I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
p
u
t
er
Science
Vo
l.
22
,
No
.
3
,
J
u
n
e
202
1
,
p
p
.
16
29
~
163
4
I
SS
N:
2
5
0
2
-
4
7
5
2
,
DOI
: 1
0
.
1
1
5
9
1
/i
j
ee
cs.v
2
2
.i
3
.
pp
16
29
-
163
4
1629
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs.ia
esco
r
e.
co
m
Cha
ra
cter
s r
ecog
nition us
ing
k
ey
s
po
ints and
c
o
nv
o
lutiona
l
neura
l net
w
o
rk
M
.
B
o
ut
o
un
t
e
1
, Y
.
O
ua
did
2
1
Na
ti
o
n
a
l
S
c
h
o
o
l
o
f
Bu
si
n
e
ss
a
n
d
M
a
n
a
g
e
m
e
n
t,
Ib
n
Z
o
h
r
Un
iv
e
rsit
y
,
Da
k
h
la,
M
o
ro
c
c
o
2
De
p
a
rtme
n
t
o
f
Co
m
p
u
ter S
c
ien
c
e
S
u
lt
a
n
,
F
a
c
u
lt
y
o
f
S
c
ien
c
e
a
n
d
T
e
c
h
n
iq
u
e
s,
L
a
b
o
ra
to
ry
o
f
In
f
o
rm
a
ti
o
n
P
r
o
c
e
ss
in
g
a
n
d
De
c
isio
n
a
n
d
S
u
p
p
o
rt,
M
o
u
lay
S
li
m
a
n
e
Un
iv
e
rsit
y
,
Be
n
i
-
M
e
ll
a
l
,
M
o
ro
c
c
o
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Feb
1
,
2
0
2
1
R
ev
i
s
ed
Mar
29
,
2
0
2
1
A
cc
ep
ted
A
p
r
6
,
2
0
2
1
In
t
h
is
p
a
p
e
r,
t
h
e
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
(CNN
)
is
u
se
d
i
n
o
r
d
e
r
t
o
d
e
sig
n
a
n
e
f
f
i
c
ien
t
o
p
ti
c
a
l
c
h
a
ra
c
ter
re
c
o
g
n
it
io
n
(OCR)
sy
ste
m
f
o
r
th
e
T
i
f
in
a
g
h
c
h
a
ra
c
ters
.
in
d
e
e
d
,
t
h
is
a
p
p
ro
a
c
h
h
a
s
p
ro
v
e
d
a
g
re
a
ter
e
ff
icie
n
c
y
b
y
g
iv
in
g
an
a
c
c
u
ra
c
y
o
f
9
9
%
,
th
is
a
p
p
ro
a
c
h
b
a
se
d
in
k
e
y
s
p
o
in
t
s
d
e
tec
ti
o
n
u
sin
g
Ha
rris
c
o
r
n
e
r
m
e
th
o
d
,
t
h
e
d
e
tec
ted
p
o
in
ts
a
re
a
u
t
o
m
a
ti
c
a
ll
y
a
d
d
e
d
t
o
th
e
o
rig
in
a
l
im
a
g
e
to
c
re
a
te
a
n
e
w
d
a
tab
a
se
c
o
m
p
a
re
d
to
th
e
b
a
sic
m
e
th
o
d
th
a
t
u
se
d
irec
tl
y
th
e
d
a
t
a
b
a
se
a
f
ter
a
p
re
p
ro
c
e
ss
in
g
ste
p
c
o
n
sistin
g
o
n
n
o
rm
a
li
z
a
ti
o
n
a
n
d
th
in
n
in
g
th
e
c
h
a
ra
c
ters
.
Us
in
g
th
is
m
e
th
o
d
,
w
e
c
a
n
b
e
n
e
f
it
f
ro
m
th
e
p
o
w
e
r
o
f
th
e
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
a
s
c
las
si
f
ier
in
im
a
g
e
th
a
t
h
a
s
a
lrea
d
y
th
e
f
e
a
tu
re
.
T
h
e
tes
t
w
a
s
p
e
rf
o
r
m
e
d
o
n
t
h
e
M
o
r
o
c
c
a
n
Ro
y
a
l
In
stit
u
te
o
f
Am
a
z
i
g
h
Cu
lt
u
re
(
IRCA
M
)
d
a
tab
a
se
c
o
m
p
o
se
d
o
f
3
3
0
0
0
c
h
a
ra
c
ters
o
f
d
iff
e
r
e
n
t
siz
e
a
n
d
st
y
le
w
h
a
t
p
re
se
n
t
t
h
e
d
if
f
icu
lt
y
,
th
e
k
e
y
s
p
o
i
n
ts
a
re
th
e
sa
m
e
in
th
e
p
rin
ted
a
n
d
h
a
n
d
w
rit
ten
c
h
a
ra
c
ters
so
th
is
m
e
th
o
d
c
a
n
b
e
a
p
p
ly
in
b
o
t
h
ty
p
e
w
it
h
so
m
e
m
o
d
if
ica
ti
o
n
s
.
K
ey
w
o
r
d
s
:
C
o
n
v
o
lu
tio
n
al
n
e
u
r
al
n
e
t
w
o
r
k
Har
r
is
co
r
n
er
Ke
y
p
o
in
t
T
if
in
ag
h
O
C
R
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Me
h
d
i B
o
u
tao
u
n
te
Natio
n
al
Sc
h
o
o
l o
f
B
u
s
i
n
es
s
a
n
d
Ma
n
a
g
e
m
e
n
t
I
b
n
Z
o
h
r
U
n
iv
er
s
it
y
Dak
h
la,
Mo
r
o
cc
o
E
m
ail: b
o
u
tao
u
n
te.
m
e
h
d
i@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
O
n
e
o
f
th
e
m
o
s
t
lead
in
g
f
ield
o
f
r
esear
ch
is
th
e
o
p
tical
ch
ar
ac
ter
r
ec
o
g
n
itio
n
,
h
o
w
ev
er
t
h
e
m
aj
o
r
it
y
o
f
r
esear
ch
ar
e
tar
g
eti
n
g
th
e
p
o
p
u
lar
lan
g
u
a
g
e
s
u
ch
as
o
t
h
er
s
lan
g
u
a
g
e
s
li
k
e
L
ati
n
an
d
A
r
ab
ic
a
n
d
ig
n
o
r
in
g
o
th
er
s
la
n
g
u
a
g
es
s
p
ea
k
i
n
g
b
y
m
i
n
o
r
it
y
l
ik
e
a
m
az
i
g
h
w
h
ic
h
i
s
a
h
o
m
ito
-
s
e
m
itic
l
an
g
u
ag
e
c
o
m
in
g
f
r
o
m
a
n
cie
n
t
B
er
b
er
lan
g
u
ag
e.
t
h
is
la
n
g
u
a
g
e
h
av
e
s
o
m
a
n
y
v
ar
iet
y
o
f
d
i
alec
ts
f
r
o
m
Mo
r
o
cc
o
,
L
ib
y
a,
E
g
y
p
t,
T
u
n
is
ia
an
d
Ma
li.
T
h
is
lan
g
u
a
g
e
is
s
p
o
k
en
to
d
ay
b
y
ab
o
u
t
1
4
m
i
llio
n
p
eo
p
le,
m
ai
n
l
y
i
n
th
e
Ma
g
h
r
eb
.
