Internati
o
nal
Journal of Ele
c
trical
and Computer
Engineering
(IJE
CE)
V
o
l.
5, N
o
. 4
,
A
ugu
st
2015
, pp
. 86
9
~
87
8
I
S
SN
: 208
8-8
7
0
8
8
69
Jo
urn
a
l
h
o
me
pa
ge
: h
ttp
://iaesjo
u
r
na
l.com/
o
n
lin
e/ind
e
x.ph
p
/
IJECE
Autom
a
t
i
c Vehicl
e T
r
acki
ng
Sy
stem Ba
se
d o
n
Fix
e
d
Thresholding and Histogra
m Ba
sed Edg
e
Pr
oce
ssing
N
.
S
h
o
b
ha
Ran
i
,
N
e
et
hu
O.
P
.
,
N
ila
Po
nn
at
h
Department of
Computer Scien
ce, Amrita Vishwa
Vi
dy
a
p
ee
t
h
am
M
y
sore C
a
mpus, M
y
sor
e
-57002
6, Karn
atak
a, In
dia
Article Info
A
B
STRAC
T
Article histo
r
y:
Received
Mar 19, 2015
Rev
i
sed
Ap
r
30
, 20
15
Accepted
May 25, 2015
Autom
a
tic det
e
ction
,
extra
c
tion
and recognitio
n of vehicle nu
m
b
er plate
region in traff
i
c control s
y
stems is
one of t
h
e prominent application in
Com
puter vis
i
on. The dras
t
i
c inc
r
eas
e in
number
of vehicles in the curren
t
genera
tion gr
eat
l
y
incr
eas
es
the
com
p
lexit
y
in tr
acking
the v
e
hi
c
l
es
throug
h
the human vis
u
al s
y
s
t
em, manual pro
cedure of controlling
traff
i
c
and
enforcem
ent
of
various
laws
and
rules
is
not suff
icient for smooth control of
traffic. This urg
e
s the need for devel
opment of technolog
y
that can automate
this process. This paper mainly
f
o
cu
ses on the d
e
velopment of
an
automatic
num
ber plat
e
extra
c
tion
and
recogn
ition
a
l
gorithm
b
y
i
n
corporat
ing
constructs like
edge detection
,
horizont
al and vertical edge processing using
fixed thr
e
shold
techn
i
que. Th
e extr
ac
ted nu
mber plate region is again
processed using template matching al
gorithm for the recogn
ition of the
chara
c
t
e
rs
em
bos
s
e
d on the num
ber plate with
res
p
ect to
ever
y ind
i
vidua
l
piec
e of num
ber
plat
e.
The
algo
r
ithm
develop
e
d
has
ach
ieved
an
accur
a
c
y
o
f
around 100%
an
d works for both
front
and r
ear
images of th
e
car
.
Keyword:
Ch
aracter recog
n
ition
Edg
e
Detectio
n
Fi
xed
p
o
i
n
t
s
t
h
resh
ol
di
ng
Tem
p
late matc
h
i
ng
Vehicle license
plate ext
r
action
Copyright ©
201
5 Institut
e
o
f
Ad
vanced
Engin
eer
ing and S
c
i
e
nce.
All rights re
se
rve
d
.
Co
rresp
ond
i
ng
Autho
r
:
N
.
Shob
h
a
Rani,
Depa
rt
m
e
nt
of
C
o
m
put
er Sci
e
nce,
Am
rits Vishwa
Vidy
a
p
eetham
,
M
y
sore
C
a
m
pus,
M
y
s
o
re
, 57
00
2
6
, Ka
rnat
a
k
a, I
ndi
a.
Em
a
il: n
.
sho
b
a
1
985
@g
m
a
il.c
o
m
1.
INTRODUCTION
Licen
se
Plate Ex
traction
and
Reco
gn
itio
n sy
ste
m
s are
u
s
ed
to
track
and
mo
n
itor t
h
e m
o
vin
g
v
e
h
i
cles
by
aut
o
m
a
t
i
call
y
ext
r
act
i
ng t
h
e num
ber pl
at
e
s
. A
u
t
o
m
a
t
i
c
license pl
at
e rec
o
g
n
i
t
i
on
bec
o
m
e
very
im
por
t
a
nt
i
n
the
curre
nt
tec
h
nology gene
ration b
ecause
of t
h
e
unlim
ite
d inc
r
ease
of c
a
rs a
n
d transportation system
s
which
causes
di
ffi
c
u
l
t
y
i
n
t
r
acki
n
g
v
e
hi
cl
es fo
r t
h
e
pu
r
pose
of
pa
r
k
i
n
g sy
st
em
, t
r
affi
c m
a
nagem
e
nt
sy
st
em
and
l
a
w
enforcem
ent e
s
pecially at state bord
ers
etc. T
h
e
drastic
increas
e in num
b
er of
ve
hicles in last c
o
uple
of
decade
s
really com
p
licated the
j
o
b
of trac
king the
ve
hic
l
es
m
a
nually
by authorities of traffic c
ontrol and
m
a
nagem
e
nt
.
The m
a
nual
m
e
t
h
o
d
fo
r co
nt
r
o
l
l
i
ng t
r
af
fi
c and e
n
f
o
rcem
ent
of l
a
ws i
s
not
suffi
ci
e
n
t
eno
u
g
h
t
o
man
a
g
e
the traffic esp
ecially in
m
e
tro
p
o
litan cities.
An au
t
o
m
a
ted
syste
m
is n
ecessary to
i
d
en
tify v
e
h
i
cles a
n
d fetch t
h
e
vehi
cles inform
ation since
the
sy
st
em
pl
ay
s an i
m
port
a
nt
r
o
l
e
i
n
det
ect
i
n
g
securi
t
y
t
h
re
at
by
i
d
e
n
t
i
f
y
i
ng
t
h
e i
n
di
vi
d
u
al
ow
ni
n
g
t
h
e ve
hi
cl
e.
Th
is prov
id
es m
u
ch
sco
p
e
for track
i
n
g
th
e crimin
als who
are trespa
ssing the traffic rules
and ca
using
ha
rm
to
ot
he
r i
n
di
vi
d
u
a
l
s i
n
t
h
e soci
et
y
.
The a
ppl
i
cat
i
ons
of a
u
to
m
a
ted
traffic con
t
ro
l system
s als
o
include trac
king
of
st
ol
en/
s
u
s
pi
ci
o
u
s
ve
hi
cl
es i
.
e.
, sy
st
em
whi
c
h
i
s
de
pl
oy
e
d
o
n
t
h
e
r
o
a
d
si
de
can
pe
rf
orm
a m
a
t
c
h bet
w
ee
n t
h
e
passi
n
g
cars a
nd t
h
e bl
ac
k l
i
s
t
whi
c
h co
nt
a
i
ns a l
i
s
t of st
ol
en cars o
r
u
npai
d
fi
nes
.
