Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
8
, No
.
6
,
Decem
ber
201
8
, p
p.
4230
~
4238
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v8
i
6
.
pp
4230
-
42
38
4230
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Geometr
ic a
nd
G
raysc
ale Templat
e Matchi
ng for S
audi
Arabian
Riyal P
ap
er
Cu
rren
cy Re
cognitio
n
Suci A
uli
a
1
, B
agus Bu
dhi L
2
, An
gga Rus
d
inar
3
, Yu
yu
n
Siti R
4
1,4
Depa
rtment
of
Applie
d
Sci
ence
,
Telkom
Unive
rsit
y
,
Indon
esia
2,3
Depa
rtment
of
Elec
tr
ical Engi
n
ee
ring
,
Te
lkom
Univer
sit
y
,
Indo
nesia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
N
ov
28
, 201
7
Re
vised
Ju
l
9
,
201
8
Accepte
d
Aug
3
, 2
01
8
Dete
c
ti
ng
the
au
the
nticity
of
pap
er
cur
ren
c
ie
s
using
aut
om
at
ed
ba
sed
Paper
Curre
nc
y
Re
cog
nit
ion
(PCR
)
wi
th
image
pro
ce
s
sing
te
chn
ique
s
was
stil
l
a
hot
topi
c
of
d
isc
uss
ion,
due
to
th
e
ci
r
cul
a
ti
on
of
c
ounte
rfe
it
cur
r
en
c
y
tha
t
was
stil
l
over
whelmi
ng
in
som
e
c
oun
tri
es.
Th
ere
was
a
downs
ide
al
on
g
with
thi
s
adva
nc
ement
in
te
chno
log
y
in
t
he
field
of
co
lo
r
print
ing
,
duplicat
ion
,
an
d
sca
nning,
be
ca
u
se
it
was
bec
a
m
e
one
of
the
supporting
factors
of
the
inc
re
asing
cri
m
e
rat
e
in
produc
t
ion
of
count
erf
e
it
m
one
y
.
Our
sy
s
te
m
has
per
form
ed
a
PC
R
appr
oac
h
bas
ed
on
image
pr
oce
ss
ing
techniq
ues
.
In
th
is
stud
y
,
the
SA
R
banknot
e
was
th
e
object
to
be
r
ec
ogni
ze
d
and
d
et
e
ct
ed
it
s
aut
henticit
y
wit
h
the
dev
el
op
m
ent
of
the
pr
evi
ous
m
et
hod,
which
was
inc
orpora
ti
ng
th
e
Geom
et
ric
Tem
pla
te
Matc
hin
g
and
Gra
y
sca
l
e
Te
m
pla
t
e
Matc
hing
.
In
ad
dit
ion
to
th
e
p
a
tt
ern
rec
ogn
it
ion
proc
ess,
the
class
ifi
ca
t
ion
proc
ess
on
1
S
A
R,
2
SA
R,
5
SAR,
and
10
SA
R
was
al
so
per
form
ed.
From
PC
R
te
st
up
to
1
00
sam
ple
da
ta
,
for
each
t
este
d
b
anknot
e
val
u
e
o
bta
ine
d
th
e
av
era
g
e
va
lue
of
the
be
st
ac
cur
acy
le
vel
from
i
ncor
pora
t
ing
GeoMat
chi
ngSc
ore
and
Gra
y
M
at
chi
ngScor
e
fo
r
the
cl
assificat
ion
proc
ess
was
95.
25%.
W
hil
e
the
ave
rag
e
le
vel
of
s
y
st
e
m
ac
cur
acy
in
rec
ogni
zi
ng
count
erf
ei
t
m
oney
on
e
ac
h
bankn
ote
ob
ta
in
ed a
m
axi
m
um
val
u
e
o
f
100%.
Ke
yw
or
d:
Ba
nknote
Counter
feit
Geo
m
et
ric
t
e
m
plate
m
a
tc
hin
g
Gr
ay
scal
e
t
em
plate
m
a
tc
hin
g
PCR
SA
R
Copyright
©
201
8
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Su
ci
Au
li
a,
Dep
a
rtm
ent o
f Appli
ed
Scie
nc
e
,
Tel
ko
m
U
ni
versi
ty
,
Tel
eko
m
un
ika
si R
oad,
Dayeu
h Ko
l
ot,
Ba
ndung, Ja
wa
Ba
rat 4025
7,
I
ndone
sia
.
Em
a
il
:
su
ci
a@ta
ss.tel
ko
m
un
i
ver
sit
y.ac
.id
1.
INTROD
U
CTION
Counter
feit
ing
has
beco
m
e
t
he
case
in
al
m
os
t
eve
ry
co
un
try
,
thu
s
m
otiv
at
ing
rese
arc
he
rs
to
detect
counter
feit
curre
ncy
base
d
on
i
m
age
pr
oc
e
ssing
[
1]
-
[4
]
.
Fo
r
e
xam
ple,
especial
ly
in
In
dia
co
unte
rf
ei
ti
ng
has
beco
m
e
a
crit
ic
al
issue.
T
herefo
re,
m
any
re
searche
rs
f
ocus
on
detect
in
g
the
a
uth
e
ntici
ty
of
100,
500,
a
nd
1000
r
upees
ba
sed
on
Ne
ur
al
Netw
ork
[
5]
-
[
1
0
]
a
nd
oth
e
r
i
m
age
proce
ssing
ba
sed
m
et
ho
ds
m
or
ph
olog
ic
al
ly
[1
1]
-
[
1
8
]
.
I
n
a
dd
it
io
n
to
t
he
I
nd
ia
n
Ru
pee,
ot
her
c
urren
ci
es
are
w
i
dely
stu
died,
inclu
ding
Eu
ro
[
19]
-
[
2
0
]
,
U
S
Do
ll
ar
[2
1]
-
[
2
3
]
,
Sa
ud
i
Ar
a
bi
an
Ri
ya
l
[2
2
]
,
Ind
on
esi
a
n
R
up
ia
h
[
2
4
]
an
d
Japa
n
Ye
n
[2
5
].
On
e
stu
dy
of
t
he
currency
rec
ogniti
on
base
d o
n im
age pro
ces
s
ing
was
a
stu
dy
that has
bee
n d
on
e
b
y
Sawa
nt [2
6
]
.