T
h
e
t
if
in
ag
h
ch
ar
ac
ter
s
is
n
o
r
m
al
l
y
w
r
it
ten
lef
t
to
r
i
g
h
t
a
n
d
f
r
o
m
to
p
to
b
o
tto
m
,
i
t
is
a
n
o
n
c
u
r
s
i
v
e
w
r
iti
n
g
s
i
m
p
li
f
y
i
n
g
t
h
e
s
eg
m
e
n
tatio
n
o
f
ch
ar
ac
ter
s
i
n
a
tex
t
i
m
ag
e
Am
az
i
g
h
.
b
ec
au
s
e
o
f
th
e
s
i
m
ilar
it
y
in
s
h
ap
e
o
f
th
e
ch
ar
ac
ter
s
an
d
th
e
t
h
er
e
n
u
m
b
er
(
3
3
ch
ar
ac
te
r
s
)
th
e
p
r
ep
r
o
ce
s
s
in
g
ti
m
e
i
s
v
er
y
h
i
g
h
,
t
h
o
s
e
ch
ar
ac
ter
s
p
r
e
s
en
t
a
c
h
alle
n
g
e
to
r
esear
ch
to
cr
ea
te
an
ef
f
icie
n
t
s
y
s
te
m
f
o
r
o
p
tical
ch
ar
ac
ter
r
ec
o
g
n
itio
n
.
I
n
th
is
cu
r
r
en
t
p
a
p
e
r
,
w
e
f
i
r
s
t
g
i
v
e
a
r
ev
i
ew
o
f
o
t
h
e
r
s
p
r
o
p
o
s
ed
t
if
in
ag
h
ch
a
r
a
c
t
e
r
s
r
e
c
o
g
n
i
ti
o
n
s
y
s
t
em
s
d
e
v
e
l
o
p
e
d
b
y
o
th
e
r
s
s
e
a
r
ch
e
r
s
s
u
ch
as
,
B
en
ch
a
r
ef
e
t
a
l
.
[
1
]
u
s
e
d
in
f
e
atu
r
e
ex
t
r
a
ct
i
o
n
th
e
r
i
em
an
n
i
an
m
e
t
r
i
c
t
h
es
e
m
e
t
r
i
c
d
e
s
c
r
i
p
t
o
r
s
a
r
e
r
e
li
a
b
l
e
t
o
w
a
r
d
ch
an
g
e
s
ca
l
e
.
B
o
u
t
a
o
u
n
t
e
e
t
a
l
.
[
2
]
a
r
e
u
s
in
g
th
e
s
im
p
l
e
s
t
p
o
s
s
i
b
l
e
g
e
o
m
e
t
r
i
c
e
l
em
en
t
s
(
s
eg
m
en
ts
a
n
d
k
ey
p
o
in
ts
)
t
o
c
r
e
a
te
th
e
g
r
a
p
h
an
d
ex
t
r
a
c
tin
g
th
e
im
p
a
c
t
m
a
t
r
ix
[
3
]
f
o
r
r
ec
o
g
n
itio
n
.
E
s
s
aa
d
y
et
a
l
.
[
4
]
w
h
o
g
i
v
e
a
p
r
o
p
o
s
itio
n
u
s
e
d
an
an
al
y
tical
m
et
h
o
d
u
s
ed
f
r
ee
m
a
n
co
d
in
g
f
o
r
f
ea
t
u
r
e
ex
tr
ac
tio
n
an
d
f
i
n
ite
au
to
m
at
io
n
in
r
ec
o
g
n
iti
o
n
s
te
p
.
Ay
ac
h
i
et
a
l
.
[
5
],
[
6
]
g
iv
ed
tw
o
m
e
th
o
d
s
f
o
r
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
22
,
No
.
3
,
J
u
n
e
2
0
2
1
:
16
29
-
1
6
3
4
1630
o
p
tical
ch
ar
ac
ter
r
ec
o
g
n
itio
n
(
OC
R
)
s
y
s
te
m
w
h
ich
u
s
ed
Mo
m
en
ts
in
v
ar
ia
n
ts
a
n
d
W
alsh
tr
an
s
f
o
r
m
f
o
r
f
ea
t
u
r
e
ex
tr
ac
tio
n
,
a
n
d
th
e
m
u
l
tila
y
er
n
eu
r
o
n
n
et
w
o
r
k
s
an
d
d
y
n
a
m
ic
p
r
o
g
r
am
m
i
n
g
f
o
r
th
e
cla
s
s
i
f
ic
atio
n
.
I
n
th
e
ir
w
o
r
k
n
eu
r
al
n
et
w
o
r
k
s
co
m
b
in
ed
w
it
h
w
als
h
tr
a
n
s
f
o
r
m
s
h
o
w
e
d
q
u
ite
i
n
ter
esti
n
g
r
es
u
lts
.
Ou
j
ao
u
r
a
et
a
l
.
[
7
]
p
r
o
p
o
s
ed
a
s
y
s
te
m
b
ased
o
n
t
h
r
ee
f
ea
t
u
r
es
ex
tr
ac
to
r
GI
ST
d
esc
r
ip
to
r
s
,
tex
tu
r
e,
w
al
s
h
,
an
d
b
ay
esia
n
n
et
w
o
r
k
s
in
t
h
e
clas
s
if
icatio
n
f
o
r
t
if
i
n
ag
h
c
h
ar
ac
ter
s
r
ec
o
g
n
itio
n
,
b
u
t
t
h
e
y
h
a
v
e
s
o
m
e
p
r
o
b
le
m
s
in
ter
m
o
f
ce
n
tr
al
p
r
o
ce
s
s
in
g
u
n
it
(
C
P
U
)
ti
m
e
an
d
r
ec
o
g
n
itio
n
r
ate.
Am
r
o
u
ch
et
a
l
.
[
8
]
u
s
ed
a
d
i
r
ec
tio
n
al
f
ea
t
u
r
e
v
ec
to
r
s
eq
u
en
ce
a
n
d
co
n
ti
n
u
o
u
s
h
id
d
en
m
ar
k
o
v
m
o
d
els
a
n
d
t
h
e
s
e
m
o
d
els
ar
e
n
o
t
a
f
a
v
o
r
itis
m
s
i
n
ce
ea
ch
h
id
d
en
m
ar
k
o
v
m
o
d
els
u
s
es
t
h
e
l
ea
r
n
in
g
o
f
a
s
i
n
g
le
ch
ar
ac
ter
,
Z
h
i
y
i
et
a
l
.
[
9
]
,
u
s
ed
Fu
zz
y
p
atter
n
r
ec
o
g
n
i
tio
n
i
n
w
h
ic
h
th
e
y
m
ai
n
l
y
cla
s
s
i
f
ied
ea
ch
ch
ar
ac
ter
s
b
y
m
et
h
o
d
s
a
cc
o
r
d
in
g
t
o
m
a
x
i
m
u
m
m
e
m
b
e
r
s
h
ip
w
h
ich
p
o
s
ed
a
p
r
o
b
lem
o
f
r
ec
o
g
n
i
tio
n
e
f
f
ec
ac
it
y
an
d
i
n
ter
m
s
o
f
p
r
o
ce
s
s
in
g
ti
m
e
b
ec
au
s
e
o
f
t
h
e
d
es
cr
ip
to
r
v
ec
to
r
s
ize.
B
o
u
tao
u
n
te
et
al
.