The '
b
l
ack l
i
s
t'
can be
up
dat
e
d i
n
real
t
i
m
e
and pr
ov
i
d
e im
m
e
di
at
e
al
arm
or si
ren t
o
t
h
e pol
i
ce f
o
rce
,
t
o
l
l
i
ng an
d b
o
r
d
er c
ont
r
o
l
i
.
e.
,
th
e
b
o
rd
er cro
s
sin
g
s
will
b
e
m
o
n
ito
red
b
y
t
h
e system
wh
i
c
h
m
a
k
e
s use
o
f
th
e car
nu
mb
ers reg
i
stered in
t
h
e
entry or exits to the
country.
It reduces the t
i
m
e
and helps
to cal
culate the travel fee
in
toll road a
n
d access
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
J
ECE
Vo
l. 5
,
N
o
. 4
,
Aug
u
s
t 2
015
:
86
9
–
87
8
87
0
cont
rol
a
nd
p
a
rki
ng i
.
e
., T
h
i
s
sy
st
em
wi
l
l
be de
pl
oy
e
d
on t
h
e gat
e
o
f
a secu
re
d ar
ea so t
h
at
, t
h
e gat
e
autom
a
tically
m
onitors the vehicles num
be
r plates an
d
open
s
on
ly for au
tho
r
ized
m
e
mb
ers. Th
e au
tomatio
n
o
f
all th
e ab
ove app
licatio
n
s
assists in
reso
l
v
ing
m
a
n
y
con
f
licts th
at m
a
y in
cur
d
u
ring th
e en
v
i
si
o
n
o
f
t
h
e
vehicle
trac
king. The
tracing of sus
p
icious vehicles
w
ill be
done
ve
ry ec
onom
ica
lly without c
o
ns
um
ption of
lo
ts of
m
a
n
pow
er.
An e
fficient a
nd
reliable sy
stem
fo
r licen
se p
l
ate ex
traction
and
reco
gn
it
io
n
is to
b
e
in
t
r
odu
ced
to
sol
v
e t
h
e c
h
al
l
e
nge
s i
n
m
a
nual
m
e
t
hod t
r
ac
ki
n
g
t
h
e ve
hi
cl
es. Thi
s
m
a
kes us t
o
m
o
t
i
v
ate for t
h
e resea
r
ch o
n
devel
opm
ent
o
f
a
u
t
o
m
a
t
e
d t
echn
o
l
o
gy
t
h
at
que
nc
hes t
h
e
m
u
lt
i
p
l
e
needs
i
n
t
r
af
fi
c c
ont
r
o
l
an
d m
a
nage
m
e
nt
.
2.
RELATED WORK
Nu
m
b
er
p
l
ate ex
traction
and
recog
n
ition
h
a
d
un
d
e
rg
on
e n
u
m
erou
s exp
e
rim
e
n
t
atio
n
s
b
y
v
a
riou
s
researc
h
er
s.
M
a
ny
ap
pr
oac
h
es
were i
m
pl
em
ent
e
d an
d
had
u
n
d
er
go
ne m
a
ny
change
s i
n
t
h
e
vi
ew
of
im
provising a
ccuracy i
n
license
pl
at
e ext
r
act
i
on;
s
o
m
e
of
t
h
e a
p
pr
oa
ches
de
vel
o
pe
d
by
resea
r
c
h
ers a
r
e
revi
e
w
ed
bel
o
w.
Sarb
j
it Kau
r
et
. al [1
] h
a
s
pro
p
o
s
ed
an
al
go
rith
m
fo
r au
t
o
m
a
t
i
c n
u
m
b
e
r d
e
tection
and
recog
n
ition
usi
n
g m
o
rph
o
l
ogi
cal
o
p
erat
i
ons
fo
r p
r
e-
pr
ocessi
n
g
, S
o
be
l
operat
o
r f
o
r
vert
i
cal
edge
det
ect
i
o
n
,
co
n
n
ect
ed
com
pone
nt a
n
alysis for se
gm
entation a
n
d rec
o
gnition.
Th
e al
gorithm
ha
ve attaine
d
a ove
r
all acc
uracy
of
aro
u
n
d
9
7
%
.
Ser
k
an
Oz
bay
et
. al
.
[
2
]
ha
d
pr
o
ppe
d
a s
i
m
p
l
e
al
gori
t
h
m
usi
ng e
d
ge
det
ect
i
o
n al
g
o
ri
t
h
m
s
,
sm
earin
g
alg
o
rith
m
s
an
d
te
m
p
la
te
match
i
n
g
b
a
sed
r
ecog
n
ition
app
r
o
a
ch
for ch
aracter recogn
itio
n
.
Th
e
al
go
ri
t
h
m
i
s
speci
fi
cal
l
y
desi
gned f
o
r t
h
e rec
o
g
n
i
t
i
on r
e
q
u
i
r
em
ent
s
of Tu
rk
i
s
h l
i
cense pl
at
es and ha
d ach
i
e
ved
an ove
r
all accuracy
of a
r
ound
97%
.
D.G.
Bailey et. al [
3
] ha
d de
signe
d
a m
odular
struct
ure i
n
terfa
ce for
eval
uat
i
o
n an
d
com
p
ari
s
on a
n
al
y
s
i
s
of va
ri
ous
fo
r n
u
m
b
er pl
at
e reco
gni
t
i
on al
g
o
ri
t
h
m
s
. Joh
n
s
on et
al
[4]
ha
d
d
e
v
i
sed
an
algo
rith
m
fo
r num
b
e
r p
l
ate reco
gn
itio
n b
y
em
p
l
o
y
in
g
op
tical ch
aracter reco
gn
itio
n tech
n
i
q
u
e
s.
Nai
k
ur
B
h
a
r
at
kum
ar G
o
hi
l
e
t
.al
[5]
has
p
r
o
ppe
d
an
ap
p
r
o
ach
o
n
ca
r l
i
c
e
n
se
pl
at
e
det
ect
i
on
usi
n
g
hi
st
og
ram
base
d approac
h
, it proce
sse
s each
fram
e
indivi
dually
and provide
s
the
co-ordin
at
es of location with
m
a
xim
u
m
probability of ha
ving a num
b
er plate. Rinku
solanki et.al [6] has proposed a
m
e
thod of a
u
tom
a
tic
licen
se p
l
ate reco
gn
itio
n, in
wh
ich
licen
se p
l
ate is ex
tr
acted
b
a
sed
o
n
so
me featu
r
es lik
e
co
lor, th
e bou
nd
ary
,
or t
h
e existe
nc
e of t
h
e cha
r
a
c
ters. In the li
cense
pl
ate segm
entation stage, the c
h
a
r
acters are e
x
tract
ed by
p
r
oj
ectin
g
th
ei
r co
lor in
formatio
n
,
b
y
lab
e
lin
g
th
em
, o
r
by
m
a
tch
i
n
g
th
eir p
o
s
ition
s
with
tem
p
late.
