Indonesi
an
Journa
l
of
El
ect
ri
cal Engineer
ing
an
d
Comp
ut
er
Scie
nce
Vo
l.
9
, No
.
2
,
Febr
ua
ry
201
8
,
pp.
35
4
~
356
IS
S
N:
25
02
-
4752
, DO
I: 10
.11
591/
ijeecs
.
v9.i
2
.
pp
354
-
356
354
Journ
al h
om
e
page
:
http:
//
ia
es
core.c
om/j
ourn
als/i
ndex.
ph
p/ij
eecs
Image R
esolution
Enhanc
ement U
sing T
ransform
Sy
ed
Naz
eebu
rrehma
n
1
, Mo
ha
mee
d A
li
H
ussa
in
2
1
Resea
r
ch
scho
l
ar,
Inform
at
ion
Te
chn
o
log
y
,
AM
ET
Univer
si
t
y,
Chenn
ai
2
Depa
rtment
of
computer
sci
ence,
KL
Univer
si
t
y,
Vij
a
y
awa
d
a
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Oct
18
, 201
7
Re
vised
Dec
2
3
, 2
01
7
Accepte
d
Ja
n
10
, 2
01
8
In
thi
s
proj
ect,
i
nte
rrupt
ion
base
d
image
r
esolut
i
on
enha
n
ce
m
ent
techniqu
e
using
Discre
te
W
ave
le
t
Tr
ansf
orm
(D
W
T)
with
high
-
fre
quency
sub
bands
obta
in
ed
is
prop
osed.
Input
ima
ges
are
dec
om
p
osed
b
y
using
DW
T
in
thi
s
proposed
enha
n
ce
m
ent
techniqu
e.
Inve
rse
DW
T
is
used
to
gene
rat
e
a
n
ew
resolut
ion
enha
n
ce
d
image
from
the
in
te
rpo
la
t
ion
of
high
-
fr
eque
n
c
y
sub
band
ima
ges
and
the
input
low
-
reso
lut
ion
image
.
I
nte
rm
edi
a
te
sta
ge
has
been
proposed
for
est
imati
ng
th
e
h
igh
fre
quen
c
y
sub
bands
to
ac
hi
ev
e
a
sharpe
r
image.
I
t
has
be
en
te
st
ed
on
ben
chmark
images
from
publi
c
da
tabase.
Pe
ak
Signal
-
To
-
Noise
Rat
io
(PS
NR)
and
visual
r
esult
s
show
the
dom
ina
nce
of
the
proposed
techni
que
over
the
pr
edi
c
ta
bl
e
and
st
at
e
-
of
-
art
image
resolut
ion
enha
nc
ement te
c
hnique
s.
Ke
yw
or
d
s
:
D
WT
PSN
R
Sate
ll
it
e
E
nh
an
ce
m
ent
Copyright
©
201
8
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights
reserv
ed
.
Corres
pond
in
g
Aut
h
or
:
Syed
Nazeeb
ur
reh
m
an,
Re
search
schol
ar, Inf
or
m
at
ion
Tech
nolo
gy,
AMET
Un
i
versi
ty
,
Chen
nai
.
1.
INTROD
U
CTION
W
it
h
t
he
recen
t
ad
van
ces
in
low
-
c
os
t
im
agi
ng
s
olu
ti
ons
a
nd
in
creasi
ng
st
or
a
ge
ca
pacit
ie
s,
the
re
is
a
n
increase
d
dem
and
for
bette
r
i
m
age
qu
al
it
y
in
a
wide
var
ie
t
y
of
ap
plica
ti
on
s
in
vo
l
ving
both
im
age
and
vid
e
o
processi
ng.
Wh
il
e
it
is
pr
efe
r
able
to
ac
qu
i
re
i
m
age
data
at
a
higher
res
olut
ion
to
be
gin
w
it
h,
one
ca
n
im
agine
a w
ide
r
a
nge
of sce
nar
i
os
where it
is tec
hnic
al
l
y no
t feasi
bl
e.