In this
stud
y,
the
c
urren
cy
use
d
was
an
I
ndia
n
cu
rr
e
ncy
with
a
n
acc
ura
cy
le
vel
ap
pro
achin
g
90%
ba
sed
on
4
par
a
m
et
ers,
nam
ely
Do
mi
nant C
olo
r, Asp
ect
Ra
ti
o,
M
ark
I
D
a
nd
L
aten
t Imag
e
. S
im
i
l
ar r
esea
rch
wa
s sti
ll
ab
ou
t t
he
I
ndia
n
currency
recogn
it
io
n,
wh
ic
h
us
e
d
the
D
WT
al
gorithm
fo
r
fe
atu
re
ext
r
action
a
nd
cl
a
ssific
at
ion
bas
ed
on
Probabil
ist
ic
Neur
al Net
work
(P
N
N
)
[
2
7
].
The
res
ult
obta
ined
by
appr
oach
i
ng
both
m
et
ho
ds
was
qu
it
e
go
od
tha
t
was
with
the
accuracy
of
90.38%
.
I
n
a
dd
it
io
n
to
the
Indian
c
urre
ncy
rec
og
niti
on,
wh
ic
h
has
done
a
lot
because
of
it
s
high
counter
feit
ing
crim
e
rates,
cur
re
ncy
rec
ogniti
on
on
doll
ars
al
so
wi
dely
popu
la
r
am
on
g
re
searc
he
rs
as
stud
i
e
d
by
F.
Take
da
[
2
5
]
.
I
n
his
j
ou
rn
al
,
it
was
pr
opos
e
d
a
ne
w
te
chn
iq
ue
to
c
onduct
a
pap
e
r
cur
re
ncy
rec
ogniti
on
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Geometri
c
and Gr
aysc
ale Te
mp
l
ate M
atchi
ng for
Sa
ud
i A
ra
bi
an Riyal
P
ap
e
r Cur
rency
...
(
Su
ci
A
ulia)
4231
and
cl
assi
ficat
ion
on
Ja
pan
Y
en
an
d
US
D
ollar,
w
hich
was
us
in
g
neural
ne
tworks
.
T
he
NN
structure
c
ouple
d
with
the
r
ando
m
m
as
k
m
et
hod
sho
ws
it
s
effe
ct
iveness
for
processi
ng
dat
a
viewe
d
f
ro
m
tim
e
and
fr
e
quency
do
m
ai
n.
I
n
a
ddit
ion
t
o
India
n
R
up
ee
an
d
US
doll
ar,
P
aper
C
ur
re
ncy
Reco
gn
it
io
n
(PC
R)
on
S
au
di
Ar
a
bian
Ri
ya
l
based
on
the
correla
ti
on
betwee
n
i
m
ages
has
al
so
been
done.
T
he
m
et
ho
d
us
e
d
fo
r
the
cl
assif
ic
at
ion
was
Ra
dia
l
B
asi
s
Fun
ct
io
n
Ne
tw
or
k
and
the
aver
a
ge
accu
r
acy
le
vel
ob
ta
ined
was
quit
e
sat
is
factor
y
tha
t
was
91.51%
[2
8
]
.
Othe
r
resea
rc
h
this
pa
per
refe
rr
e
d
to
was
P
CR
and
cl
assif
ic
at
ion
on
five
curren
ci
es
at
on
ce
,
includi
ng
US
Do
ll
ar
(
USD
),
Au
st
rali
an
D
ollar
(AUD),
Sa
ud
i
A
ra
bian
Ri
ya
l
(S
AR),
E
uro
(E
UR),
a
nd
I
nd
ia
n
Rup
ee
(
IN
R
)
[
2
2
]
.
I
n
t
his
re
sea
rch,
patte
rn
rec
ogniti
on
in
Regi
on
of
I
nterest
(RO
I)
usi
ng
ne
ural
net
work
,
wh
il
e
t
he
proc
ess
of
cl
assi
ficat
ion
us
in
g
te
mp
l
ate
m
atc
hin
g
m
et
ho
d.
Th
e
syst
e
m
has
s
uccess
fu
ll
y
rec
ogniz
e
d
5
I
NR,
20
IN
R
, 20
Eu
r
o,
50 E
uro,
20
A
UD
a
nd
50 AUD
ba
nknote
s.
Patt
er
n
rec
o
gnit
io
n
on 1
S
AR a
nd
5 SAR
bank
no
te
s
are st
il
l i
n
process,
whil
e p
at
te
r
n r
ecognit
ion
on
USD
bank
no
te
s w
as
f
ai
le
d.
Ba
sed
on
s
ome
of
the
stu
dies
prese
nted
a
bove
,
PCR
is
sti
ll
a
chall
en
ging
to
pic
to
stu
dy
.
The
refor
e
,
in
this
resea
rc
h
we
will
stud
y
the
cu
rr
e
nc
y
recogn
it
io
n
te
chn
iq
ue
a
nd
it
s
cl
assifi
cat
ion
us
i
ng
te
m
pla
te
m
at
ching
m
et
ho
d,
becau
se
b
a
sed on resea
rc
h
[
2
2
]
it
h
as s
uc
cessf
ully
r
eco
gn
iz
e
d
fi
ve
c
urren
ci
es
nam
el
y
USD,
AUD,
I
NR,
E
UR
and
S
AR.
In
the
pre
vious
stud
y
[2
8
]
,
PC
R
and
cl
assifi
cat
ion
of
USD
ba
sed
on
Ca
nny
Edge
Detect
ion
a
nd
Tem
plate
Mat
chin
g
ob
ta
ine
d
an
a
ver
a
ge
ac
cur
acy
le
vel
o
f
95.
625%
.
I
n
this
stud
y,
the
SA
R
bank
no
te
is
th
e
obj
ect
t
o
be
recog
nized
a
nd
detect
ed
it
s
authe
ntici
ty
with
the
de
velo
pm
ent
of
the
previo
us
m
et
ho
d, w
hich
is inc
orporati
ng the
G
e
om
et
ri
c Tem
plate
Ma
tc
hin
g an
d G
ra
ysc
al
e Tem
plate
Ma
tc
hing.
2.
RESEA
R
CH MET
HO
D
In
this stu
dy, th
e res
ults of
edge d
et
ect
ion
usi
ng
can
ny edge
d
et
ect
ion
obtai
ned
sati
sfacto
r
y resu
lt
s as
reco
m
m
end
ed
by
[
29]
-
[
3
0
]
it
see
m
s
to
pro
du
ce
false
detect
ion
in
no
isy
en
vir
on
m
ent.