Ku
m
a
r
Paras
u
ram
a
n et
.al
[7]
has co
nt
ri
b
u
t
e
d a
n
app
r
oach
on
Indian ve
hicle license plat
e ext
r
action a
nd c
h
aracter
segm
ent
a
t
i
on b
a
sed
on
t
h
e m
o
rp
h
o
l
o
gi
cal
al
g
o
ri
t
h
m
s
and
co
nnect
e
d
c
o
m
ponent
s a
n
al
y
s
i
s
.
San
d
ra
Si
va
na
nd
a
n
et.al [
8
]
h
a
s co
n
t
r
i
bu
ted
a
meth
o
d
o
l
og
y
o
n
au
to
m
a
tic
v
e
h
i
cle i
d
en
tificatio
n
thr
ough
edg
e
d
e
tectio
n an
d
m
o
rp
ho
log
i
cal op
eration
s
, a
Scan
lin
e algorith
m
fo
r se
g
m
en
tatio
n
and
reco
gn
itio
n of seg
m
en
ted
ch
aracters.
C
h
et
an S
h
arm
a
et
.al
[9]
and
has de
vi
se
d a m
e
t
hod
on
In
di
an Ve
hi
cl
e l
i
cense Pl
at
e Ext
r
a
c
t
i
on usi
ng
hi
st
og
ram
equal
i
zat
i
o
n m
e
t
h
o
d
,
m
o
rph
o
l
ogi
cal
ope
rat
i
ons
a
n
d
ed
ge
Detection techniques. Sheetal Mithun
Kawa
de et.a
l
[1
0]
has
c
ont
ri
but
e
d
a
n
a
p
p
r
o
ach
of
a real
t
i
m
e
vehi
cl
e l
i
cense
pl
at
e rec
o
g
n
i
t
i
on
by
em
pl
oy
i
n
g
an
ap
p
r
o
p
ri
at
e
th
resh
o
l
d
techn
i
qu
e for seg
m
en
tatio
n
and
tem
p
la
te
m
a
tch
i
n
g
techn
i
qu
e fo
r
reco
gn
itio
n. Pratish
t
h
a
Gup
t
a et.al
[1
1]
has
de
vel
ope
d a
n
aut
o
m
a
t
e
d sy
st
em
usi
ng S
I
M
U
LI
N
K
m
odel
i
n
M
a
t
l
a
b whi
c
h
ext
r
act
s t
h
e
n
u
m
b
er
p
l
ate and
recogn
ize alph
anu
m
eric ch
aracters
an
d
reco
gn
ition
is p
e
rform
e
d
u
s
ing
tem
p
late
match
i
n
g
.
R
o
nak
P.
Patel et.al [12] h
a
s
p
r
op
o
s
ed
an
ap
pro
ach on
au
to
m
a
tic
licen
ses
p
l
ate reco
gn
itio
n usin
g m
o
rp
ho
log
i
cal
ope
rat
i
o
ns
an
d edge det
ect
i
o
n t
echni
q
u
es
.
A
.
Ak
o
u
m
et
.al
[13]
has
pr
op
ose
d
a ne
w a
p
p
r
oa
ch f
o
r
det
ect
i
o
n an
d
Ide
n
t
i
f
i
cat
i
o
n
o
f
vehi
cl
e num
ber by
com
b
i
n
i
n
g feat
ure
s
of h
o
ri
z
ont
al
gra
d
i
e
nt
s
a
n
d
m
e
t
hod
sy
m
m
et
ry
.
Th
e list of ex
perim
e
n
t
at
io
n
s
o
n
ex
tractio
n
an
d reco
gnition of ve
hicle license
plate are
very wide
.
It
is clearly not
iceable that
many
m
e
thodologies are
devised usi
n
g
techniques
of pre-processi
ng like
m
o
rph
o
l
o
gi
cal
ope
rat
i
ons
, C
a
nny
ed
ge det
e
ct
i
on an
d hi
st
o
g
ram
m
a
t
c
hi
ng
et
c. The sci
e
nt
i
f
i
c
obser
vat
i
o
ns
o
f
exi
s
t
i
ng m
e
t
hod
ol
o
g
y
wi
t
h
t
h
e pr
o
p
o
s
ed
dat
a
set
s
em
ploy
i
ng ca
n
n
y
edge
det
ect
i
on
and s
o
bel
d
g
e
edg
e
d
e
tectio
n are as fo
llo
ws.
B
o
t
h
t
h
e S
o
bel
edge
det
ect
or
and C
a
nny
e
d
ge det
ect
o
r
u
s
es a pai
r
o
f
3
x
3
co
n
vol
ut
i
on
m
a
t
r
i
ces for
id
en
tifying
th
e in
ten
s
ity d
i
scon
tin
u
ities in the i
m
ag
es. Th
e
v
e
ry im
p
o
r
tan
t
pro
s
p
ect
o
f
sob
e
l edg
e
d
e
tctor is, i
t
i
s
dedi
cat
ed f
o
r eval
uat
i
o
n o
f
vert
i
cal
edges
and h
o
r
i
z
o
n
t
a
l edges. It is v
e
ry
effici
ent com
p
ared
to
any ed
ge
det
c
t
i
on
o
p
erat
or
w
h
en i
t
c
o
m
e
s t
o
i
d
ent
i
f
i
cat
i
on
of
ho
ri
z
ont
al
or
vert
i
c
al
edges
,
si
nce
canny
e
d
ge d
e
t
ect
i
on
t
o
ext
r
act
us
ef
ul
st
r
u
ct
u
r
al
i
n
fo
rm
ati
on
fr
o
m
di
fferent
vi
s
i
on
o
b
ject
s
an
d
dram
at
i
call
y
re
duce
t
h
e am
ou
nt
of
dat
a
t
o
be p
r
oc
essed.
The re
d
u
ct
i
on
of
num
ber
of s
m
al
l
e
r edges i
s
not
ne
gl
i
g
i
b
l
e
i
n
case of
so
bel
ed
ge det
e
c
t
i
on an
d t
h
e
sm
al
l
e
r or m
i
nut
e ed
ges
are
v
e
ry
i
m
port
a
nt
i
n
t
h
e
re
searc
h
e
s
rel
a
t
e
d
t
o
t
h
e
ext
r
act
i
o
n
o
f
num
ber
pl
at
e r
e
gi
o
n
because the characters em
bos
sed on the
num
b
er plate are com
posed
of
very
thin or minute edges. More ove
r
the m
a
jor foc
u
s of t
h
e e
x
peri
mentation is t
o
extract
the
num
ber
plate
re
gion form
a car
im
age and rec
o
gnize
the various c
h
aracters em
bosed on it.