2.
BACKG
ROU
ND
Con
tra
st
enh
a
nc
e
m
ent
us
ing
m
ini
m
u
m
m
ea
n
br
i
gh
t
ness
er
ror
bi
-
histo
gr
a
m
equ
al
iz
at
ion
is
pr
esented
in
[
1].
T
he
c
onser
vato
ry
of
Bi
-
Histo
gr
am
Eq
ualiz
at
ion
(
BB
HE)
ref
e
rr
e
d
to
as
le
ast
a
m
ou
nt
of
bi
-
histogram
equ
al
iz
at
io
n
of
m
ean
br
ig
h
tn
ess
error
to
giv
e
highest
bri
ghtness
c
onser
va
ti
on
.
It
se
par
a
te
s
the
histogr
a
m
of
input
i
m
ages
i
nto
tw
o
de
pend
on
input
m
ean
be
fore
eq
ualiz
ing
them
in
par
al
le
l.
Th
reshold
le
vel
bas
e
d
par
ti
ti
on is
perform
ed
wh
ic
h would
g
i
ve
le
a
st am
ou
nt of a
bs
ol
ute m
ean br
ig
htn
ess
er
ror.
Con
tra
st
en
ha
nc
e
m
ent
of
m
a
m
m
og
ram
s
based
on
re
gion
is
desc
ribe
d
in
[
2].
Co
ntrast
enh
a
ncem
ent
of
m
a
m
m
og
rap
hic
feat
ur
es
of
cha
ng
i
ng
siz
e
and
sh
a
pe
,
ad
aptive
m
et
ho
d
is
us
ed.
T
he
re
gion
is
grown
by
this
m
et
ho
d
us
i
ng
e
ach p
ixel
i
n
th
e
i
m
age.
T
he
point
a
nd
s
hap
e
o
f
the r
egi
on get
us
e
d
to
local
i
m
age,
va
riat
ion
s
in
gr
ey
le
vel
.
By
app
ly
in
g
a
n
e
m
pirical
transf
or
m
,
con
t
rast
is
en
han
ce
d
bas
ed
on
e
ve
ry
re
gions
see
d
pixe
l
valu
e
,
it
s
back
gro
und
and
it
s
con
t
ra
st.
In
te
r
-
s
ub
ba
nd
c
orrelat
ion
based
im
age
reso
luti
on
en
ha
ncem
ent
in
wav
el
et
do
m
ai
n
is
disc
us
se
d
in
[3
]
.
I
nter
-
sub
band
correla
ti
on
bas
ed
res
ol
ution
enh
a
ncem
ent
m
et
ho
d
is
m
easur
e
d.
In
te
r
pola
ti
on
f
i
lt
ers
are u
se
d
t
o
a
naly
ze
the
c
orrelat
ion
s
bet
ween
sub b
a
nd
s
ha
ving d
ive
rs
e
sam
pling
pha
ses
in
the lo
wer l
evel
, and give
n
t
o
t
he
c
orrelat
ed
s
ub b
a
nds i
n
the
h
ig
he
r
le
vel.
Cy
cl
e
sp
inn
in
g
an
d
ed
ge
m
od
el
ing
bas
ed
wa
velet
dom
ai
n
i
m
age
reso
luti
on
e
nh
a
ncem
ent
is
exp
la
ine
d
in
[
4].
An
init
ia
l
high
-
res
olu
ti
on
est
i
m
a
te
to
t
he
or
i
gin
al
im
age
is
ob
ta
ine
d
by
m
eans
of
zero
-
paddin
g
i
n
the
wa
velet
dom
a
in.
Cy
cl
e
-
s
pinnin
g
m
et
ho
dolog
y
is
us
e
d
for
furthe
r
proces
sing
w
hich
dec
rease
s
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Image Res
olu
ti
on E
nhanceme
nt U
si
ng Tr
ans
form
(
Syed Na
zeeb
ur
re
hma
n)
355
rin
ging.