[
1
0
]
p
r
esen
ted
an
ap
p
r
o
ac
h
b
ased
o
n
ch
ar
ac
ter
s
d
ec
o
m
p
o
s
itio
n
in
to
m
u
ltip
les
g
eo
m
etr
i
c
s
h
ap
es
(
s
e
g
m
en
t,
ar
c)
b
y
d
ete
ctin
g
th
e
p
o
in
t
s
o
f
b
r
an
c
h
an
d
en
d
an
d
u
s
i
n
g
a
co
m
p
ar
i
s
o
n
b
et
w
ee
n
d
i
f
f
e
r
e
n
t
m
et
h
o
d
s
o
f
cla
s
s
i
f
icat
io
n
s
u
c
h
as
n
e
u
r
al
n
e
t
w
o
r
k
(
NN)
,
k
-
me
an
an
d
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
s
(
SVM
)
.
O
th
er
w
o
r
k
s
u
ch
as
W
a
r
d
an
i
,
[
1
1
]
a
n
d
A
h
a
r
r
a
n
e
e
t
al
.
[
1
2
]
a
r
e
b
a
s
ed
i
n
th
e
p
o
w
e
r
o
f
th
e
c
o
n
v
o
lu
t
io
n
a
l
n
eu
r
al
n
e
t
w
o
r
k
t
o
b
u
i
l
d
th
e
OC
R
s
y
s
tem
,
s
o
th
e
a
p
p
l
i
c
at
i
o
n
o
f
t
h
e
c
o
n
v
o
lu
t
i
o
n
a
l
n
eu
r
a
l
n
etw
o
r
k
(
C
N
N
)
o
r
th
e
d
ee
p
l
e
a
r
n
in
g
in
g
en
e
r
a
l
a
r
e
d
i
r
e
c
t
as
o
t
h
e
r
w
o
r
k
in
o
t
h
e
r
l
an
g
u
ag
e
a
s
th
e
w
o
r
k
o
f
F
i
n
jan
et
al
.
[
1
3
]
o
r
E
l
-
S
a
w
y
et
al
.
[
1
4
]
ap
p
lied
in
a
r
ab
ic
lan
g
u
a
g
e,
K
ad
ir
et
al
,
[
1
5
]
u
s
ed
th
e
C
NN
f
o
r
n
u
m
b
er
r
ec
o
g
n
i
tio
n
i
n
co
m
p
ar
is
o
n
w
it
h
b
ag
o
f
f
ea
t
u
r
es,
S
ad
o
u
k
et
al
.
[
1
6
]
u
s
ed
a
h
an
d
w
r
itte
n
d
atab
ase
b
u
t
th
e
r
ec
o
g
n
itio
n
r
ate
w
er
e
b
et
wee
n
9
5
%
an
d
9
8
%
b
u
t B
en
ad
d
y
et
al
.
[
1
7
]
,
u
s
ed
th
e
C
NN
a
s
a
f
ea
t
u
r
e
ex
tr
ac
to
r
an
d
ac
h
ie
v
e
a
r
a
te
o
f
9
9
%.
A
ll
t
h
e
r
ea
lized
w
o
r
k
s
ar
e
u
s
i
n
g
t
h
e
f
ea
t
u
r
es
ex
tr
ac
t
io
n
a
n
d
w
it
h
a
class
if
ier
lik
e
n
eu
r
al
n
et
w
o
r
k
o
r
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e,
o
th
er
s
ar
e
u
s
in
g
th
e
co
n
v
o
l
u
tio
n
a
l
n
eu
r
al
n
et
w
o
r
k
as
f
ea
t
u
r
e
ex
tr
ac
to
r
o
r
as
class
if
ier
th
at
g
a
v
e
g
o
o
d
r
esu
lts
,
b
u
t
th
e
p
r
o
b
lem
ar
e
th
e
ti
m
e
to
ex
tr
ac
t
th
e
f
ea
t
u
r
es
an
d
to
d
o
th
e
lean
in
g
p
h
ase
a
n
d
th
e
r
ate
o
f
r
ec
o
g
n
itio
n
th
a
t
w
h
y
w
e
c
h
o
s
e
to
cr
ea
te
a
n
e
w
m
e
th
o
d
th
at
u
s
e
d
ir
ec
tl
y
i
m
ag
e
w
it
h
k
e
y
s
p
o
i
n
ts
a
s
f
ea
t
u
r
es
an
d
t
h
e
p
o
w
er
o
f
C
NN
as
clas
s
i
f
ier
to
d
ec
r
ea
s
e
th
e
ti
m
e
a
n
d
in
cr
ea
s
e
th
e
r
ate.
T
h
is
p
ap
er
is
o
r
g
an
ized
as
f
o
llo
w
s
:
s
ec
tio
n
2
p
r
esen
ts
th
e
tech
n
iq
u
e
s
u
s
ed
to
d
etec
t
th
e
k
ey
s
p
o
in
ts
i
n
ea
ch
ch
ar
ac
ter
s
.
T
h
e
u
s
ed
C
NN
i
s
p
r
esen
ted
i
n
s
ec
tio
n
3
.
I
n
th
e
s
ec
tio
n
4
th
e
e
x
p
er
im
e
n
tal
r
es
u
lt
s
an
d
d
is
cu
s
s
io
n
s
ar
e
m
e
n
tio
n
ed
an
d
th
e
n
th
e
co
n
c
lu
s
io
n
.
2.
P
RO
P
O
SE
D
M
E
T
H
O
D
T
h
e
p
r
e
p
r
o
c
e
s
s
in
g
c
o
n
s
is
ts
o
n
b
i
n
a
r
i
z
a
ti
o
n
,
n
o
r
m
a
l
iz
a
t
i
o
n
a
n
d
th
in
n
in
g
[
1
8
]
t
h
e
im
a
g
e
i
n
o
r
d
e
r
t
o
r
e
d
u
c
e
th
e
n
u
m
b
e
r
o
f
d
a
t
a
u
s
ed
i
n
th
e
r
e
c
o
g
n
i
t
i
o
n
s
e
ct
i
o
n
b
y
k
e
e
p
in
g
o
n
ly
t
h
e
im
p
o
r
ta
n
t
in
f
o
r
m
a
ti
o
n
.
Sin
c
e
o
u
r
d
a
t
a
b
a
s
e
is
c
o
m
p
o
s
e
d
o
f
m
o
n
o
c
h
r
o
m
e
im
ag
es
.
T
h
e
h
a
r
r
is
m
et
h
o
d
[
1
9
]
a
r
e
u
s
e
d
t
o
d
e
t
e
c
t
th
e
k
ey
s
p
o
in
ts
w
h
i
ch
a
r
e
a
d
d
e
d
d
i
r
e
c
t
ly
t
o
th
e
im
ag
e
b
e
f
o
r
e
u
s
i
n
g
a
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
r
a
l
n
etw
o
r
k
(
C
N
N
)
[
2
0
]
t
o
c
la
s
s
if
y
th
em
.
2
.
1
.
No
rm
a
liza
t
io
n
a
nd
t
hin
nin
g
No
r
m
a
lizatio
n
i
s
a
f
ir
s
t
s
tep
o
f
p
r
e
-
p
r
o
ce
s
s
i
n
g
to
r
e
m
o
v
e
th
e
u
n
w
a
n
ted
ar
ea
an
d
p
u
t
t
h
e
ch
ar
ac
ter
i
n
th
e
ce
n
ter
o
f
t
h
e
i
m
a
g
e.