Thus t
h
e cur
r
e
n
t
experi
m
e
nt
at
ion
has co
nsi
d
ered t
h
e S
obe
l
edg
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Au
toma
tic Vehicle Tra
cking
System
Ba
sed
on
Fixed
Th
resho
l
d
i
ng
an
d H
i
sto
g
r
a
m
Ba
sed
… (N. S
hobh
a
Ran
i
)
87
1
d
e
tectio
n
as ideal fo
r
d
e
tectio
n
o
f
in
ten
s
ity d
i
scon
ti
n
u
ities in
th
e i
m
ag
es. Th
e Figure 1
(
a) and
Figu
re 1
(
b)
sho
w
s t
h
e res
u
l
t
s
of so
bel
an
d
canny
ed
ge
de
t
ect
i
on o
n
t
h
e i
n
p
u
t
i
m
ages im
bi
bed. The
p
r
o
p
o
sed m
e
t
h
o
dol
ogy
cont
ri
b
u
t
e
s a no
vel
al
g
o
ri
t
h
m
whi
c
h m
a
kes use o
f
a
S
obel
e
dge
det
ect
i
on m
e
t
hod and a fi
xed t
h
resh
ol
d
tech
n
i
qu
e fo
r th
e
d
e
tectio
n of licen
se
pl
at
e r
e
gi
o
n
a
n
d a
n
a
u
t
o
m
a
t
i
c
im
ag
e cropping m
e
thod
for the
ext
r
action
of license
plate re
gion.
Fi
gu
re
1(a
)
.
R
e
sul
t
s
o
f
C
a
nny
and
S
obel
e
d
ge
det
ect
i
o
n
Fig
u
re
1
(
b). The v
e
ry sm
al
ler d
e
tails are
no
t
retain
ed in
t
h
e
left im
ages where as i
n
right image the
fine
r
edge
det
a
i
l
s
are
pre
s
erve
d as
a
resu
l
t
of
So
bel
ope
rat
o
r
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
J
ECE
Vo
l. 5
,
N
o
. 4
,
Aug
u
s
t 2
015
:
86
9
–
87
8
87
2
3.
R
E
SEARC
H M
ETHOD
In
th
e
p
r
op
o
s
ed
m
e
th
o
d
o
l
og
y th
e ex
t
r
actio
n
an
d
reco
gn
ition
of
n
u
m
b
e
r
p
l
ate reg
i
on
on
car im
ag
es is
accom
p
lished in three
stages.
In sta
g
e
1, pre-processi
ng
of t
h
e input im
age
is perform
e
d to convert the i
n
put
im
age into an
enha
nce
d
or
noise-fr
ee
gray
scale im
age. The ext
r
action
of
num
b
er plate
regi
ons is ac
hieved
t
h
r
o
u
g
h
hi
st
og
ram
based
h
o
ri
zont
al
a
n
d
vert
i
cal
edge
p
r
oc
e
ssi
ng
u
s
i
n
g a
fi
xed
t
h
resh
ol
d t
echni
que
i
n
st
age
2
to
id
en
tify the p
r
o
b
a
b
l
e reg
i
on
s of
number
plate and the stage 3
encom
p
asses the segm
entation a
nd
r
ecogn
itio
n
o
f
ch
ar
acters em
b
o
ssed
on
t
h
e nu
m
b
er
p
l
ate r
e
g
i
on
b
y
ap
p
l
ying
boud
ing
box
m
e
t
h
od
an
d
te
m
p
late
matc
hing algorithm
for re
c
o
gnition. The use
r
interface desi
gned
for the
proposed system
is as
depi
ct
ed
i
n
t
h
e
Fi
gu
re
2.
Figure
2. Graphical use
r
inte
rface for a
u
to
matic num
ber
plate extraction a
n
d rec
o
gnition
3.
1 Pre
-
Pr
oce
ssi
ng:
The pre-proces
sing is a pre-re
quisite operati
on
for
any of the im
age proce
ssi
ng system
to
convert the
input im
ages into t
h
e
fo
rm suitable
for
succee
ding sta
g
es
of
pr
oces
sing. In the
proposed system
pre-
p
r
o
cessi
n
g
en
co
m
p
asses th
e stag
e 1 of th
e au
to
m
a
tic n
u
m
b
e
r p
l
ate ex
tract
io
n
and
recog
n
itio
n
.
In
itially a RGB
i
m
ag
e is co
n
s
i
d
ered
as inpu
t to
th
e pro
p
o
s
ed
system
wh
ich
will b
e
tran
sfo
r
m
e
d
to
a gray scale i
m
ag
e. Fu
rt
h
e
r
Gau
s
sian
filterin
g
[2
] is app
l
ied
to
elim
in
at
e th
e
n
o
i
se fro
m
th
e i
m
ag
e and
h
i
gh
lig
h
t
th
e
h
i
gh
frequ
en
cy
com
pone
nt
s i
n
t
h
e im
age, si
nce im
ages capt
u
re
d i
n
real
t
i
m
e
m
a
y
consi
s
t
s
of l
o
t
s
of
ba
ck g
r
o
u
nd
noi
s
e
and
u
n
b
a
lan
c
ed
illu
m
i
n
a
tio
n
artifacts. Mo
rph
o
l
o
g
i
cal op
eratio
n
is app
lied
on
th
e im
ag
e to
co
nn
ect th
e
brok
en
edge
an
d t
h
e
g
r
adi
e
nt
o
p
erat
i
ons
[
4
]
are
pe
r
f
o
r
m
e
d t
o
t
h
i
c
ken
t
h
e e
d
ges
of t
h
e i
m
age. Fi
rst
o
r
de
r
der
i
vat
i
v
e
m
a
sk, So
bel
i
s
appl
i
e
d t
o
hi
ghl
i
g
ht
t
h
e ed
ges o
r
i
e
nt
e
d
at
ho
ri
zo
nt
al
an
d ve
rt
i
cal
di
re
ct
i
ons. T
h
e Fi
gu
re 3
sh
ows th
e i
n
pu
t d
a
tasets
u
s
ed
for au
to
m
a
ti
c nu
m
b
er
p
l
ate ex
traction
and reco
gn
itio
n. Th
e fi
gu
re
4
p
r
o
v
i
d
e
s
t
h
e o
u
t
p
ut
of
p
r
e-
pr
ocessi
ng
.
3.2.