Ba
sed
on
oth
e
r
ref
e
ren
ces
[3
1
]
,
can
ny
ed
ge
de
te
ct
ion
in
pre
-
processi
ng
give
s
a
fairly
good
cl
assifi
cat
ion
resu
lt
.T
he
ne
xt
ste
p
f
or
i
m
age
s
m
oo
the
ning
proce
ss
in
this
rese
arch
is
done
w
it
h
two
a
ppr
oach
e
s,
w
hich
are
m
edian
filt
er
an
d
g
au
ssian
filt
er.
Me
dian
filt
ering
is
a
non
-
li
ne
ar
di
gital
filt
erin
g
te
ch
nique
w
hich
is
of
te
n
us
ed
to
redu
ce
an
d
even
el
im
inate
no
ise
[
29
]
.
Me
dian
filt
erin
g
is
of
te
n
us
e
d
in d
igit
al
i
m
age
pr
oces
sin
g
beca
us
e
of
it
s
su
pe
r
iority
on
m
ai
ntaining
an
e
dg
e
val
ue
durin
g
no
ise
r
e
m
ov
al
pr
oc
ess [
3
2
]
.
Fi
gure 1
sh
ows
a
n
il
lust
rati
on
of
the
m
edian
filt
ering
.
Figure
1
.
I
nput
and
ou
t
pu
t i
ll
ust
rati
on
of m
edian
filt
er
with
out cha
ngin
g
t
he
edge
pix
el
va
lue
Ba
sed
on
Fi
gure
1
ab
ove
,
m
edian
value
re
placed
t
he
cente
r
of
m
edian
filt
er
m
a
trix
based
on
equ
at
io
n (
1)
[3
3]
-
[
3
5
]
:
[
,
]
=
{
,
[
,
]
,
(
,
)
}
(1)
Wh
e
re
repres
ents
a
neig
hbour
hood
ce
ntere
d
on
l
ocati
on.
Anothe
r
a
ppr
oa
ch
was
g
auss
ian
fi
lt
er,
it
w
as
a
sm
oo
thing
te
c
hn
i
qu
e
on
ed
ge
detect
ion
process
[2
4
]
.
T
he
eq
uatio
n
f
or
e
dg
e
m
at
ching
al
gorit
hm
of
the
i
m
age
[3
6
]
is s
how
n on eq
uat
ion
(
2
)
:
(
,
)
=
−
1
4
[
1
−
2
+
2
2
2
]
−
2
+
2
2
2
(2)
Wh
e
re
(
,
)
represe
nts
the
posit
ion
of
eac
h
pix
el
s
of
t
he
im
age
and
σ
wa
s
the
S
ta
nd
a
rd
Ga
us
si
an
De
viati
on.
In
im
age
pr
oc
essing,
Ga
us
si
an
filt
er
us
ed
was
a
two
di
m
ension
al
Ga
us
sia
n
filt
er.
Ther
e
f
or
e,
eac
h
pix
el
directi
on
has o
ne dim
ension
al
G
a
us
sia
n eq
ua
ti
on
as
foll
ow
s:
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4231
-
4238
4232
(
,
)
=
1
√
2
2
−
2
2
2
(3)
Wh
e
re
σ
re
pr
e
sented
the
sta
ndar
d
de
viati
on
of
val
ue
distri
bu
ti
on.
T
he
di
stribu
ti
on
wa
s
assum
ed
to
hav
e
a
n
aver
a
ge value
equ
al
t
o 0. The
il
lustrati
on of
Gaussi
an
distri
bu
ti
on is
sho
w
n on Fig
ure
2
a
nd Fig
ure
3.
Figu
r
e
2
.
V
al
ue
d
ist
rib
utio
n g
raph
of G
a
us
si
an 1D
Figure
3
.
V
al
ue
d
ist
rib
utio
n g
raph
of G
a
us
si
an 2D
The
seco
nd
ste
p
in
this
resear
ch
after
the
sm
oo
t
hing
proces
s
was
par
am
eter
op
ti
m
iz
ation.
Te
m
plate
Ma
tc
hin
g
was
a
m
at
ching
te
chn
i
qu
e
us
e
d
in
the
stud
y,
te
m
plate
m
at
ching
was
oft
en
acc
ur
at
e
by
c
om
bin
in
g
geo
m
et
ric
score
and
gray
scal
e
scor
e
pa
ram
et
ers.
G
hazi
[
3
7
]
has
te
ste
d
the
accu
racy
le
vel
of
Ge
ome
tric
Tem
plate
Ma
tch
in
g
(G
e
TeM
)
to
de
te
ct
the Di
nar
c
urre
ncy
f
or
10
0
ti
m
es
a
nd
it
has
the
ac
cur
acy
le
vel o
f
91%
.
GeTeM
w
orks
by
c
om
par
ing
this
x
value:
=
〈
〉
=
1
with
this
x
’
value:
′
=
〈
′
〉
=
1
in
ti
m
e
dom
ai
n
[
38]
.
In
h
is pa
per
[3
8
]
,
F
rank
m
entions
G
eTeM
ne
eds
t
o
be
c
on
s
idere
d
a pow
e
r
fu
l
a
ddit
ion
t
o
the
s
uite
of
to
ol
s
that
a
tim
e
series
analy
st
has
at
th
ei
r
dis
po
sal
for
the
ne
xt
fu
t
ure
w
ork,
s
o
t
hat
in
this
st
ud
y
t
rial
s
wer
e
com
bin
e
d
betwee
n
GeTe
M
and
G
raysc
al
e
Tem
plate
Ma
tc
hin
g
(
Gr
a
yT
eM
)
f
or
feat
ur
e
e
xtr
act
io
n
process
.
T
he
e
qu
at
io
n
of the m
at
ch
sc
or
e
it
sel
f
is s
hown b
y t
he follo
wing e
qu
at
io
n (4)
[3
6
]:
ℎ
=
ℎ
1000
(4)
-4
-3
-2
-1
0
1
2
3
4
5
x
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
Gaus
si
an
1D
0
3
0
.
0
5
2
3
0
.
1
1
(
x
,
y
)
2
2
D
G
a
u
s
s
i
a
n
y
0
.
1
5
0
1
x
0
0
.
2
-1
-1
-2
-2
-3
-3
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Geometri
c
and Gr
aysc
ale Te
mp
l
ate M
atchi
ng for
Sa
ud
i A
ra
bi
an Riyal
P
ap
e
r Cur
rency
...
(
Su
ci
A
ulia)
4233
3.