T
o
d
o
it,
th
e
v
er
tical
an
d
h
o
r
izo
n
tal
h
is
to
g
r
a
m
o
f
t
h
e
i
m
a
g
e
ar
e
ca
l
cu
lated
to
f
i
n
d
th
e
b
o
r
d
er
o
f
th
e
ch
ar
ac
ter
,
(
Fig
u
r
e
1
)
.
T
h
e
h
o
r
izo
n
tal
an
d
v
er
tical
h
is
to
g
r
a
m
ar
e
s
ca
n
n
ed
to
f
i
n
d
th
e
co
r
n
er
s
o
f
th
e
ch
ar
ac
ter
s
Fig
u
r
e
2
illu
s
tr
a
te
an
ex
a
m
p
le
o
f
c
h
ar
ac
ter
i
m
ag
e
n
o
r
m
a
lizatio
n
.
Fig
u
r
e
1
.
V
e
rti
c
a
l
a
n
d
h
o
rizo
n
tal
h
isto
g
ra
m
o
f
th
e
ima
g
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
C
h
a
r
a
cters
r
ec
o
g
n
itio
n
u
s
in
g
ke
ys p
o
in
ts
a
n
d
c
o
n
vo
l
u
tio
n
a
l
n
eu
r
a
l n
etw
o
r
k
(
M B
o
u
to
u
n
te
)
1631
D
u
e
t
o
i
ts
r
o
b
u
s
tn
es
s
an
d
s
p
ee
d
th
e
Z
h
an
g
an
d
W
an
g
[
2
1
]
a
l
g
o
r
ith
m
is
a
d
o
p
t
e
d
.
W
h
i
ch
is
a
p
a
r
a
l
l
el
a
l
g
o
r
ith
m
t
h
a
t
p
r
o
d
u
c
es
a
p
e
r
f
e
c
t
s
k
e
l
et
o
n
8
-
c
o
n
n
e
c
t
e
d
in
a
s
in
g
l
e
i
te
r
a
t
i
o
n
an
d
w
h
i
ch
o
p
e
r
a
t
e
s
th
e
c
o
ll
is
i
o
n
s
.
T
h
e
p
r
o
p
o
s
e
d
s
y
s
t
em
f
o
r
r
e
c
o
g
n
i
ti
o
n
is
m
a
d
e
u
p
o
f
th
r
e
e
m
a
in
p
h
as
e
s
in
F
ig
u
r
e
3
,
th
e
p
r
e
-
p
r
o
c
e
s
s
i
n
g
p
h
as
e
th
e
n
o
r
m
a
l
iz
a
t
i
o
n
an
d
t
h
in
n
in
g
,
th
e
i
s
o
l
a
ti
o
n
o
f
t
h
e
ch
a
r
a
ct
e
r
b
l
o
c
k
b
y
e
lim
in
a
t
i
o
n
o
f
u
n
u
s
e
d
in
f
o
r
m
a
t
i
o
n
an
d
t
h
en
t
h
e
ch
a
r
a
c
t
e
r
r
e
c
o
g
n
i
ti
o
n
b
l
o
c
k
.
I
n
o
th
e
r
w
o
r
d
s
,
th
e
p
r
e
p
r
o
c
e
s
s
i
n
g
s
t
e
p
,
f
e
a
tu
r
e
e
x
t
r
a
c
ti
o
n
an
d
t
h
e
r
ec
o
g
n
it
i
o
n
.
T
h
e
th
is
w
o
r
k
th
e
s
k
eleto
n
o
f
th
e
ch
ar
ac
ter
s
w
it
h
th
e
k
e
y
s
p
o
in
ts
is
p
r
esen
ted
in
th
e
f
ir
s
t
s
tep
,
an
d
th
er
ea
f
ter
it
w
ill
b
e
tr
an
s
f
o
r
m
ed
to
g
eo
m
e
tr
ic
s
i
m
p
le
s
h
ap
e
s
u
c
h
as
lin
es
o
r
ar
cs
[
2
2
]
.
T
h
e
g
o
al
is
to
p
r
o
v
id
e
a
s
tr
u
ct
u
r
a
l
d
escr
ip
tio
n
o
f
t
h
e
c
h
ar
ac
ter
.
First,
th
e
k
e
y
p
o
in
ts
o
f
th
e
c
h
ar
ac
ter
ar
e
ex
tr
ac
ted
u
s
i
n
g
h
ar
r
is
co
r
n
er
d
etec
to
r
in
Fig
u
r
e
4
.
Fig
u
r
e
2
.
R
esu
lt o
f
n
o
r
m
aliza
t
io
n
an
d
th
i
n
n
i
n
g
Fig
u
r
e
3
.
T
if
in
ag
h
ch
ar
ac
ter
r
ec
o
g
n
itio
n
s
y
s
te
m
Fig
u
r
e
4
.
Ke
y
p
o
in
ts
d
etec
ter
b
y
h
ar
r
is
m
eth
o
d
T
h
e
h
ar
r
is
alg
o
r
ith
m
is
an
i
m
p
r
o
v
e
m
e
n
t
o
f
t
h
e
m
o
r
av
ec
co
r
n
er
s
d
etec
to
r
[
2
3
]
.
T
h
e
p
r
in
cip
le
o
f
m
o
r
a
v
ec
alg
o
r
it
h
m
is
to
co
n
s
id
er
a
r
ec
tan
g
u
lar
lo
ca
l
w
in
d
o
w
in
t
h
e
i
m
ag
e
to
d
eter
m
i
n
e
th
e
av
er
ag
e
o
f
th
e
ch
an
g
es
i
n
th
e
i
n
te
n
s
it
y
i
n
im
ag
e
b
y
m
o
v
i
n
g
it
to
s
ev
er
al
d
ir
ec
tio
n
s
.
W
h
en
th
e
w
i
n
d
o
w
u
n
d
er
g
o
es
a
lar
g
e
s
eq
u
en
ce
o
f
c
h
an
g
e
t
h
e
m
in
i
m
u
m
o
f
it,
i
s
t
h
e
co
r
n
er
o
r
th
e
p
o
in
t
o
f
i
n
ter
est.
Ass
u
m
e
t
h
at
I
is
th
e
in
te
n
s
it
y
o
f
th
e
i
m
ag
e,
t
h
e
ch
a
n
g
in
g
E
p
r
o
d
u
ce
d
b
y
t
h
e
m
o
v
e
m
en
t (
x
,
y
)
is
g
i
v
e
n
b
y
:
,
=
∑
,
|
+
,
+
−
,
|
²
,
(
1
)
w
h
e
r
e
,
,
=
−
(
2
+
2
)
/
2
²
(
2
)
T
h
e
n
ex
t
s
tep
is
d
i
s
ti
n
ct
b
et
w
e
en
t
h
e
ar
c
an
d
s
e
g
m
e
n
t,
s
o
f
o
r
ea
ch
t
w
o
k
e
y
p
o
i
n
ts
t
h
at
ar
e
co
n
n
ec
ted
in
t
h
e
c
h
ar
ac
ter
,
a
s
e
g
m
en
t
is
d
r
a
w
i
n
g
b
et
w
ee
n
th
er
ea
f
ter
t
h
e
d
is
ta
n
ce
a
m
o
n
g
th
e
d
r
a
w
i
n
g
s
e
g
m
en
t
an
d
t
h
e
b
lack
p
ix
el
s
o
f
t
h
e
c
h
ar
ac
ter
t
h
at
co
n
n
ec
t
th
e
t
w
o
p
o
in
t
s
i
n
i
m
ag
e
is
ca
lcu
lated
.