De
tection
of license plate region
The st
a
g
e 2
o
f
t
h
e
pr
o
p
o
s
ed
m
e
t
hod
ol
o
g
y
em
pl
oy
s hi
st
og
ram
based
pr
ocessi
ng
f
o
r
det
ect
i
o
n
o
f
ho
ri
zo
nt
al
an
d
vert
i
cal
ed
ges
t
h
r
o
u
g
h
w
h
i
c
h
t
h
e p
r
o
b
a
b
l
e
num
ber
pl
at
e r
e
gi
o
n
ca
n be
d
e
t
ect
ed. R
e
gi
o
n
wi
t
h
the
m
a
xim
u
m
histogram
value is considere
d
as the
m
o
st
pr
o
b
abl
e
can
di
dat
e
fo
r n
u
m
b
er plate, since in the
p
r
e-pro
cessed
g
r
ay scale im
a
g
e ho
rizon
t
al an
d
v
e
rtical
edge are h
i
gh
ligh
t
ed
and
wh
ich
will h
a
v
e
d
o
m
i
n
atin
g
cont
i
n
u
o
u
s
hi
g
h
f
r
eq
ue
nci
e
s
defi
ned t
h
an a
n
y
ot
he
r e
dges
i
n
t
h
e v
e
hi
cl
e
im
age. The
v
e
rt
i
cal
and
ho
r
i
zont
al
hi
st
o
g
ram
of t
h
e sha
r
pe
ne
d i
m
age i
s
o
b
t
a
i
n
ed
an
d
whi
c
h i
s
h
a
s de
pi
ct
ed i
n
t
h
e Fi
gu
re
5.
All th
e reg
i
ons in
th
e im
ag
e
are
p
r
o
cessed u
s
in
g
ro
w
-
wi
se an
d c
o
l
u
m
n
-wi
s
e t
o
fi
nd
a com
m
on
regi
on
havi
ng
m
a
xim
u
m
hori
z
ont
al
an
d ve
rt
i
cal
hi
st
ogram
val
u
e. T
h
e c
o
m
m
on regi
o
n
i
s
t
h
e regi
on
w
h
ere t
h
e
ho
ri
zo
nt
al
an
d
vert
i
cal
hi
st
og
r
a
m
freq
u
e
n
ci
es
m
a
tches or i
n
tersects
with
one anothe
r.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Au
toma
tic Vehicle Tra
cking
System
Ba
sed
on
Fixed
Th
resho
l
d
i
ng
an
d H
i
sto
g
r
a
m
Ba
sed
… (N. S
hobh
a
Ran
i
)
87
3
Fi
gu
re
3.
I
n
p
u
t
dat
a
set
s
fo
r a
u
t
o
m
a
t
i
c
num
ber
pl
at
e ext
r
act
i
o
n
an
d
rec
o
g
n
i
t
i
o
n
(
a
)
Origi
n
al Im
age
(
b
)
Gray scale
Im
age
Fi
gu
re
4.
O
u
t
p
ut
o
f
pre
-
pr
oce
ssi
ng
st
age
The c
o
m
m
on regi
o
n
w
h
i
c
h i
s
obt
ai
ned
usi
n
g hi
st
og
ram
m
a
t
c
hi
ng
[
5
]
t
echni
que i
s
c
o
nsi
d
ere
d
as
regi
o
n
wi
t
h
h
i
gh
est
p
r
o
b
a
bilit
y o
f
con
t
ainin
g
a licen
se
p
l
ate n
u
m
b
e
r.
Furt
her the e
x
traction of com
m
on region in the ve
hicle image is accom
p
lish
ed through aut
o
m
a
tic
im
age cro
p
p
i
n
g
usi
n
g a
fi
xe
d
ra
nge
m
e
t
hod.
Fi
xe
d R
a
nge
[
14]
i
s
a
m
e
t
hod
by
w
h
i
c
h
we
det
e
rm
i
n
e t
h
e
ran
g
e
bl
oc
k
whi
c
h c
ont
ai
n
s
t
h
e
sp
an
of c
o
nve
r
g
ence
regi
ons
wi
t
h
r
e
spect
t
o
bot
h t
h
e
h
o
r
i
z
o
n
t
a
l
an
d
v
e
rt
i
cal
h
i
stog
ram
s
b
y
ap
p
e
nd
ing
with
sp
an
o
f
ad
d
ition
a
l rows an
d
co
lu
m
n
s in
to
p
,
b
o
ttom
,
left an
d
rig
h
t
of
con
v
e
r
ge
nce re
gi
o
n
i
d
e
n
t
i
f
i
e
d
wi
t
h
respect
t
o
a
t
h
res
h
ol
d val
u
e ‘N
1’
an
d ‘
N
2’
.
3.2.1 Determi
nati
on of thre
shold value ‘N1’
and ‘N2’:
The t
h
res
hol
d
val
u
e
‘N
1
’
an
d
‘N
2’
use
d
i
n
t
h
e fi
xed
ra
nge
m
e
t
hod i
s
a
val
u
e w
h
i
c
h i
s
det
e
rm
i
n
ed by
appl
y
i
n
g
di
f
f
er
ences o
p
er
at
i
o
n bet
w
ee
n t
h
e
pi
xel
i
n
t
e
ns
ity v
a
lu
es av
ailab
l
e with
in
th
e con
v
e
rg
en
ce
reg
i
o
n
of
the im
age. Consider a pi
xel p1,
p2, p3
…
pn i
n
the converge
nce re
gion of a
gray scale im
a
g
e with
respe
c
t to a
ro
w o
r
col
u
m
n
. The di
ffe
renc
e bet
w
ee
n t
h
e
pi
xel
pi
-
pi
+1
whe
r
e i
=
1,
2,
3…
n i
s
com
p
u
t
ed wi
t
h
res
p
e
c
t
t
o
each pair of
a
d
jace
nt pixels in
the
c
o
nvergence regi
on
and st
ore
d
in a
row vect
or
and similarly the colum
n
wi
se i
n
t
e
nsi
t
y
di
ffe
re
nces a
r
e
al
so
cal
cul
a
t
e
d a
n
d st
o
r
e
d
i
n
a c
o
l
u
m
n
vect
or
. T
h
e m
i
nim
u
m
di
ffere
nce
val
u
e
M
i
n(r
)
an
d m
a
xi
m
u
m
di
ffere
nce val
u
e M
a
x
(
r
)
i
s
det
e
rm
i
n
ed f
r
om
t
h
e ro
w vect
or a
n
d
M
i
n(c) a
n
d M
a
x(c
)
ar
e
determ
ined from the colum
n
vect
o
r
. T
h
e
m
i
nim
u
m
and m
a
xim
u
m
di
ff
er
ences
obtained from
the row and
col
u
m
n
vect
or
can be acc
or
de
d as t
h
e m
i
nimum
fi
xed t
h
res
hol
d an
d m
a
xim
u
m
fi
xed t
h
r
e
sh
ol
d. T
h
e t
h
r
e
sh
ol
d
val
u
e ‘
N
1’ an
d
‘N
2’ i
s
em
pi
rical
l
y
gi
ven by
avera
g
e o
f
M
i
n
(
r
)
, M
a
x
(
r
)
, M
i
n(c
)
an
d M
a
x(
c) wi
t
h
res
p
ect
t
o
i
t
s
r
o
w
and
co
lu
mn
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
J
ECE
Vo
l. 5
,
N
o
. 4
,
Aug
u
s
t 2
015
:
86
9
–
87
8
87
4
i.e
., N
1
=
Min(
r)
+ Mi
n(
c)
2
i.e
.