Linea
r
reg
res
sio
n
us
i
ng
a
le
ast
trai
nin
g
set
of
hi
gh
reso
l
ution
or
i
gi
nals
is
la
st
ly
e
ng
a
ge
d
to
recti
fy
the
degra
ded
e
dg
e
s.
Ad
a
ptive
co
ntrast
e
nhance
m
ent
of
m
edical
x
-
ray
i
m
ages
based
on
re
gi
on
is
presente
d
in
[5
]
.
Con
tra
st
en
ha
nc
e
m
ent
of
X
-
Ra
y
i
m
ages
and
pr
ese
nts
her
e
a
n
a
ppro
a
ch
f
or
co
ntrast
e
nh
a
nc
e
m
ent
base
d
upon
adap
ti
ve
neig
hbor
hood
te
chn
i
qu
e
[6
]
.
A
n
ou
tc
om
e
of
pe
rio
dized
sm
al
l
si
de
gam
es
wi
th
and
with
out
m
ental
i
m
ager
y
on
pla
yi
ng
abili
ty
a
m
on
g
intercoll
egiat
e
le
vel
so
ccer
play
ers
is
al
so
descr
i
bes
that
[6
]
.
Im
age
Su
pe
r
Re
so
luti
on
Usi
ng
W
a
velet
T
r
ansfo
rm
ation
Ba
sed
Gen
et
ic
Algo
rithm
exp
la
ined
in
[
7].
Local
Bi
na
ry
Patt
ern
s
(LBP)
an
d
D
WT
te
c
hn
i
qu
e
s
are
a
naly
zed
for
obj
ect
rec
ogniti
on
in
[
8].
LBP
is
us
e
d
to
ext
ract
the
detai
le
d
inf
or
m
at
ion
'
s
of
ob
j
ect
s
f
ro
m
it
s
m
ulti
-
scal
e
represe
ntati
on
.
Using
the
e
xt
racted
featur
e
s
,
the
rec
ogniti
on
of
obj
ect
s
can
be
done
by
the
cl
assifi
er
kn
own
a
s
the
near
e
st
n
ei
ghbour
cl
assifi
er.
S
o
D
WT
can
be
us
e
d
f
or
bo
t
h
enh
a
ncem
ent
and
cl
assifi
cat
ion
ty
pe.
Te
xt
Re
gio
n
Extra
ct
ion
in
a
D
oc
um
ent
I
m
age
Ba
sed
on
D
WT
is
exp
la
ine
d
in
[
9]
.
Bl
ind
Stegnogra
ph
y
in
Col
or
Im
ages
by
Dou
ble
W
avel
et
Tran
sf
orm
a
nd
Im
pr
ov
e
d
Arno
l
d
Transf
or
m
is
pr
esente
d
in
[10
]
.
VLS
I
A
rch
it
ect
ur
es
for
D
WT
based
on
li
fting
–
A
Surve
y
[11].
He
re
D
WT
is
al
so
use
d
i
n
V
LSI
dom
ai
n
wo
r
ks
.
Fr
e
quenc
y
Ba
nd
Sup
pr
e
ssion
an
d
T
hro
ughput
E
nhanc
e
m
ent
base
d
I
m
age
Com
pr
ession u
sing Disc
rete
Wav
el
et
T
ra
nsfo
rm
is d
isc
us
s
ed
in
[1
2
]
.
3.
THE
PROB
LE
M
The
basic
lim
i
ta
ti
on
s
inclu
din
g
a
re
that
it
cannot
sim
ultan
eo
us
ly
en
hance
al
l
par
ts
of
i
m
age
ver
y
well
an
d
it
is
a
lso
diff
ic
ult
to
autom
at
e
the
im
age
en
ha
nce
m
ent
proce
dur
e.
Work
i
ng
of
tradit
ion
al
m
eth
ods
of
i
m
age
enh
a
nce
m
ent
is
to
en
ha
nce
the
a
ppe
aran
ce
of
lo
w
qu
al
it
y
i
m
age
[6]
.