RESU
LT
S
A
ND AN
ALYSIS
The
syst
em
con
sist
ed
of
3
m
ai
n
pa
rts,
nam
el
y
1)
m
echan
ic
al
syst
e
m
of
t
he
sca
nn
e
r
box
the
siz
e
of
18
cm
x10cm
x
15
cm
to
ta
ke
t
he
im
age
of
th
e
ba
nknote
wit
h
a
li
ghti
ng
from
4
le
d
piece
s
an
d
180
de
grees
of
li
gh
t
distrib
ution,
2)
we
b
ca
m
era
and
PC
that
serv
es
a
s
the
i
m
age
pr
oces
sin
g
syst
e
m
,
an
d
3)
ac
tuator
con
t
ro
ll
ers
w
hi
ch
co
ntro
ls
by
assigni
ng
se
rial
values
(ser
ia
l
ou
t
of
PC)
us
ing
ar
duin
o
U
N
O
an
d
DC
m
oto
r
to
pu
ll
the
ba
nkno
te
a
fter
being
scan
ne
d
w
it
h
a
we
bcam
.
The
th
ree
sy
stem
s
are
connecte
d
se
rial
ly
to
th
e
com
pu
te
r
as
a
com
m
un
ic
at
i
on
ce
nter
,
ei
ther
as
a
syst
em
’s
database
serv
e
r,
or
as
an
e
xisti
ng
ac
tuato
r
con
t
ro
ll
er.
3.1.
Sy
s
tem Tes
tin
g
Sce
na
ri
o
To
f
ound
ou
t
the
best
pa
ram
e
te
rs
that
was
use
d
to
te
st
the
authe
ntici
ty
of
the
bank
no
te
(
PCR
and
it
s
cl
assifi
cat
ion
)
,
the
first
ste
p
was
to
m
at
ch
the
filt
er
m
edian
va
riable
X
siz
e
(1,
5,
10,
15,
a
nd
20)
a
nd
Y
siz
e
(1,
5,
10,
15,
a
nd
20)
each
co
m
bin
at
ion
an
d
Gaussi
an
fi
lt
er
with
div
i
der
pa
ram
et
ers
(d
5,
d10,
an
d
d15)
on
a
nu
m
ber
of
trai
ning
set
of
te
m
pla
te
i
m
ages.
Af
te
r
te
sti
ng
on
t
he
trai
ning
set
data,
then
the
outp
ut
was
in
the
form
of
Geom
et
ric
Matchin
g
Sco
re
(
Geo
M
S)
an
d
Grays
cale
Matc
hing
Sco
re
(
Gr
ay
MS).
T
he
val
ues
of
Geo
MS
an
d
G
rayM
S
are
a
na
ly
zed
an
d
use
d
as
t
he
th
res
ho
l
d
value
for
each
bank
no
t
e
1
S
AR,
5
S
AR,
10
SA
R,
an
d
50
S
AR.
From
the
resu
lt
s
of
the
tr
ai
nin
g
set
data
te
st,
ta
ble
X
a
nd
ta
ble
Y
eac
h
sh
ows
Ge
oMS
an
d
Gr
ay
MS data
f
or
PCR 1 SAR
. Th
e
data wer
e taken
ba
sed on
t
he
lowe
st scor
e
for
eve
ry 1
0
at
te
m
pts.
B
ased
on
Table
1,
t
he
th
reshold
val
ue
of
Ge
oMS
PC
R
1
SA
R
ta
ke
n
from
the
lo
we
st
value
on
t
he
top
ed
ge
col
um
n
was
901,4
88.
Sim
ilarly
,
the
t
hr
es
hold
valu
e
of
G
rayM
S
PCR
1
SA
R
base
d
on
T
able
2
w
as
858,7
8
2.
Furthe
rm
or
e,
after
Ge
oMS
and
Gr
ay
MS
wer
e
ob
ta
i
ned
for
each
SAR
PCR
,
a
te
st
for
the
cl
assi
ficat
ion
proces
s
was
cond
ucted.
Table
1
.
G
e
om
et
ry
Ma
tc
hin
g
Sc
ore
of
PC
R 1
S
AR
Nu
m
b
e
r
o
f
Test
Min Valu
e
Max
Valu
e
Top
E
d
g
e
10
8
2
3
.837
9
0
1
.988
9
0
1
.488
20
9
0
2
.184
9
0
8
.172
9
0
7
.672
30
9
0
9
.341
9
1
3
.821
9
1
3
.321
40
9
1
3
.889
9
1
9
.059
9
1
8
.559
50
9
1
9
.536
9
2
1
.716
9
2
1
.216
60
9
2
2
.004
9
2
5
.7
9
2
5
.2
70
9
2
6
.04
9
3
0
.747
9
3
0
.247
80
9
3
1
.152
9
3
4
.558
9
3
4
.058
90
9
3
5
.195
9
4
2
.113
9
4
1
.613
100
9
4
2
.192
9
5
9
.826
9
5
9
.326
Table
2
.
Gray
scal
e
Ma
tc
hing
Score
of
PCR
1
S
AR
Nu
m
b
e
r
o
f
Test
Min Valu
e
Max Valu
e
Top
E
d
g
e
10
8
1
9
.712
8
5
9
.282
8
5
8
.782
20
8
5
9
.339
8
5
9
.833
8
5
9
.333
30
8
5
9
.874
8
6
0
.377
8
5
9
.877
40
8
6
0
.401
8
6
0
.798
8
6
0
.298
50
8
6
0
.803
8
6
1
.044
8
6
0
.544
60
8
6
1
.076
8
6
1
.429
8
6
0
.929
70
8
6
1
.456
8
6
1
.883
8
6
1
.383
80
8
6
1
.965
8
6
2
.317
8
6
1
.817
90
8
6
2
.332
8
6
2
.822
8
6
2
.322
100
8
6
2
.848
8
6
4
.557
8
6
4
.057
3.2.
Experim
en
ta
l
Results
The
process
of
i
m
age
acqu
isi
ti
on
wa
s
done
by
captu
rin
g
t
he
ba
nknote
c
om
ing
into
the
scan
ner
box
with
th
e
li
ghti
ng
co
ndit
ion
s
of
4
LE
Ds
a
rrang
e
d
i
nto
2
x
2
LE
Ds.
Fro
m
the
capt
ur
e
resu
lt
,
t
he
syst
e
m
then
detect
ed
the
va
lue
of
the
num
ber
with
the
la
rg
est
f
ont
s
iz
e
and
t
hen
c
rop
ped
it
as
s
een
in
Fi
gure
4
an
d
Figure
5.