, N2 =
Max(
r)
+ M
a
x(
c)
2
The
Fi
g
u
re
6
s
h
o
w
s t
h
e s
n
a
p
s
hot
o
f
t
h
e
f
r
eq
uenci
e
s
vai
l
a
bl
e wi
t
h
i
n
t
h
e
co
nve
r
g
ence
re
gi
on
o
f
t
h
e i
m
age.
Fi
g
u
re
5.
H
o
ri
z
ont
al
a
n
d
Ve
rt
i
cal
hi
s
t
og
ram
s
of Fi
g
u
re
4
Fi
gu
re 6.
Fre
q
uenci
e
s
i
n
t
h
e con
v
e
r
ge
nce re
gi
o
n
of
i
m
age
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Au
toma
tic Vehicle Tra
cking
System
Ba
sed
on
Fixed
Th
resho
l
d
i
ng
an
d H
i
sto
g
r
a
m
Ba
sed
… (N. S
hobh
a
Ran
i
)
87
5
3.
3 E
x
tr
acti
on
o
f
l
i
cense pl
at
e regi
o
n
In
itially th
e g
r
ay scale i
m
a
g
e is pro
c
essed
horiz
on
tally
to
d
e
term
in
e th
e ho
rizo
n
t
al ed
g
e
s
by
traversing each and eve
r
y row
by cal
culating the di
ffe
rence eve
r
y two adj
ace
nt pixels and if di
fference
b
e
tween
p
i
x
e
l in
ten
s
ities d
o
e
s n
o
t
falls with
in th
e th
resho
l
d
N1
an
d
N2
th
en
, tho
s
e p
i
x
e
ls
will b
e
m
a
rk
ed with
‘0’ in
ten
s
ity(b
l
ack
). Th
is
p
r
o
c
essed
will b
e
co
n
tin
u
e
d
u
n
til
all th
e ro
ws are exh
a
usted
i
n
t
h
e im
ag
e.
To
d
e
term
in
e th
e v
e
rtical ed
ges in
th
e p
r
o
b
ab
le
license plate region, eac
h colum
n
is traverse
d and
colum
n
wise
differe
nces
are
com
puted,
if t
h
e differe
n
ce
be
tween any t
w
o ad
j
acen
t
p
i
x
e
l
s
do
es no
t
fall with
in
th
e thresho
l
d ‘N1’ and
‘N2
’
th
en
t
h
ose
p
i
xels are m
a
rk
ed
with
‘0’
in
t
e
n
s
ity(b
lack). Th
is p
r
o
cess will
b
e
co
n
tinu
e
d
till all th
e co
l
u
m
n
s in
th
e im
ag
e ex
h
a
u
s
ts. Fu
rt
h
e
r th
e resu
ltan
t
i
m
ag
e is sm
o
o
t
h
e
n
e
d b
y
app
l
yin
g
a
sm
o
o
t
h
i
n
g
filter [6
] to
elimin
ate th
e d
i
stin
ctly
m
a
rk
ed
p
i
x
e
l
s
wh
ich
are con
s
id
ered
as unwan
ted
reg
i
on
s. The
Fi
gu
re
7
depi
ct
s t
h
e res
u
l
t
s
o
b
t
a
i
n
ed
aft
e
r h
o
r
i
z
o
n
t
a
l
and
ve
rt
i
cal
ed
ge
pr
ocessi
n
g
usi
n
g
fi
xe
d t
h
resh
ol
d
m
e
t
hod.
Figure
7.
Out
p
ut
o
f
H
o
ri
zo
nt
a
l
and
ve
rt
i
cal
edge
p
r
ocessi
n
g
usi
n
g
fi
xe
d t
h
r
e
sh
ol
d
O
u
r
n
e
x
t
step
is ex
tr
actio
n
o
f
licen
se p
l
ate r
e
g
i
on
f
r
o
m
th
e r
e
su
ltan
t
im
ag
e. I
n
th
e pr
oposed
syste
m
, w
e
f
i
r
s
t
crop
th
e im
ag
e
h
o
rizon
t
ally a
n
d
th
en
v
e
rtically. In
h
o
ri
zo
nt
al
crop
pi
n
g
we
pr
oce
ss th
e imag
e m
a
trix
co
lu
m
n
-
wise an
d co
m
p
are its horizon
tal h
i
stog
ram
valu
e with
t
h
e
c
o
m
m
on regi
on
fre
q
u
ency
val
u
es.
I
f
cert
a
i
n
val
u
e
i
n
t
h
e h
o
r
i
z
o
n
t
a
l
hi
st
og
ram
i
s
m
o
re t
h
a
n
t
h
re
shol
d,
we m
a
ke i
t
as our st
a
r
t
i
ng
poi
nt
f
o
r c
r
op
pi
n
g
a
nd c
o
nt
i
n
u
e
u
n
til th
resh
o
l
d v
a
lu
e w
e
find less th
an
th
at
is o
u
r
end
po
i
n
t. In
th
is pro
c
ess w
e
en
counter
m
a
n
y
areas w
h
ich
have
val
u
e m
o
re t
h
an t
h
resh
o
l
d. So
we st
or
e are st
art
i
ng a
nd e
nd
poi
nt
i
n
t
h
e m
a
t
r
i
x
and c
o
m
p
are wi
dt
h o
f
each area; width is calculated differe
n
ce of start and en
d
poi
nts. The
n
we crop im
age horizontally by using
t
h
e st
art
a
nd e
n
d
p
o
i
n
t
.
Thi
s
new
h
o
ri
z
ont
al
l
y
crop
pe
d i
m
age i
s
pr
ocess
e
d f
o
r ve
rt
i
cal
cro
p
p
i
n
g.
In
v
e
rt
i
cal
cro
p
p
i
n
g we
u
s
e t
h
e sam
e
t
h
resh
ol
d c
o
m
p
ari
s
on m
e
t
hod
, b
u
t th
e
o
n
l
y d
i
fferen
ce is t
h
at th
is ti
m
e
we p
r
o
c
ess
i
m
ag
e
m
a
trix
ro
w-wise and
co
m
p
are th
resho
l
d
v
a
lu
e
with
vert
i
cal
hi
st
o
g
r
a
m
val
u
es.