I
f
th
e
d
i
s
t
an
ce
i
s
eq
u
al
to
0
it
m
ea
n
s
t
h
at
th
e
g
eo
m
e
tr
ic
s
h
a
p
es
b
et
w
ee
n
th
e
t
w
o
p
o
in
t
s
ar
e
a
s
i
m
p
le
s
e
g
m
en
t,
if
n
o
t
t
wo
p
o
s
s
ib
ly
ca
s
es
ca
n
b
e,
f
ir
s
t
a
s
i
m
p
le
g
eo
m
etr
ica
l
s
h
ap
e
in
f
o
r
m
o
f
ar
c
co
n
n
ec
te
d
th
e
m
o
r
m
o
r
e
th
a
n
o
n
e
co
n
n
ec
ted
s
eg
m
e
n
t
d
u
e
to
an
er
r
o
r
in
th
e
m
et
h
o
d
o
f
s
e
g
m
e
n
tat
io
n
i
n
Fi
g
u
r
e
5
.
I
n
o
r
d
e
r
t
o
r
e
s
o
l
v
e
th
i
s
is
s
u
e
,
t
h
e
an
g
l
e
Y
b
e
tw
ee
n
th
e
t
w
o
k
ey
p
o
in
ts
a
n
d
th
e
p
o
in
t
th
a
t
h
av
e
th
e
b
i
g
g
es
t
d
is
t
an
c
e
b
etw
e
en
t
h
e
d
r
aw
in
g
s
eg
m
en
t
a
n
d
th
e
b
l
a
ck
l
i
n
e
th
a
t
c
o
n
n
e
ct
th
e
k
ey
p
o
in
ts
is
c
a
l
cu
l
a
t
e
d
an
d
w
e
s
et
a
th
r
esh
o
l
d
f
o
r
t
h
i
s
a
n
g
le
t
o
d
e
t
e
r
m
in
e
th
e
ty
p
e
o
f
th
a
t
g
e
o
m
e
t
r
i
c
s
h
a
p
es
.
T
h
e
F
ig
u
r
e
6
s
h
o
w
s
th
e
p
r
o
c
ess
o
f
p
r
o
b
l
em
d
e
t
ec
t
i
o
n
f
r
o
m
t
h
e
s
t
a
r
t
t
o
e
x
t
r
a
c
ti
n
g
th
e
k
ey
p
o
in
ts
an
d
al
s
o
t
h
e
s
im
p
l
e
g
e
o
m
et
r
i
c
s
h
a
p
e
s
[
2
4
]
.
T
h
e
in
f
o
r
m
at
i
o
n
c
o
l
l
e
c
te
d
in
th
i
s
s
t
e
p
w
il
l
b
e
u
s
e
d
t
o
h
el
p
th
e
C
NN
in
r
e
c
o
g
n
i
ti
o
n
s
t
e
p
,
a
c
o
m
p
a
r
is
o
n
w
as
d
o
n
e
b
etw
e
en
s
im
p
l
e
im
ag
e
s
an
d
o
th
e
r
s
w
ith
k
ey
s
p
o
in
t
s
t
o
d
et
e
r
m
in
e
th
e
u
til
i
ty
o
f
o
u
r
m
et
h
o
d
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
22
,
No
.
3
,
J
u
n
e
2
0
2
1
:
16
29
-
1
6
3
4
1632
Fig
u
r
e
5
.
Dete
r
m
i
n
atio
n
o
f
th
e
Y
an
g
le
to
d
etec
t
th
e
s
e
g
m
en
ta
tio
n
er
r
or
Fig
u
r
e
6
.
E
x
tr
ac
tin
g
k
e
y
p
o
in
t
s
an
d
s
i
m
p
le
g
eo
m
etr
ic
s
h
ap
e
3.
E
XP
E
R
I
M
E
NT
A
L
RE
SUL
T
S
W
e
test
ed
o
u
r
ap
p
r
o
ac
h
o
f
T
if
in
ag
h
c
h
ar
ac
ter
r
ec
o
g
n
itio
n
o
n
th
e
d
atab
ase
o
f
I
R
C
A
M.
T
h
is
d
atab
ase
in
cl
u
d
es
3
3
0
0
0
ch
ar
ac
ter
s
p
r
i
n
ted
in
d
if
f
er
en
t
s
izes
an
d
wr
itin
g
s
t
y
les.
I
n
Fi
g
u
r
e
7
ea
ch
ch
ar
ac
ter
w
ill
b
e
tr
ea
ted
to
d
etec
t
th
e
k
e
y
s
p
o
in
ts
u
s
i
n
g
h
ar
i
s
co
r
n
e
r
m
et
h
o
d
,
th
e
id
en
ti
f
icatio
n
e
f
f
ec
ted
b
y
a
co
n
v
o
l
u
tio
n
al
n
eu
r
al
n
et
w
o
r
k
.
T
o
im
p
r
o
v
e
t
h
e
ef
f
icie
n
c
y
o
f
o
u
r
m
et
h
o
d
,
w
e
u
s
e
a
C
NN
w
h
ite
t
w
o
co
n
v
o
lu
tio
n
la
y
er
s
an
d
2
p
o
o
lin
g
la
y
er
s
,
t
h
e
tr
ain
in
g
p
r
o
ce
s
s
is
p
er
f
o
r
m
ed
o
n
a
p
er
s
o
n
al
co
m
p
u
ter
(
PC
)
eq
u
ip
p
ed
w
it
h
I
n
tel
UH
D
Gr
ap
h
ics 6
2
0
an
d
a
p
r
o
ce
s
s
o
r
i7
8
th
g
en
er
atio
n
,
th
e
ac
cu
r
ac
y
o
b
tain
ed
o
f
tr
ai
n
in
g
a
n
d
tes
ti
n
g
is
s
h
o
w
ed
i
n
th
e
T
ab
le
1
.
Fig
u
r
e
7
.
Si
m
ilar
e
c
h
ar
ac
ter
T
ab
le
1
.
A
cc
u
r
ac
y
V
a
r
i
a
b
l
e
A
c
c
u
r
a
c
y
(
%)
L
o
ss (%)
S
i
mp
l
e
me
t
h
o
d
9
6
.
1
2
3
.
8
8
K
e
y
s p
o
i
n
t
me
t
h
o
d
9
9
.
3
9
0
.
6
1
T
h
e
o
l
d
m
et
h
o
d
th
at
u
s
es
f
ea
t
u
r
e
ex
tr
ac
tio
n
in
ti
f
i
n
ag
h
ch
ar
ac
ter
r
ec
o
g
n
itio
n
h
a
v
e
s
ev
er
al
p
r
o
b
lem
s
s
p
ec
iall
y
w
it
h
ch
ar
ac
ter
in
F
i
g
u
r
e
7
th
at
h
a
v
e
a
s
i
m
ilar
it
y
o
f
8
0
%
o
r
m
o
r
e,
an
d
to
r
eso
lv
e
th
e
p
r
o
b
lem
w
e
n
ee
d
an
ad
d
itio
n
al
tr
ea
tm
e
n
t
s
u
ch
as
o
u
r
p
r
ev
io
u
s
w
o
r
k
[
2
5
]
o
r
th
e
r
ec
o
g
n
itio
n
r
ate
w
ill
b
e
in
f
er
io
r
to
9
5
%.
I
n
th
is
p
ap
er
th
is
k
in
d
o
f
p
r
o
b
lem
w
as
s
o
l
v
ed
b
y
ad
d
in
g
k
e
y
p
o
in
ts
.