Fig
ure
4
re
presents
the
Geo
m
et
ric
Tem
plate
M
at
ching
f
or
ea
ch
1
S
AR,
5
S
AR,
10
S
AR,
and
50
SA
R
bank
note
.
A
s
f
or
G
raysca
le
Tem
plate
Ma
tc
hin
g, it
is
represe
nted
i
n Fi
gure
5.
(a)
(b)
(c)
(d)
Figure
4. Im
age te
m
plate
g
eo
m
et
ric
m
at
ching
(a)
1 SAR
, (b
) 5 SAR,
(c) 1
0 SAR,
(d) 5
0 S
AR
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4231
-
4238
4234
(a)
(b)
(c)
(d)
Figure
5.
I
m
age te
m
plate
g
ra
ysc
al
e
m
at
ching
.
(a) 1
S
AR, (
b
)
5 SA
R
, (c)
10 S
AR, (
d)
50
SA
R
Param
et
ers
us
ed
as
a
ref
e
renc
e
in
this
stud
y
include
m
edian
filt
er
pa
ram
e
te
rs,
di
vid
er
pa
ram
et
ers
on
Gau
s
sia
n
filt
er
,
Geo
MS
a
nd
Gr
ay
MS.
T
he
m
edian
filt
er
par
am
et
er
wh
ic
h
repea
te
dly
m
od
i
fied
was
th
e
valu
e
of
X,
Y
.
Fig
ure
6
shows
one
of
the
m
in
-
m
ax
valu
e
sea
rc
h
res
ults
from
Geo
MS
for
P
CR
5
SA
R
fro
m
10
0
exp
e
rim
ents
by
te
sti
ng
each
siz
e
of
t
he
m
e
dian
filt
er
pa
ra
m
et
ers
and
Ga
us
sia
n
di
vid
er
par
am
et
ers.
Th
e
sam
e
is sh
own
i
n
Fi
gure
7 f
or m
in
-
m
ax
value
s
of
Gr
ay
MS
PCR
1
S
AR.
Figure
6. Ge
om
et
ric Ma
tc
hin
g
Sc
ore
gr
a
ph for PCR
5
S
AR
Figure
7. G
raysca
le
Ma
tc
hing
Score
gr
a
ph
for
PCR
1
S
AR
Figure 7
r
e
pres
ents m
in
-
m
ax
value
f
or
GeoM
S P
CR
5
SAR
is 9
27
.
12
-
994.7
13, while
fro
m
Figu
re
8
m
in
-
m
ax
value
for
G
rayM
S
PCR
1
S
AR
is
941.2
92
-
959.7
96.
T
he
m
at
ching
sc
ore
values
a
re
st
or
e
d
as
database
w
hic
h
are
the
n
use
d
as
ref
e
ren
c
e
scor
e
s
w
he
n
pe
rfor
m
ing
PC
R
te
sti
ng
or
cl
assifi
cat
ion
.
Ba
sed
on
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Geometri
c
and Gr
aysc
ale Te
mp
l
ate M
atchi
ng for
Sa
ud
i A
ra
bi
an Riyal
P
ap
e
r Cur
rency
...
(
Su
ci
A
ulia)
4235
the
te
st
res
ults
of
10
0
trai
ning
data
set
,
the
c
om
par
ison
of
s
cor
e
val
ue
obta
i
ned
f
or
each
Geo
MS
a
nd
G
rayM
S
tem
plate
s o
n
e
ach
bank
no
te
i
s sho
wn in Fi
gure
8, Fi
gure
9, an
d
Fi
gure
10.
Figure
8. Com
par
is
on of
Ge
oM
S and Gray
MS o
n
PCR
1 SAR
Figure
9. Com
par
is
on of
Ge
oM
S and Gray
MS o
n
PCR
5 SAR
Figure
10. C
om
par
ison
of
G
eoMS a
nd
Gr
a
yM
S o
n PCR
10 S
AR
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4231
-
4238
4236
Table
3
.
Accur
acy
Le
vel
of
Currency R
ec
og
niti
on
a
nd it
s
Cl
assifi
cat
ion
No
Ban
k
n
o
tes
Tr
u
e Pos
itiv
e
False Neg
ativ
e
Accurac
y
L
ev
el
Su
ccess rate
of
detectin
g
co
u
n
terfeit
m
o
n
e
y
1.
1
SAR
99
1
9
9
%
1
0
0
%
2.
5
SAR
86
14
8
6
%
1
0
0
%
3.
1
0
SAR
97
3
9
7
%
1
0
0
%
4.
5
0
SAR
99
1
9
9
%
1
0
0
%
Fr
om
PCR
te
st
to
100
sam
ple
data
for
eac
h
bank
no
te
val
ue
,
obta
ine
d
a
r
esult
as
s
how
n
in
Ta
ble
3.
Fr
om
Table
3
ob
ta
ine
d
the
a
ver
a
ge
val
ue
of
accura
cy
le
vel
of
com
bin
ing
Geo
m
et
ric
Te
m
pla
te
Ma
tc
hi
ng
a
nd
Gr
ay
scal
e
Tem
plate
Ma
tc
hin
g
fo
r
the
cl
assifi
cat
ion
proces
s
was
95.
25%.
Wh
il
e
the
ave
r
age
value
of
s
yst
e
m
accuracy
le
ve
l
in
the
recog
niti
on
of
c
ounter
feit
m
on
ey
on
each
bank
no
te
(1
SA
R
,
5
S
AR,
10
SA
R,
and
50
SA
R)
obtai
ne
d a m
axi
m
u
m
v
al
ue
of
100%
.
4.
CONCL
US
I
O
N
In
this
pap
e
r,
we
ha
ve
s
ucc
essfu
ll
y
dem
on
strat
ed
a
PC
R
(P
ape
r
Cu
rrency
Re
cognit
ion)
an
d
it
s
cl
assifi
cat
ion
s
te
st on
Sa
udi
Ar
a
bian
Ri
ya
l 1
S
AR, 5 S
AR
, 10
S
AR,
a
nd
50
SA
R
with
pro
posed
te
c
hniqu
e
s b
y
com
bin
ing
Ge
om
et
ric
Te
m
pl
at
e
Ma
tc
hin
g
and
G
raysca
le
Tem
plat
e
Matching
m
et
ho
ds
to
produce
a
quit
e
sat
isfyi
ng
res
ult
in
c
ounter
feit
m
on
ey
rec
ogni
ti
on
a
nd
the
cl
assifi
cat
ion
of
it
s
own
cu
rr
e
nc
y.