Ag
ai
n we
get
set
vert
i
cal
st
art
an
d e
nd
poi
nt
,
we fi
nd
t
h
at
set
w
h
i
c
h m
a
p l
a
rgest
hei
g
ht
a
nd c
r
o
p
t
h
e
i
m
age. A
f
t
e
r
vert
i
c
a
l
an
d
ho
ri
zo
nt
al
cro
ppi
ng
we get
exact area of
num
ber plate from
original
i
m
age in RGB form
at. The extracted
license plate
re
gion is
as sh
own
in Figur
e
8
.
Fi
gu
re 8. Ext
r
act
e
d
n
u
m
ber
pl
at
e
re
gi
on
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
J
ECE
Vo
l. 5
,
N
o
. 4
,
Aug
u
s
t 2
015
:
86
9
–
87
8
87
6
3.
4
Reco
gni
t
i
o
n o
f
L
i
cense
P
l
ate
The e
x
tracte
d
num
ber
plate is proces
sed again th
e sta
g
e
3 to rec
ognize the c
h
aracters e
m
bossed
on
the license pla
t
e. The e
x
tract
ed num
b
er pla
t
e im
age from
stage 2 is
bina
rized [2] and t
h
en
segm
entation
of
t
h
e chara
c
t
e
rs
i
n
t
h
e n
u
m
b
er pl
at
e i
s
per
f
o
rm
ed usi
n
g v
e
rt
i
cal
pro
j
ect
i
on
pr
o
f
i
l
e
s [3]
and
bo
u
ndi
ng
bo
x
app
r
oach
[
5
]
.
The
Fi
g
u
re
9 a
n
d
fi
gu
re
10
s
h
ows
t
h
e
o
u
t
p
ut
s o
f
bi
na
ri
zat
i
on a
n
d
se
gm
entat
i
on.
Figure
9. Binarize
d im
age
Fi
g
u
r
e
10
.
Segm
ent
e
d c
h
a
r
act
ers
di
spl
a
y
e
d i
n
o
u
t
p
ut
fol
d
er
Fu
rt
h
e
r tem
p
late
m
a
tch
i
n
g
[11
]
b
a
sed
classificatio
n
is u
s
ed
for recog
n
ition
of indiv
i
d
u
a
l alph
anu
m
eric
characte
r
.
In t
e
m
p
late based algorithm
,
segm
ented im
ag
e is co
m
p
ared
with
th
e im
ag
es which a
r
e s
t
ore
d
i
n
d
a
tab
a
se.
Duri
n
g
classificatio
n, th
e test imag
e
for
wh
ich th
e co
rrelation
co
efficien
t
fo
r tem
p
late i
m
ag
e is
maxim
u
m
,
that im
age is c
onsi
d
ere
d
as
a
best
match. Th
e
eac
h a
n
d e
v
ery c
h
aracter classifi
ed is
post
proc
essed
and di
spl
a
y
e
d i
n
edi
t
a
bl
e
fo
r
m
at
i
n
a notepad. The Figure 11(a), Figure
1
1
(
b)
, Figu
re 1
1
(
c)
pr
ov
ides th
e
r
ecogn
itio
n
r
e
su
lts
and
Figu
r
e
12
p
r
ov
id
es
t
h
e rec
o
gnized c
h
aracters in a l
o
g file.
Figu
re 1
1
. (a)
Out
put o
f
GU
I interface
Figu
r
e
11
.
(b
) O
u
t
p
u
t
of
G
U
I
in
ter
f
ace
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Au
toma
tic Vehicle Tra
cking
System
Ba
sed
on
Fixed
Th
resho
l
d
i
ng
an
d H
i
sto
g
r
a
m
Ba
sed
… (N. S
hobh
a
Ran
i
)
87
7
Figure
11. (c
) Out
put
of GUI interface
Fi
gure
12. L
o
g
file of rec
o
gnize
d
c
h
aracte
r
s
4.
R
E
SU
LTS AN
D ANA
LY
SIS
Th
e propo
sed alg
o
rith
m
was tested
with 6
5
in
pu
t
im
ag
es having
differe
n
t
res
o
lutions
and ca
ptured at
d
i
fferen
t
illu
min
a
tio
n
co
nd
itio
n
s
. Th
e im
ag
es con
t
ain
e
d
veh
i
cles o
f
d
i
fferen
t
co
lo
rs and
in
clud
es
fron
t and
rear
view
of the car. T
h
e
front im
ages
of ca
r tested are a
r
ound
45 and
rea
r
im
ages of a
b
out
20 are test
ed a
nd
100% acc
urac
y is obtained for the front
images of car,
where as rear im
a
g
es has
give
n an accuracy of
98% in
reco
g
n
i
t
i
on
of
num
ber
pl
at
e. The al
g
o
ri
t
h
m
i
s
fl
exi
b
l
e
t
o
wo
rk
with
all
ty
p
e
s o
f
im
ages. The acc
uracy
in the
pr
o
pose
d
sy
st
e
m
co
m
p
l
e
t
e
l
y
depe
n
d
s
up
o
n
t
h
e hi
st
o
g
r
am
pr
ocessi
ng i
n
i
d
ent
i
f
i
cat
i
o
n
o
f
co
n
v
er
ge
nce
regi
on
wi
t
h
res
p
ect
t
o
b
o
t
h
h
o
r
i
z
ont
al
an
d ve
r
t
i
cal
hi
st
ogra
m
. The sobel
edge
det
ect
i
on
fo
r t
h
e c
u
r
r
en
t
expe
ri
m
e
nt
ai
on ha
d
pr
o
duce
d
ve
ry
g
o
od
o
u
t
c
om
es, whi
c
h as re
sul
t
l
ead t
o
war
d
s a
n
err
o
r
free
rec
o
gni
t
i
o
n
p
r
o
cess of
num
b
e
r p
l
ate reco
gn
itio
n.. Th
e efficien
cy of
th
e system
is d
e
fin
e
d
as nu
m
b
er of test
i
m
ag
es
recogn
ized
correctly to
th
e to
tal nu
m
b
er of im
ages considere
d
for test
i
n
g. T
h
e al
gori
thm
has achie
ved an
ove
rall accurac
y
of 99%
with
all the im
ages tested.
5.
CO
NCL
USI
O
N
The det
ect
i
o
n and e
x
t
r
act
i
o
n of l
i
cense
pl
at
e regi
o
n
of car i
m
ag
es is o
n
e
th
e ch
allen
g
i
ng
task
in
the
pr
ocess o
f
l
i
cense pl
at
e reco
g
n
i
t
i
on. T
h
e hi
st
og
ram
based
pro
cessing
for th
e id
en
tification
o
f
p
r
o
b
ab
le licen
se
pl
at
e regi
on i
s
core
ope
rat
i
o
n
i
n
t
h
e p
r
o
p
o
se
d sy
st
em
. The
det
e
rm
i
n
at
i
on of
fi
xe
d t
h
res
h
ol
d
N1 a
nd
N
2
usi
n
g
adjace
nt pixel
diffe
re
nces is at the
heart
of t
h
e pr
o
p
o
s
ed sy
st
em
usi
ng wh
i
c
h we can ext
r
act the license
plate
reg
i
o
n
i
n
th
e
i
m
ag
e. Th
e al
g
o
rith
m
can
be furth
e
r ex
ten
d
e
d
for th
e reco
gn
itio
n of
licen
se p
l
ates
in
two
wheele
r
ve
hicle im
ages. The
algorithm
can also be enhanced further to
recognize lice
n
se plate of images
captured at longer focal lengt
h
s.
