n
ma
ny
fi
e
l
ds
i
n
ou
r
l
i
v
es
are
c
orr
up
ted
by
th
e
two
oth
er
ty
pe
s
of
no
i
s
e
(
ad
di
t
i
v
e
an
d
mu
l
t
i
p
l
i
c
a
ti
v
e), f
or ex
am
pl
e
the
m
ed
i
c
al
i
m
ag
e
tha
t
i
s
u
s
ua
l
l
y
c
orr
up
ted
by
th
e
s
pe
c
k
l
e n
o
i
s
e.
In
t
hi
s
de
n
oi
s
i
ng
tec
h
ni
qu
e,
t
he
r
e
are
fo
ur
p
oi
nts
w
hi
c
h
are
bri
efl
y
me
nti
on
e
d
i
n
t
hi
s
s
ec
ti
on
an
d
c
an
be
us
e
d
to
de
v
e
l
o
p
th
i
s
m
eth
o
d
du
e
t
o
t
he
i
nf
l
ue
nc
e
on
t
he
f
i
l
teri
ng
pe
r
forma
nc
e.
T
he
fi
r
s
t
po
i
n
t
i
s
t
he
nu
mb
er
of
arit
hm
e
ti
c
progr
es
s
i
o
n
t
erms
,
c
ho
s
e
n
to
de
term
i
ne
thi
s
s
eq
ue
nc
e
.
T
he
s
ec
on
d
i
s
the
ma
nn
er
of
es
ti
m
ati
n
g
t
he
c
om
mo
n
di
f
ferenc
e
,
of
th
e
A
P
.
I
n
thi
s
s
tud
y
,
i
s
es
ti
ma
ted
b
y
the
av
erage
of
th
e
di
f
fer
en
c
es
be
twe
en
th
e
pi
x
e
l
s
of
thi
s
A
P
,
where
i
t
i
s
p
os
s
i
bl
e
to
us
e
oth
er
m
an
n
ers
.
T
he
c
ho
i
c
e
of
the
mo
s
t
s
u
i
tab
l
e
s
e
qu
e
nc
e
i
s
s
el
ec
t
ed
to
r
es
tor
e
t
he
c
orr
up
te
d
pi
x
el
.
T
he
l
as
t
po
i
nt
i
s
the
c
h
arac
teri
s
ti
c
s
of
the
A
P
(
c
on
v
ergenc
e
of
a
s
eq
ue
nc
e)
wh
i
c
h
c
an
be
u
s
ed
to
s
el
ec
t
the
ap
propr
i
a
te
A
P
an
d
th
e
m
os
t
c
orr
ec
t
pi
x
e
l
v
a
l
ue
to
r
es
tore th
e
c
orr
up
te
d o
n
e.
Ackno
w
ledg
em
ent
T
hi
s
w
ork
was
fi
n
an
c
i
al
l
y
s
up
po
r
te
d
by
M
i
n
i
s
tr
y
of
E
du
c
at
i
o
n
Ma
l
ay
s
i
a
(
MO
E
)
un
de
r
F
un
da
me
n
tal
Res
e
arc
h Gr
an
t
S
c
he
m
e (F
RG
S
)
(
(
Ref:
F
RG
S
/1/
20
1
9/S
T
G
0
6/UNI
MA
P
/02
/3).
Ref
er
en
ce
s
[1
]
Ku
m
a
r A
,
Sh
a
i
k
F.
I
m
a
g
e
p
ro
c
e
s
s
i
n
g
i
n
d
i
a
b
e
ti
c
re
l
a
te
d
c
a
u
s
e
s
.
Sp
r
i
n
g
e
r
.
2
0
1
5
.
[2
]
Li
X.
I
m
a
g
e
re
s
to
ra
t
i
o
n
:
Fu
n
d
a
m
e
n
t
a
l
s
a
n
d
a
d
v
a
n
c
e
s
:
CR
C Pre
s
s
.
2
0
1
2
.
[3
]
Bo
y
a
t
A
K
,
J
o
s
h
i
BK
.
A
re
v
i
e
w
p
a
p
e
r:
n
o
i
s
e
m
o
d
e
l
s
i
n
d
i
g
i
ta
l
i
m
a
g
e
p
ro
c
e
s
s
i
n
g
.
a
rXi
v
p
re
p
ri
n
t
2
0
1
5
.
a
rXi
v
:
1
5
0
5
.0
3
4
8
9
.
[4
]
Ku
n
d
u
r
D
,
Ha
tz
i
n
a
k
o
s
D.
Bl
i
n
d
i
m
a
g
e
d
e
c
o
n
v
o
l
u
ti
o
n
.
IE
EE
s
i
g
n
a
l
p
ro
c
e
s
s
i
n
g
m
a
g
a
z
i
n
e
.
1
9
9
6
;
1
3
(3
):
43
-
6
4
.
[5
]
Li
X.
Im
a
g
e
re
c
o
v
e
r
y
v
i
a
h
y
b
ri
d
s
p
a
r
s
e
r
e
p
re
s
e
n
ta
ti
o
n
s
:
A
d
e
te
r
m
i
n
i
s
t
i
c
a
n
n
e
a
l
i
n
g
a
p
p
r
o
a
c
h
.
IEE
E
J
o
u
rn
a
l
o
f
Se
l
e
c
t
e
d
T
o
p
i
c
s
i
n
Si
g
n
a
l
Pro
c
e
s
s
i
n
g
.
2
0
1
1
;
5
(
5
):
953
-
9
6
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.