I
n
t
he
proc
ess
of
recog
nizing
c
ounte
rf
ei
t
m
on
e
y
on
each
ba
nkno
te
syst
em
ca
n
detect
100%,
wh
il
e
the
syste
m
per
form
ance
fo
r
PCR
cl
assifi
cat
ion
with
t
he
best
pa
ram
et
er
s
of
c
om
bin
in
g
Ge
oMS
a
nd
Gr
ay
MS
reac
hed
95.
25
%
posit
ive
recog
niti
on
rat
e an
d 4.75%
ne
gative
recog
niti
on
rate.
ACKN
OWLE
DGE
MENTS
This
wor
k has
been s
upported
b
y
In
te
r
nal F
und R
esearc
h 2
016 I
from
Tel
ko
m
U
ni
ver
sit
y
.
REFERE
NCE
S
[1]
T.
Chakr
abor
t
y
,
et
al
.
,
“
Review
o
f
Vari
ous
Im
age
Proce
ss
ing
Te
ch
nique
s
for
Curre
nc
y
Note
Authe
nti
c
at
ion
,
”
In
t.
J.
Comput.
Eng
.
R
es.
Tr
ends
,
vo
l
/is
sue:
3
(
3
)
,
pp.
1
19
–
122,
2016
.
[2]
O.
R.
Devi
,
et
al
.
,
“
Surve
y
on
Pa
per
Curre
nc
y
R
e
cogni
ti
on
S
y
ste
m
,
”
Int.
J
.
Eme
r
g.
Tr
ends
Technol
.
Comput.
S
ci.
,
vol
/i
ss
ue:
5
(
2
)
,
p
p.
105
–
108
,
201
6.
[3]
P.
S.
Nil
a,
et
al.
,
“
Revi
ew
on
Co
unte
rfe
it
Ind
ia
n
Paper
Curre
n
c
y
Rec
ognition
S
y
s
te
m
,
”
In
t.
J
.
Sc
i. R
es.
Comput
.
S
c
i.
Eng.
In
f
.
Techno
l.
,
vol
/
issue:
2
(
2
)
,
pp
.
88
–
91
,
201
7.
[4]
B.
Sharm
a,
“
Recogniti
on
of
Ind
ian
Paper
Curre
n
c
y
b
ase
d
on
LBP,
”
Int.
J
.
Comput.
Appl
.
,
vol
/i
ss
ue:
59
(
1
)
,
pp.
24
–
27,
2012
.
[5]
E.
Pila
n
ia
and
B.
Arora,
“
Rec
o
gnit
ion
of
Fake
Curre
nc
y
Based
on
Secur
ity
Th
rea
d
Feat
u
re
of
Curre
nc
y
,
”
Int
.
J.
Eng.
Comput
.
S
c
i
.
,
vol
/
issue:
5
(
1
7136
)
,
pp
.
1713
6
–
17140,
2016
.
[6]
K.
Chakra
bort
y
,
et
al.
,
“
Rec
en
t
Deve
lopments
in
Paper
Curre
nc
y
rec
ogn
it
ion
S
ystem,”
Int.
J
.
Res
.
Eng.
Technol
.,
vol
/i
ss
ue:
2
(
11
)
,
pp.
222
–
226
,
20
13.
[7]
V.
K.
Jain,
“
Indi
an
Curre
nc
y
De
nom
ina
ti
on
Ide
n
ti
ficat
ion
Us
ing
Im
age
Proce
ss
ing
Te
chni
qu
e,”
In
t.
J
.
Comput.
Sci.
Inf.
Te
chnol.
,
v
ol
/i
ss
ue:
4
(
1
)
,
pp.
126
–
128,
2013
.
[8]
P.
J.
Grac
e
and
D.
Ph,
“
A
Survey
on
Fak
e
Ind
ian
Paper
Curre
n
c
y
Id
ent
if
icati
on
S
y
stem,”
Int.
J
.
Adv
.
Re
s.
Comput.
Sci
.
Soft
w
.
Eng
.
,
vol
/i
ss
ue:
6
(
7
)
,
p
p.
340
–
345
,
201
6.
[9]
R.
Rat
he
e,
“
A
Revi
ew
Paper
o
n
Curre
nc
y
R
ecogniti
on
S
y
s
t
em,”
J
.
Net
w.
Com
mun.
Eme
rg.
Technol
.
,
vol
/i
ss
ue
:
6
(
8
)
,
pp
.
5
–
6
,
20
16.
[10]
S
.
Singh,
et
al.
,
“
Curre
nc
y
Recogniti
on
on
Mobile
Phones,”
2
014
22nd
Inte
rnational
Confe
r
enc
e
on
Pattern
Re
cogn
it
ion
IE
E
E
,
pp
.
2661
–
266
6
,
2014
.
[11]
M.
S.
V
.
Vashis
htha
,
“
A
pape
r
c
urre
nc
y
r
ec
ogn
ition
s
y
st
em
using
image
p
roc
essin
g
to
improve
th
e
reliab
i
li
t
y
with
pca
m
et
hod
,
”
Int
.
J
.
Eng. Sci.
Re
s
.
Techno
l
.
,
vo
l
/i
s
sue:
9655
(
6
)
,
pp
.
172
–
175,
2015.
[12]
P.
D.
Pawar
and
S.
B.
Kal
e,
“
Re
cogni
ti
on
of
Ind
ia
n
Curre
n
c
y
N
ote
Based
on
HS
V
Par
amete
rs,”
Int.
J.
S
ci
.
Re
s
.
,
vol
/i
ss
ue:
3
(
6
)
,
p
p.
132
–
137
,
201
4.
[13]
S.
E.
Ali
and
S.
Mukherje
e
,
“
Chal
l
enge
s
In
Indian
Curre
nc
y
Den
om
ina
ti
on
Rec
o
gnit
ion
&
Auth
e
nti
c
at
ion
,
”
Int
.
J
.
Re
s. Eng.
Techn
ol
.
,
vol
/
issue:
3
(
11
)
,
pp
.
477
–
48
3,
2014
.
[14]
Rashm
i
C
.
and
D.
H.
K.
,
“
Im
age
Proce
ss
ing
Ap
proa
ch
for
INR
Curre
nc
y
Not
e
Num
ber
Rec
ogn
it
ion
S
y
stem
for
Autom
at
ed
T
el
l
e
r
Mac
hin
es,
”
Int
.