REFERE
NC
ES
[1]
Sarbjit Kaur
, S
ukhvir Kaur, "An Efficient
Approach for
Num
b
er Plate
Extr
act
ion from
Vehicl
es Im
age un
der
Image Processin
g
", Intern
ational Journal of Co
mputer
Scien
ce
and Information
Tec
hnologies,
Vol. 5 (3)
, 201
4
,
2954-2959.
[2]
Serkan Ozba
y
,
Ergun Erce
leb
i
, "Autom
atic Vehic
l
e Ident
i
fi
ca
tion b
y
Pla
t
e R
ecogni
tion", W
o
rld acad
em
y
o
f
Science, Engin
eering and Technolog
y
In
ter
n
ation
a
l
Journal of Electr
i
cal, Computer, Electronics and
Communication
Engineering Vol:1 No:9, 2007.
[3]
D.G. Bail
e
y
, D.
Irecki
,
B.K. L
i
m
and
L. Yang "Tes
t bed for num
ber plat
e recogn
iti
on applic
ations
",
P
r
oceedings
of
First IEEE Inter
n
ation
a
l W
o
rkshop on Electroni
c Design,
Test a
nd Applications
( DELTA'
02 ), IEEE Com
puter
Society
,
2002
.
[4]
A.S. Johnson, B.M., "Bird
,
1
990,
"Number-plate matching
for Automa
tic Vehicle Iden
tification
,
" IEEE
Colloquium on
Electronic Imag
e an
d
Image Processing in
Secur
ity
and
Forensic,
April, 1990
.
[5]
Naikur Bharat Kumar Gohil, "Car
license plate detection", Dharmsinh
Desai University
"
,
India, 200
6.
[6]
Rinku Solanki,
Rajesh Kumar R
a
i,
Teena R
a
ikw
a
r, "Th
e
au
tomatic licens
e
plate
r
ecognition", In
ternation
a
l journ
a
l
of engin
eer
ing r
e
search
and
app
l
ications,
ISSN
: 2248-9622, vol. 3
,
Issue 6
,
Nov-D
ec 2013
, pp
304-
310.
[7]
Kum
a
r P
a
ras
u
ram
a
n, P
.
Vas
a
n
t
ha Kum
a
r, "An
effi
cien
t m
e
tho
d
of Indi
an veh
i
cl
e l
i
cens
e
pl
at
e ex
trac
tion
and
chara
c
t
e
r segm
e
n
tation"
, IE
EE
Interna
tiona
l co
nferenc
e
on
co
m
putational
int
e
lligen
ce
and co
m
puting resear
c
h
,
2010.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
J
ECE
Vo
l. 5
,
N
o
. 4
,
Aug
u
s
t 2
015
:
86
9
–
87
8
87
8
[8]
S
a
ndra S
i
van
a
n
d
an, As
hwini
Dhanait
,
Yogit
a
Dhepal
e,
"Aut
om
atic v
e
hic
l
e
identif
ic
ation
using li
cense p
l
a
t
e
recogni
tion for Indian vehi
cles"
,
IJCA proceedi
ngs on
e
m
erging trends in com
puter science
and Inform
ation
techno
log
y
, 201
2.
[9]
Chetan S
h
arm
a
,
Am
ardeep Kaur,
"Indian Vehi
cle
lic
ens
e
Plat
e Ex
trac
tion and seg
m
entation",
Inte
rnation
a
l journ
a
l
of Com
puter Sci
e
nce
and
Com
m
unica
tion,
Jul
y
-
D
ecem
ber 2011
, Vol. 2, pp. 593-
599.
[10]
Sheetal Mithun
Kawade, M.M
.
Mukhedkar,
"A Real time
veh
i
cle'
s license
plate recogniti
on s
y
s
t
em", Internation
a
l
journal of Scien
ce
and
Engineering, Vol. 1(2)
, pp
. 41-48
.
[11]
Pratishtha Gupta, G. N. Purohit,
M
a
nis
h
a Ra
thor
e, "Num
ber P
l
at
e extr
ac
ti
on usin
g Template ma
tching technique"
,
International
jou
r
nal of
Computer applica
tions, ISSN: 0975-8887,
Vol. 8(3)
, Febru
a
r
y
2014.
[12]
Ronak P Patel,
Narendra M Patel, Key
u
r Br
ahmbhatt, "Automatic
License Plate
Recognition"
, In
ternational jOur
nal
of Computer Science and
Mobile Computing, Apr
il 2013
.
[13]
A. Akoum, B. Day
a
, P. Chauv
e
t, "A New app
r
oach for
detection and iden
tif
ication of vehicl
e plate numbers",
Journal of sof
t
ware
engin
eering
and app
licat
ions, Vol. 3(2)
, pp
. 9
9
-108, Febru
a
r
y
2010.
[14]
Edward R Doughert, "Digital Image
Processing
Methods", CRC Pr
ess, 12-Jan-1994 - Techno
log
y
& Engin
eering
-
504 pages.
BIOGRAP
HI
ES
OF AUTH
ORS
N. Shobha Rani has complete
d M.Sc (CS) and
currently
pursu
ing her Ph.D degree in
Computer Scien
ce and
technolo
g
y
at Mah
a
ra
ja
Research
Foundation
,
MIT, M
y
sore and
working as
a facul
t
y
in th
e departm
e
nt
of
Computer Science at Amrita Vishwa
Vid
y
ap
ee
tham
,
M
y
sore Cam
pus. Her ar
ea of
i
n
terests is Opti
cal
char
act
er re
cognition
,
Computer vision
and Do
cument image processing.
Neethu O.P. is currently
pursuin
g her Master’s degree in Com
puter applications (MCA) i
n
department of C
o
mputer Scien
c
e at Amrita
Vishwa Vidy
apeetham
, M
y
s
o
re Cam
pus
. H
e
r
area
of inte
rests is
Com
puter
vision and
image p
r
ocessing.
Neethu O.P. is currently
pursuin
g her Master’s degree in Com
puter applications (MCA) i
n
department of C
o
mputer Scien
c
e at Amrita
Vishwa Vidy
apeetham
, M
y
s
o
re Cam
pus
. H
e
r
area
of inte
rests is
Com
puter
vision and
image p
r
ocessing.
Evaluation Warning : The document was created with Spire.PDF for Python.