T
h
e
d
if
f
er
en
ce
b
et
w
ee
n
th
e
s
i
m
p
le
m
eth
o
d
an
d
th
e
n
e
w
m
e
th
o
d
u
s
in
g
t
h
e
k
e
y
p
o
in
t
s
is
clo
s
e
to
4
%
with
an
ac
c
u
r
ac
y
o
f
9
9
%.
T
h
e
Fig
u
r
e
s
8
an
d
9
ar
e
co
m
p
let
in
g
t
h
e
tab
le
an
d
s
h
o
win
g
t
h
e
tr
ai
n
i
n
g
p
r
o
g
r
ess
o
f
o
u
r
C
NN
u
s
in
g
t
h
e
t
w
o
m
et
h
o
d
s
.
Fig
u
r
e
8
.
A
cc
u
r
ac
y
a
n
d
lo
s
s
u
s
in
g
t
h
e
s
i
m
p
le
d
atab
ase
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
C
h
a
r
a
cters
r
ec
o
g
n
itio
n
u
s
in
g
ke
ys p
o
in
ts
a
n
d
c
o
n
vo
l
u
tio
n
a
l
n
eu
r
a
l n
etw
o
r
k
(
M B
o
u
to
u
n
te
)
1633
Fig
u
r
e
9
.
A
cc
u
r
ac
y
a
n
d
lo
s
s
u
s
in
g
t
h
e
k
e
y
s
p
o
in
t
s
d
atab
ase
4.
CO
NCLU
SI
O
N
I
n
th
i
s
w
o
r
k
w
e
p
r
esen
ted
a
n
e
w
tec
h
n
iq
u
e
f
o
r
T
if
in
ag
h
ch
ar
ac
ter
r
ec
o
g
n
itio
n
,
u
s
i
n
g
t
h
e
k
e
y
p
o
in
ts
an
d
th
e
g
eo
m
etr
ic
to
co
r
r
ec
t
th
e
p
o
in
ts
d
etec
tio
n
er
r
o
r
.
T
h
e
o
b
tain
ed
r
esu
lts
co
n
f
ir
m
s
th
at
w
e
h
a
v
e
i
m
p
le
m
en
ted
a
n
ef
f
icie
n
t
m
et
h
o
d
f
o
r
ch
ar
ac
ter
s
r
ec
o
g
n
it
io
n
th
at
ca
n
b
e
u
s
ed
w
ith
o
t
h
er
class
i
f
ier
,
in
f
u
tu
r
e
w
e
ca
n
ad
d
m
o
r
e
d
etails to
t
h
e
ch
ar
ac
ter
s
i
m
ag
e
to
i
m
p
r
o
v
e
t
h
e
ef
f
icie
n
c
y
o
f
t
h
e
C
NN,
s
u
c
h
as
k
e
y
p
o
i
n
ts
a
n
d
t
y
p
es
g
eo
m
e
tr
ic
s
h
ap
es
o
r
ien
t
atio
n
.
T
h
e
o
b
tain
ed
r
esu
lt
ar
e
en
co
u
r
ag
in
g
u
s
to
tr
y
t
h
is
m
e
t
h
o
d
o
n
h
an
d
w
r
itte
n
ch
ar
ac
ter
s
th
a
t n
ee
d
m
o
r
e
p
r
ep
r
o
ce
s
s
in
g
s
tep
s
.
R
E
F
E
R
E
NC
E
S
[1
]
O
.
Be
n
c
h
a
re
f
,
M
.
F
a
k
ir,
N
.
I
d
ri
ss
i,
B
.
Bo
u
ik
h
a
len
,
a
n
d
B
.
M
i
n
a
o
u
i
,
“
A
p
p
li
c
a
ti
o
n
o
f
Riem
a
n
n
ian
g
e
o
m
e
tr
y
to
T
i
f
in
a
g
h
c
h
a
ra
c
ter
re
c
o
g
n
it
io
n
i
n
F
ra
n
ç
a
ise
:
A
p
p
li
c
a
ti
o
n
d
e
la
g
é
o
m
é
tri
e
rie
m
a
n
n
ien
n
e
à
la
re
c
o
n
n
a
issa
n
c
e
d
e
s
c
a
ra
c
tèr
e
s T
i
f
in
a
g
h
,
”
T
h
ird
i
n
ter
n
a
ti
o
n
a
l
sy
mp
o
si
u
m o
n
A
u
to
m
a
ti
c
Ama
zig
h
p
r
o
c
e
ss
in
g
(
S
IT
ACA
M
’1
1
)
,
2
0
1
1
.
[2
]
M
.
B
o
u
ta
o
u
n
te
a
n
d
Y
.
O
u
a
d
i
d
;
“
Do
c
u
m
e
n
t
Im
a
g
e
Co
n
v
e
rter
to
XML
F
il
e
,
Ca
se
o
f
T
if
in
a
g
h
,
”
In
ter
n
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
Co
m
p
u
ter
A
p
p
l
ica
ti
o
n
s
, v
ol
.
1
4
5
,
n
o.
4,
p
p
.
3
0
-
3
3
,
J
u
l
.
2
0
1
6
,
d
o
i:
1
0
.
5
1
2
0
/i
jca
2
0
1
6
9
1
0
5
6
9
.
[3
]
D.
A
rriv
a
u
lt
,
“
Co
n
tri
b
u
ti
o
n
o
f
G
ra
p
h
s
to
Un
c
o
n
stra
i
n
e
d
Re
c
o
g
n
it
io
n
o
f
A
n
c
ien
t
M
a
n
u
sc
rip
t
Ch
a
ra
c
ters
in
F
ra
n
ç
a
ise
:
A
p
p
o
rt
d
e
s
G
ra
p
h
e
s
d
a
n
s
la
Re
c
o
n
n
a
issa
n
c
e
No
n
-
C
o
n
tra
in
te
d
e
Ca
ra
c
tère
s
M
a
n
u
sc
rit
s
A
n
c
ien
s
,
”
T
h
e
sis,
Un
iv
e
rsité d
e
P
o
it
iers
,
P
ra
n
c
is,
pp
.
6
1
,
2
0
0
2
.
[4
]
Y
.
Es
S
a
a
d
y
,
A
.
Ra
c
h
id
i,
M
.
El
Ya
ss
a
,
a
n
d
D
,
M
a
m
m
a
ss
,
“
P
ri
n
te
d
A
m
a
z
ig
h
Ch
a
ra
c
ter
Re
c
o
g
n
it
io
n
b
y
a
S
y
n
tac
ti
c
A
p
p
ro
a
c
h
u
sin
g
F
in
it
e
A
u
to
m
a
ta
,
”
J
o
u
rn
a
l
o
f
Gr
a
p
h
ics
,
Vi
si
o
n
a
n
d
Ima
g
e
Pr
o
c
e
ss
in
g
,
v
o
l
.
10
,
n
o
.
2
,
p
p
.
1
-
8
,
2
0
1
0
.
[5
]
R
.
El
Ay
a
c
h
i,
M
.
F
a
k
ir,
B
.
Bo
u
i
k
h
a
len
e
,
a
n
d
S
.
S
a
f
i
,
“
Off
li
n
e
p
rin
ted
a
m
a
z
ig
h
e
s
c
rip
ts
re
c
o
g
n
it
io
n
,
”
J
o
u
r
n
a
l
o
f
T
h
e
o
re
ti
c
a
l
a
n
d
A
p
p
li
e
d
I
n
fo
rm
a
t
io
n
T
e
c
h
n
o
lo
g
y
(
J
T
AIT
)
,
v
o
l.