It
does
not
co
ntain
a
ny
high
qu
al
it
y
bac
kgr
ound
in
form
at
i
on
beca
us
e
i
n
t
he
dark
im
age
so
m
e
areas
are
so
da
rk
t
hat
al
l
the
inf
orm
ati
on
is
al
read
y l
os
t i
n t
ho
se
r
e
gions.
4.
PROP
OSE
D SOLUTI
ON
Lo
w
-
pass
filt
ered
sig
nal
ha
vin
g
s
om
e
hig
h
fr
e
qu
e
ncy
inf
orm
ation
beca
use
the
analy
sis
filt
er
ban
k
has finit
e filt
er
ta
ps
a
nd
als
o
s
om
e low
freq
ue
ncy in
form
ati
on is obtai
ne
d from
h
igh
pass
f
il
te
red
sig
nal.
Sam
e
ph
a
se
has
dow
n
sam
pling
t
he
both
high
-
pas
s
an
d
l
ow
-
pa
ss
filt
ered
sig
nals
but
sti
ll
rem
a
in
s
s
om
e
cor
re
la
ti
on
though,
t
her
e
will
be
co
rr
el
a
ti
on
at
low
w
hi
le
do
w
n
sam
pling
by
var
i
ous
phases.
T
he
r
e
are
thre
e
ste
ps
ar
e
involve
d
in
pr
opose
d schem
e:
f
il
te
r
est
i
m
at
io
n,
ba
nd esti
m
a
t
ion
,
and
rec
on
s
tructi
on.
4.1.
Fil
ter Estim
ati
on
More
tha
n
co
m
ple
te
wav
el
et
transfor
m
on
the
obta
ined
L
L0
0
is
app
li
ed
first
to
get
al
l
var
i
ou
s
ph
ase
cases
of
eve
ry
su
b
ba
nd.
Four
filt
ers
are
de
sign
e
d
in
this
ste
p:
LL1_
01,
LL
1_10,
H
L1
_
00
a
nd
L
H1_
00
.
These
are
a
ntici
pated
res
pec
ti
vely
fr
om
L
L1
_
00.Li
near
le
ast
-
sq
ua
res
reg
re
ssio
n
ba
sed
filt
er
eval
uatio
n
appr
oach al
so
util
iz
ed.
4.2.
Band
Es
tima
ti
on
Fil
te
rs
antic
ipa
te
d
in
this
sect
ion
a
re
use
d
i
n
the
lo
wer
le
ve
l
to
evaluate
t
he
associat
ed
ba
nd
s
in
the
higher
le
vel.
10
LL, 01
LL, 00
HL
a
nd
00
L
H
of
hi
gh
le
vel
from
LL00
is
ob
ta
ine
d
by
us
i
n
g
filt
ers
f
Ah,
f
A
v,
f
Bh
and
f
B
v
a
ntici
pated
i
n
th
e low
e
r
le
vel.
5.
RECONST
R
UC
TI
ON
Inp
ut
im
age
re
so
luti
on
is
en
ha
nced
by
rec
onstr
ucting
the
or
i
gin
al
im
age
from
the
perf
or
m
ance
of
inv
e
rse
wav
el
e
t
transfo
rm
wit
h
LL
00
a
nd
t
he
pr
e
dicta
ble
10
HL
a
nd
01
LH.
T
he
dia
gonal
ba
nd
of
H
H11
is
su
pp
os
e
d
to
be
zero
,
because
HH1
1
no
t
only
belo
ng
s
to
di
ver
se
s
ub
band
f
r
om
00
LL
bu
t
al
s
o
has
div
ers
e
sam
pling
phas
es.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vol
.
9
,
No.
2
,
Fe
br
uary
201
8
:
35
4
–
35
6
356
(a)
(
b)
(
c)
Figure
1. Gi
ve
n Inp
ut I
m
age: (a)
Ba
boon,
(b) Lena
, a
nd (
c
) M
andril
l
6.