J. Comput. A
pp
l
.
Techno
l. Res
.
,
vol
/i
ss
ue:
5
(
8
)
,
p
p.
539
–
542
,
201
6.
[15]
I.
A.
Suresh,
“
India
n
Curre
n
c
y
R
ec
ogni
ti
on
and
Veri
ficati
on
Us
i
ng
Im
age
Pro
ce
ss
ing,
”
Int
.
Re
s
.
J
.
Eng.
Te
chnol
.
,
vol
/i
ss
ue:
3
(
6
)
,
p
p.
87
–
91
,
2016
.
[16]
S.
Sahu
and
T.
Verm
a,
“
Ide
ntification
of
Pape
r
Curre
nc
y
Tec
hnique
s
:
A
Surve
y
,
”
In
t.
J
.
Sci.
Technol.
Eng
.
,
vol
/i
ss
ue:
2
(
12
)
,
pp.
607
–
612
,
20
16.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Geometri
c
and Gr
aysc
ale Te
mp
l
ate M
atchi
ng for
Sa
ud
i A
ra
bi
an Riyal
P
ap
e
r Cur
rency
...
(
Su
ci
A
ulia)
4237
[17]
B.
S.
Prasanthi
a
nd
D.
R.
Setty
,
“
India
n
Paper
Cu
rre
nc
y
Authen
ti
c
at
ion
S
y
st
em using
Im
age
proc
essing,
”
Int
.
J.
Sc
i
.
Re
s. Eng.
Techn
ol.
(
IJS
R
ET)
,
vol
/i
ss
ue:
4
(
9
)
,
pp
.
973
–
981,
2015
.
[18]
H.
Hass
anpour
,
et
al
.
,
“
Fea
ture
ext
ra
ct
ion
for
pa
per
cur
r
ency
r
ecogniti
on,
”
2007
9th
Inte
rnat
iona
l
Symposium
on
Signal
Proce
ss
in
g
and
Its
Applic
ati
ons IE
EE
,
pp.
1
–
4
,
2007
.
[19]
D.
P.
Univ
ersit
,
et
al.
,
“
Math
e
m
at
ic
a
l
and
Sta
ti
stical
Methods
for
Actu
arial
Scie
nc
es
and
Fi
nanc
e
,
”
Mila
no
,
Springer
Mil
an,
2010.
[20]
F.
M.
Hasanuz
z
aman,
et
al.
,
“
Robust
and
Eff
e
ctive
Com
ponent
-
Based
Banknote
Rec
ogni
ti
on
for
the
Bli
nd
,
”
IEEE
Tr
ansacti
ons On Sy
stems,
Man,
And
Cyb
erne
tics
,
vol
/i
ss
ue:
42
(
6
)
,
pp
.
1021
–
1030
,
2012
.
[21]
C.
Bhurke
,
et
a
l.
,
“
Curre
nc
y
Re
c
ognit
ion
Us
ing
I
m
age
Proce
ss
ing,
”
Int
.
J
.
Inno
v.
Re
s.
Co
mput
.
C
omm
un.
Eng.
,
p
p.
4418
–
4422,
201
5.
[22]
J.
W
.
L
ee,
e
t
al
.
,
“
A
Surve
y
on
Banknot
e
Rec
og
nit
ion
Me
thods
b
y
V
ari
ous
Sens
ors,”
S
ensors
,
v
ol
/i
ss
ue:
17
(
313
)
,
pp.
1
–
34
,
2017
.
[23]
M.
Akbar,
et
al.
,
“
Origina
l
an
d
Counte
rfe
it
Mone
y
Detect
io
n
Based
on
Edge
Dete
c
ti
on,
”
20
13
Int.
Conf.
Instrum
ent
ati
on,
Comm
un.
Inf
.
T
ec
hnol
.
Bi
omed
.
Eng
.
,
2013.
[24]
F.
Ta
k
eda
and
S.
Om
at
u,
“
High
Speed
Pape
r
Curre
nc
y
,
”
I
EEE
transacti
ons
on
neural
ne
two
rk
,
vol
/i
ss
ue:
6
(
1
),
1995
.
[25]
K.
Sawant
and
C.
Mor
e,
“
Cu
rre
nc
y
R
ec
ogn
ition
Us
ing
Im
ag
e
Proce
ss
ing
an
d
Minim
um
Distanc
e
Cla
ss
ifie
r
Te
chn
ique
,
”
In
t.
J.
Adv. Eng. Re
s
.
Sci
.
,
vol
/
issue:
6
495
(
9
)
,
pp
.
1
–
8
,
2016.
[26]
S.
S.
Sannakki
and
P.
J.
Gunjal
e,
“
Rec
ogn
it
ion
and
Cla
ss
ifi
cat
ion
of
Curre
nc
y
No
te
s
using
D
iscre
t
e
W
ave
let
Tra
nsform
,
”
In
t.
J.
Eme
rg.
Techn
ol.
Adv. Eng
.
,
vo
l
/i
ss
ue:
4
(
7
)
,
pp.
253
–
257,
2014
.
[27]
M.
Sarfra
z
,
“
An
int
e
ll
ig
ent
pap
e
r
cur
ren
c
y
re
co
gnit
ion
s
y
s
te
m
,
”
Proce
dia
-
Pro
ce
dia
Comput
.
Sci
.
,
vol
.
65
,
pp
.
538
–
545,
2015
.
[28]
B.
B.
La
ksono,
et
al
.
,
“
Opera
si
Vala
s
:
Ide
nt
ifi
k
asi
Nom
ina
l
Denga
n
Metode
Ca
nn
y
Edge
De
te
c
t
ion
dan
Te
m
pla
t
e
Matc
hing
Foreig
n
Exc
hang
e
:
Nom
ina
l
Ide
nti
f
icat
ion
Us
ing
Cann
y
Edge
Detect
ion and
Te
m
plate
Matc
h
ing
Abs
trak
Diskus
i
Sharpe
n
ing
dan
Sm
oothi
ng
Sharpe
ning
m
eru
paka
n
p
enaj
ama,
”
SNIKO
,
p
p.
10
–
11
,
2015
.
[29]
M.
Krishnave
ni
,
et
al
.
,
“
Im
prove
d
Cann
y
Edge
s
Us
ing
Cel
lul
a
r
Based
Particle
S
warm
Optimiza
t
ion
Techni
que
f
o
r
Ta
m
il
Sign
Digital
Im
age
s
,
”
Int.