20
,
n
o
.
2
,
2
0
1
0
.
[6
]
R
.
El
A
y
a
c
h
i,
M
.
F
a
k
ir
,
a
n
d
B
.
B
o
u
ik
h
a
len
e
,
“
Re
c
o
g
n
it
io
n
o
f
T
if
in
a
g
h
Ch
a
ra
c
ters
Us
in
g
D
y
n
a
m
ic
P
ro
g
ra
m
m
in
g
&
Ne
u
ra
l
Ne
tw
o
rk
,
”
Bo
o
k
o
f
Do
c
u
me
n
t
Rec
o
g
n
it
io
n
a
n
d
U
n
d
e
rs
ta
n
d
in
g
(
In
T
e
c
h
)
,
p
p
.
35
-
57
,
2
0
1
1
,
d
o
i:
1
0
.
5
7
7
2
/
1
6
2
7
8
.
[7
]
M
.
Ou
ja
o
u
ra
,
B
.
M
in
a
o
u
i
,
a
n
d
M
.
F
a
k
ir
,
“
W
a
lsh
,
T
e
x
tu
re
a
n
d
G
IS
T
De
sc
rip
to
rs
w
it
h
Ba
y
e
sia
n
Ne
tw
o
rk
s
f
o
r
Re
c
o
g
n
it
io
n
o
f
T
if
in
a
g
h
Ch
a
ra
c
ters
,
”
In
ter
n
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
Co
mp
u
ter
A
p
p
li
c
a
ti
o
n
s
(
IJ
CA)
,
v
ol
.
81
,
n
o
.
12
,
p
p
:
39
-
4
6
,
2
0
1
3
,
d
o
i
:
1
0
.
5
1
2
0
/
1
4
0
6
8
-
2
4
6
4
.
[8
]
M
.
Am
ro
u
c
h
,
Y
.
Es
-
sa
a
d
y
,
A
.
Ra
c
h
id
i,
M
.
El
Ya
ss
a
,
a
n
d
D
.
M
a
m
m
a
ss
,
“
Ha
n
d
w
rit
ten
Am
a
z
ig
h
Ch
a
ra
c
ter
Re
c
o
g
n
it
io
n
S
y
ste
m
Ba
se
d
o
n
Co
n
ti
n
u
o
u
s
HMM
s
a
n
d
Dire
c
ti
o
n
a
l
F
e
a
tu
re
s
,
”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
M
o
d
e
rn
En
g
i
n
e
e
rin
g
Res
e
a
rc
h
(
IJ
M
ER
),
v
ol
.
2
,
n
o
.
2
,
pp
.
4
3
6
-
4
4
1
,
2
0
1
2
.
[9
]
R
.
Zh
iy
i,
Z
.
Yin
g
,
H
.
Do
n
g
m
in
g
,
a
n
d
W
.
L
u
ro
n
g
,
“
V
is
u
a
li
z
a
ti
o
n
o
f
L
ice
n
se
P
late
Re
c
o
g
n
it
io
n
S
y
ste
m
,
”
In
d
o
n
e
sia
n
J
o
u
rn
a
l
o
f
E
lec
trica
l
E
n
g
i
n
e
e
rin
g
a
n
d
Co
m
p
u
ter
S
c
ien
c
e
(
IJ
EE
CS
)
,
v
o
l
.
11
,
n
o
.
11
,
p
p
:
6
7
1
4
-
6
7
2
1
,
2
0
1
3
,
d
o
i:
1
0
.
1
1
5
9
1
/t
e
lk
o
m
n
ik
a
.
v
1
1
i1
1
.
3
5
2
0
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
22
,
No
.
3
,
J
u
n
e
2
0
2
1
:
16
29
-
1
6
3
4
1634
[1
0
]
M
.
Bo
u
tao
u
n
te
a
n
d
Y
.
Ou
a
d
i
d
,
“
T
i
f
in
a
g
h
Ch
a
ra
c
ters
R
e
c
o
g
n
it
io
n
Us
in
g
S
im
p
le
Ge
o
m
e
tri
c
S
h
a
p
e
s
,
”
In
d
o
n
e
sia
n
J
o
u
rn
a
l
o
f
El
e
c
trica
l
E
n
g
in
e
e
rin
g
a
n
d
Co
m
p
u
ter
S
c
ien
c
e
(
IJ
EE
CS
)
,
v
o
l
3
,
n
o
.
1
,
p
p
:
2
3
5
-
2
3
9
,
2
0
1
6
,
d
o
i
:
1
0
.
1
1
5
9
1
/i
jee
c
s.v
3
.
i1
.
p
p
2
3
5
-
2
3
9
.
[1
1
]
N
.
A
h
a
rra
n
e
,
A
.
D
a
h
m
o
u
n
i
,
K
,
E
l
M
o
u
ta
o
u
a
k
il
,
a
n
d
K.
S
a
to
ri
,
“
P
r
in
ted
T
if
in
a
g
h
sc
rip
t
re
c
o
g
n
it
io
n
f
ro
m
W
e
b
a
n
d
n
a
tu
ra
l
sc
e
n
e
s im
a
g
e
s in
m
u
lt
il
in
g
u
a
l
e
n
v
iro
n
m
e
n
t,
”
T
ICAM
’1
8
,
N
o
v
.
2
0
1
8
.
[1
2
]
D
EL
W
a
rd
a
n
i
,
“
T
if
in
a
g
h
-
IRCAM
Ha
n
d
w
rit
ten
c
h
a
ra
c
ter
re
c
o
g
n
it
io
n
u
sin
g
De
e
p
lea
rn
in
g
,
”
a
rXi
v
:1
9
1
2
.
1
0
3
3
8
,
De
c
.
2
0
1
9
.
[1
3
]
F
.
R
.
Ha
sa
n
,
A
.
S
.
Ra
sh
e
e
d
,
A
A
.
Ha
sh
im
,
a
n
d
M
.
M
u
rt
d
h
a
,
“
A
ra
b
ic
h
a
n
d
w
rit
ten
d
ig
it
s
re
c
o
g
n
it
io
n
b
a
se
d
o
n
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
s
w
it
h
re
sn
e
t
-
3
4
m
o
d
e
l
,
”
In
d
o
n
e
sia
n
J
o
u
rn
a
l
o
f
El
e
c
trica
l
E
n
g
i
n
e
e
rin
g
a
n
d
C
o
mp
u
ter
S
c
ien
c
e
(
IJ
EE
CS
)
,
v
o
l.
2
1
,
n
o
.
1
,
p
p
.
1
7
4
-
1
7
8
,
Ja
n
.
2
0
2
1
,
d
o
i:
1
0
.
1
1
5
9
1
/i
jee
c
s.v
2
1
.
i1
.
p
p
1
7
4
-
1
7
8
.
[1
4
]
A
.
El
-
S
a
wy
,
M
.
L
o
e
y
a
n
d
H
.
EL
-
Ba
k
r
y
,
“
Ara
b
ic
Ha
n
d
w
rit
ten
Ch
a
ra
c
ters
R
e
c
o
g
n
it
io
n
u
sin
g
Co
n
v
o
lu
ti
o
n
a
l
Ne
u
ra
l
Ne
tw
o
rk
,
”
W
S
EA
S
T
ra
n
sa
c
ti
o
n
s
o
n
C
o
mp
u
ter
Res
e
a
rc
h
,
v
ol
.