RESU
LT
S
AND DI
SCUS
S
ION
S
Nu
m
ber
of w
el
l
-
know
n
te
st ima
ges
includi
ng Len
a
, Bab
oon
an
d
m
and
rill
are
ex
per
im
ented.
Used all
the
i
m
ages
in
the
database
e
xcep
t
one
im
a
ge
f
or
e
valuati
on
at
the
trai
ni
ng
phase
for
li
near
re
gr
essi
on
i
s
rep
eat
e
d.
PS
N
R
value
s
f
or
e
r
ror
betwee
n
t
he
one
ha
nd
gr
ound
t
ru
t
h
im
ages
an
d
rec
on
st
r
uction
is
m
easur
e
d
.
Ed
ge
directed
i
nter
po
la
ti
on b
a
sed
non
-
wa
vel
et
schem
e
was
al
so
c
on
si
der
e
d
to
gi
ve
a
c
om
par
ison
w
it
h
a
sta
te
-
of
-
the
-
art m
et
ho
d.
Our res
ults sh
ow that t
he pr
opos
e
d
te
c
hniqu
e
pro
vid
es
re
li
able i
m
pr
ov
e
m
ents.
Table
1.
PS
NR
Value
f
or
P
ropo
s
ed
G
i
ven I
m
age
7.
CONCL
US
I
O
NS
Wav
el
et
do
m
ai
n
base
d
im
age
reso
l
utio
n
e
nh
a
ncem
ent
al
gorithm
was
presente
d.
Ze
r
o
paddi
ng
of
high
fr
e
quency
wav
el
et
sub
ba
nds
is
the
i
m
p
or
ta
nt
el
e
m
ent
s
of
this
te
ch
niq
ue
.
Ri
ng
i
ng
a
rising
a
re
redu
ced
by
cy
cl
e
sp
inn
i
ng
from
finall
y
e
dg
e
recti
ficat
i
on
an
d
zer
o
-
pa
ddin
g
to
im
pr
ov
e
blurrin
g
du
e
to
the
una
vaila
bili
ty
of
high
sp
at
ia
l
fr
eq
uen
cy
in
form
ation
.
O
ur
res
ults
hav
e
sh
own
t
hat
the
pro
posed
m
et
ho
d
outp
er
form
s
conser
vative i
m
age in
te
r
po
la
ti
on
a
ppr
oach
e
s.
REFERE
NCE
S
[1]
Chen
S
.
D
.
and
Ramli
A
.
R.
,
“
Minim
um
m
ea
n
bright
ness
err
or
bi
-
histogra
m
equal
iz
a
ti
on
in
con
tras
t
enha
n
ce
m
ent,
”
IEE
E
transacti
o
ns on
Consum
er
El
e
ct
ronics
,
vol
/
issue:
49(4)
,
pp.
1310
-
1319
,
200
3
.
[2]
Morrow
W
.
M
.
,
et
al
.
,
“
Regi
on
-
base
d
cont
rast
enha
nc
ement
of
m
am
m
ogra
ms,
”
IEE
E
transact
ions
o
n
Me
dic
a
l
Imaging
,
vol
/i
ss
ue:
11(3)
,
pp
.
39
2
-
406
,
1992
.
[3]
P
ia
o
Y
.
and
Park
H.
,
“
Im
age
resolut
ion
enha
n
c
ement
using
inter
-
subband
cor
r
el
a
ti
on
in
wave
l
et
dom
ai
n,
”
IEEE
Inte
rnational
Co
nfe
renc
e
on
Image
Proc
essing
,
v
ol.
1
,
pp
.
I
-
445
,
2007
.
[4]
T
emizel
A
.
and
Vlac
hos
T.
,
“
W
ave
le
t
dom
ai
n
image
resolut
i
on
enha
nc
ement
using
c
y
c
le
sp
inni
ng
and
edg
e
m
odel
ing,
”
IE
E
E
13
th
Europ
ean Signal
Proce
ss
in
g
Confe
ren
ce
,
p
p.
1
-
4
,
2005
.