J.
Elec
tr.
Compu
t.
Eng
.
,
vo
l
/i
ss
ue
:
6
(
5
)
,
p
p.
2158,
2016.
[30]
R.
Supri
y
ant
i
,
e
t
al.
,
“
Bright
nes
s
and
Contra
st
Modific
a
ti
on
in
Ultra
sonograph
y
Im
age
s
Us
ing
Edge
Det
ec
t
ion
Result
s,”
TEL
KOMNIKA
Teleco
mm
unic
ati
on
Co
mput.
E
lectron.
Control
,
vo
l
/i
ss
ue:
14
(
3
)
,
pp
.
109
0
–
1098,
2016
.
[31]
A.
Fahruroz
i
,
et
al.
,
“
W
ood
Clas
sific
at
ion
Base
d
on
Edg
e
De
tecti
ons
and
T
exture
Fea
ture
s
Se
l
ec
t
ion,
”
vol
/i
ss
ue:
6
(
5
)
,
pp
.
2167
–
2
175,
2016
.
[32]
W
.
K.
Prat
t,
“
Digit
al Im
age Proc
essing,
”
vo
l
/i
ss
u
e:
5
(
11
),
2001
.
[33]
S.
Shrestha,
“
I
m
age
Denoising
Us
ing
New
Ad
apt
iv
e
Based
M
edi
an
Fil
te
r
,
”
Si
gnal
Image
Pro
ce
ss
.
An
Int
.
J
.
,
v
ol
/i
ss
ue:
5
(
4
)
,
p
p.
1
–
13
,
2014
.
[34]
S.
Sury
a
and
G.
Tha
ilam
bal
,
“
Com
par
at
ive
Stud
y
on
Curre
nc
y
Re
cogni
ti
on
S
y
s
tem
U
sing
Im
age
Proce
ss
ing,
”
Int
.
J.
Eng. Comput.
Sci
.
,
vo
l
/i
ss
ue:
3
(
8
),
pp
.
7723
–
77
26,
2014
.
[35]
S
.
Chandel
and
M.
T
y
ag
i,
“
Ev
aluate
and
Propos
e
a
Novel
Techn
ique
to
Che
ck
G
enui
nen
ess
of
th
e
Curre
n
c
y
Us
ing
Im
age
Proce
ss
in
g,
”
In
t. J.
Comp
ut.
S
ci.
Tr
ends
Technol
.
,
vol
/
issue:
5
(
1
)
,
pp
.
111
–
116,
2017
.
[36]
G.
Sharm
a,
et
al
.
,
“
Im
age
Rec
og
nit
ion
S
y
st
em
using
Geom
e
tri
c
Matc
hing
and
C
ontour
Detect
ion
,
”
Int
.
J.
Comput
.
Appl
.
,
vo
l
/i
ss
ue:
51
(
17
)
,
pp
.
48
–
5
3,
2012
.
[37]
G.
I.
R
aho,
et
al.
,
“
Cash
Curre
ncies
Rec
ogn
ition
Us
ing
k
-
Ne
are
st
Ne
ighbor
Cla
ss
ifi
er
,
”
In
t.
J.
W
eb
S
emant.
Technol
.
,
vo
l
/i
ss
ue:
6
(
4
)
,
pp
.
11
–
21,
2015
.
[38]
J.
Frank
,
et
al
.
,
“
Ti
m
e
Serie
s
Anal
y
s
is
Us
ing
G
eometr
ic
Te
m
pl
at
e
Mat
chi
ng
,
”
I
EE
E
Tr
ans.
Pat
t
ern
Anal
.
Mac
h
.
Inte
ll
.
,
vol
/i
ss
ue:
35
(
3
)
,
p
p.
1,
20
12.
BIOGR
AP
H
I
ES
OF
A
UTH
ORS
Suci
Aulia
,
M.T
.
is
an
le
ct
ur
er
a
t
Te
lkom
Univer
sit
y
,
Depa
rt
eme
nt
of
Applie
d
Scie
n
ce.
She
is
conc
ern
in
signa
l
proc
essing
sinc
e
6
y
e
ars
ago.
She
has
publi
shed
seve
ral
pap
ers
in
her
rese
arc
h
espe
cially
imag
e
signal
proc
essin
g
fie
lds
such
as
her
la
t
est
public
at
ion
under
th
e
t
it
le
“
Hog
An
d
Ica
B
ase
d
Fa
ce
Rec
ognition S
y
s
te
m
On A Surveilla
n
ce
V
id
eo” ,et
c.
He
gra
dua
te
d
fr
om
El
ectrical
E
ngine
er
ing
a
t
Telkom
Univer
sit
y
Yea
r
2015
with
the
title
Skips
i
"A
utomati
c
Mone
y
Ch
ange
r:
Id
ent
ifica
ti
on
Curr
ency
Input
And
Opera
ti
ng
Oper
a
ti
on
Of
Vala
s
Exc
hang
e
W
it
h
Cann
y
Edge
D
etec
t
ion
And
Te
m
pla
t
e
Matc
h
ing
Method
“
.
Durin
g
the
stud
y
h
e
was a
ctive
as
a
r
ese
arc
h
er at Digi
ta
l
Control L
abo
rat
or
y
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4231
-
4238
4238
Angga
Rusdina
r,
Ph.D
is
an
l
ec
tur
er
a
t
Tel
kom
Univer
sit
y
,
Depa
r
te
m
ent
of
El
e
ct
r
ic
a
l
Engi
ne
eri
ng.
His
skill
s
and
e
cxperti
se
fo
cus
on
Pa
tt
ern
R
ec
ogn
it
io
n,
Com
pute
r
Vis
ion,
Robot
ic
s
El
e
ct
roni
c
Engi
n
ee
ring
,
and
Cont
rol
S
y
st
ems
Eng
ine
er
ing
.
Yu
y
un
siti
rohm
ah
ST,
.
MT
is
an
le
ct
ur
e
at
T
el
k
om
Univer
sit
y
.
She
is
conc
ern
i
n
tra
nsm
issio
n
te
l
ec
om
m
unic
at
i
on
s
y
stem.
Sh
e
has
publ
ishe
d
pape
r
about
per
form
anc
e
of
Orthogonal
Freque
nc
y
Div
ision
Multi
p
le
x
ing
and
Orthog
onal
W
ave
le
t
Divisio
n
Multi
pl
exi
ng
.
Evaluation Warning : The document was created with Spire.PDF for Python.