[5]
Kanwal
N.
,
“
Regi
on
base
d
ad
apt
iv
e
cont
rast
enha
nc
ement
of
m
edi
ca
l
x
-
ra
y
images,
”
IEE
E
5
th
Inte
r
nati
ona
l
Confe
renc
e
on
B
ioi
nformatic
s
an
d
Bi
omed
ic
al
En
gine
ering
,
pp
.
1
-
5
,
2011
.
[6]
Kum
ar
A
.
M.
,
“
An
outc
om
e
of
per
iodized
sm
al
l
side
games
wit
h
and
without
m
ent
al
image
r
y
on
play
ing
abi
l
i
t
y
among i
nte
r
col
l
e
gia
t
e
l
evel
socc
e
r
play
e
rs
,”
India
n
Journal
o
f
S
ci
e
nce
and
Techno
l
ogy
,
vo
l/
issue:
8
(36)
,
2015
.
[7]
Panda
S
.
S
.
and
Jena
G.
,
“
Im
age
Super
Resolution
Us
ing
W
av
el
e
t
Tra
nsform
ation
Based
Gene
ti
c
Algori
thm
,”
In
Computati
ona
l
Int
el
l
ige
nc
e in D
ata
Mini
ng
,
vo
l.
2
,
pp
.
355
-
361
,
2016
.
[8]
Je
y
asudh
a
A
.
an
d
Pri
y
a
K.
,
“
Ob
je
c
t
Rec
ogni
ti
on
Based
o
n
Lbp
a
nd
Discre
t
e
W
ave
l
et
Tr
ansform
,”
Inte
rnation
al
Journal
of
Ad
va
nce
s in
Signal a
nd
Image
S
ci
en
c
es
,
vol
/i
ss
ue:
2(1
)
,
pp
.
24
-
30
,
201
6
.
[9]
Xiao
Y
.
and
Ya
n
H.
,
“
Te
xt
reg
i
on
ext
ra
ct
ion
in
a
document
image
bas
ed
on
th
e
Dela
un
a
y
te
ss
el
l
at
ion
,”
Pa
tt
er
n
Re
cogn
it
ion
,
vo
l
/i
ss
ue:
36(3)
,
pp
.
799
-
809
,
2003
.
[10]
PourA
ria
n
M
.
R
.
and
Hana
ni
A.
,
“
Bli
nd
Stega
nogra
ph
y
in
C
olor
Im
age
s
b
y
Double
W
ave
le
t
Tra
nsfor
m
and
Im
prove
d
Arnold
Tra
nsform
,”
In
donesian
Journal
of
Elec
tri
cal
E
ngine
ering
and
Comput
er
Sci
en
ce
,
vo
l/
issue:
3(3
)
,
pp.
586
-
600
,
20
16
.
[11]
Kala
i
ara
si
M
.
an
d
Vigneswara
n
T.
,
“
VLSI
Archi
te
c
ture
s
for
Li
ft
i
ng
Based
Disc
re
te
W
avele
t
Tra
n
sform
–
A
Survey
,”
Indone
sian J
our
nal
of
Elec
tric
al
Engi
ne
ering
and
Computer
Sc
ie
n
ce
,
vol
/i
ss
ue:
3(2
)
,
pp
.
323
-
330
,
2
016
.
[12]
L
.
Nara
y
a
nan
,
et
al.
,
“
Discre
t
e
W
ave
le
t
Tra
nsfo
rm
Based
Im
age
Com
pre
ss
ion
us
ing
Freque
nc
y
B
and
Suppress
ion
and
Throughpu
t
Enha
nc
ement
,”
I
nte
rnational
Jou
rnal
of
MC
Squ
are
Scientific
Res
earc
h
,
vol
/i
ss
ue
:
9(2)
,
pp.
176
-
182
,
2017
.
Para
m
eter
Bab
o
o
n
Lena
Mand
rill
PSNR
2
8
.94
3
1
.46
2
4
.58
Evaluation Warning : The document was created with Spire.PDF for Python.