5,
n
o
.
1
,
p
p
.
1
1
-
1
9
,
2
0
1
7
.
[1
5
]
N
.
H
.
M
.
Ka
d
ir
,
S
.
N
.
S
.
M
.
N
.
H
id
a
y
a
h
,
M
.
No
ra
sia
h
,
a
n
d
I
.
Zai
d
a
h
,
“
Co
m
p
a
riso
n
o
f
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
a
n
d
b
a
g
o
f
f
e
a
tu
re
s
f
o
r
m
u
lt
i
-
f
o
n
t
d
ig
it
re
c
o
g
n
it
i
o
n
,
”
In
d
o
n
e
sia
n
J
o
u
rn
a
l
o
f
El
e
c
trica
l
En
g
i
n
e
e
rin
g
a
n
d
C
o
mp
u
ter
S
c
ien
c
e
(
IJ
EE
CS
)
,
v
o
l.
1
5
,
n
o
.
3
,
p
p
.
1
3
2
2
-
1
3
2
8
,
2
0
1
9
,
d
o
i:
1
0
.
1
1
5
9
1
/i
jee
c
s.v
1
5
.
i3
.
p
p
1
3
2
2
-
1
3
2
8
.
[1
6
]
L
.
S
a
d
o
u
k
,
T
.
G
a
d
i
,
a
n
d
E
.
Ess
o
u
f
i
,
“
Ha
n
d
w
rit
ten
ti
f
in
a
g
h
c
h
a
ra
c
te
r
re
c
o
g
n
it
io
n
u
sin
g
d
e
e
p
lea
rn
i
n
g
a
rc
h
it
e
c
tu
re
s
,
”
Pro
c
e
e
d
in
g
s
o
f
t
h
e
1
st
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
I
n
ter
n
e
t
o
f
T
h
i
n
g
s
a
n
d
M
a
c
h
i
n
e
L
e
a
rn
in
g
,
2
0
1
7
,
p
p
.
1
-
11
.
[1
7
]
M
.
Be
n
a
d
d
y
,
O
.
El
M
e
slo
u
h
i
,
a
n
d
M
.
Ka
rd
o
u
c
h
i
,
“
Ha
n
d
w
rit
ten
T
if
in
a
g
h
Ch
a
ra
c
ters
Re
c
o
g
n
it
io
n
Us
in
g
De
e
p
Co
n
v
o
l
u
ti
o
n
a
l
Ne
u
ra
l
Ne
tw
o
rk
s
,
”
Co
mp
u
ter
S
c
ien
c
e
S
e
n
sin
g
a
n
d
Ima
g
in
g
(
CS
S
I)
,
M
a
r
.
2
0
1
9
,
d
o
i:
1
0
.
1
0
0
7
/s1
1
2
2
0
-
0
1
9
-
0
2
3
1
-
5
.
[1
8
]
R.
C.
G
o
n
z
a
lez
a
n
d
R.
E.
W
o
o
d
s,
Dig
it
a
l
Im
a
g
e
P
ro
c
e
ss
in
g
Us
in
g
M
AT
LAB
,
Ga
tes
m
a
rk
P
u
b
li
sh
i
n
g
,
Ja
n
.
2
0
2
0
.
[1
9
]
C.
G
.
H
a
rris
a
n
d
S
.
M
ik
e
,
“
A
C
o
m
b
in
e
d
Co
rn
e
r
A
n
d
Ed
g
e
De
tec
to
r,
”
Al
v
e
y
Vi
sio
n
Co
n
fer
e
n
c
e
,
v
o
l.
1
5
,
n
o
.
5
0
,
1
9
8
8
.
[2
0
]
D.
Cireg
a
n
,
U.
M
e
ier,
a
n
d
J.
S
c
h
m
id
h
u
b
e
r,
“
M
u
lt
i
-
c
o
lu
m
n
d
e
e
p
n
e
u
ra
l
n
e
tw
o
rk
s
f
o
r
im
a
g
e
c
las
sif
ica
ti
o
n
,
”
2
0
1
2
IEE
E
Co
n
fer
e
n
c
e
o
n
Co
m
p
u
ter
Vi
sio
n
a
n
d
Pa
tt
e
rn
Rec
o
g
n
it
io
n
,
2
0
1
2
,
p
p
.
3
6
4
2
-
3
6
4
9
,
d
o
i
:
1
0
.
1
1
0
9
/C
VP
R
.
2
0
1
2
.
6
2
4
8
1
1
0
.
[2
1
]
T
.
Y.
Zh
a
n
g
,
a
n
d
C.
Y.
S
u
e
n
,
“
A
F
a
st
P
a
ra
ll
e
l
A
l
g
o
rit
h
m
f
o
r
T
h
in
n
i
n
g
Dig
it
a
l
P
a
tt
e
r
n
s,”
Co
mm
u
n
ica
ti
o
n
s
o
f
th
e
ACM
,
v
o
l.
2
7
,
n
o
.
3
,
p
p
.
2
3
6
–
2
3
9
,
1
9
8
4
,
d
o
i:
1
0
.
1
1
4
5
/3
5
7
9
9
4
.
3
5
8
0
2
3
.
[2
2
]
Y.
Ou
a
d
i
d
,
M
.
F
a
k
ir,
a
n
d
B.
M
in
a
o
u
i,
“
T
if
in
a
g
h
P
rin
ted
C
h
a
ra
c
ter
Re
c
o
g
n
it
io
n
th
r
o
u
g
h
S
tr
u
c
tu
ra
l
F
e
a
tu
re
Ex
trac
ti
o
n
”
.
In
ter
n
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
C
o
mp
u
ter
Vi
sio
n
a
n
d
Ima
g
e
Pro
c
e
ss
in
g
,
v
o
l
.
6
,
n
o
.
2
,
p
p
:
4
2
-
5
3
,
2
0
1
6
,
d
o
i:
1
0
.
4
0
1
8
/IJC
V
I
P
.
2
0
1
6
0
7
0
1
0
3
.
[2
3
]
H
.
M
o
ra
v
e
c
,
“
Ob
sta
c
le
A
v
o
id
a
n
c
e
a
n
d
Na
v
ig
a
ti
o
n
i
n
th
e
Re
a
l
W
o
rld
b
y
a
S
e
e
in
g
Ro
b
o
t
Ro
v
e
r,
”
Diss
e
rtatio
n
,
Ro
b
o
ti
c
s In
stit
u
te,
Ca
rn
e
g
ie
-
M
e
ll
o
n
U
n
iv
e
rsity
,
P
it
tsb
u
rg
h
,
USA
,
1
9
8
0
.
[2
4
]
Y.
Ou
a
d
id
,
B.
El
b
a
lao
u
i,
M
.
B
o
u
tao
u
n
te,
M
.
F
a
k
ir
,
a
n
d
B
.
M
i
n
a
o
u
i
,
“
Ha
n
d
w
rit
ten
ti
f
in
a
g
h
c
h
a
ra
c
ter
re
c
o
g
n
it
io
n
u
sin
g
sim
p
le
g
e
o
m
e
tri
c
sh
a
p
e
s
a
n
d
g
ra
p
h
s
,
”
I
n
d
o
n
e
si
a
n
J
o
u
r
n
a
l
o
f
El
e
c
trica
l
E
n
g
in
e
e
rin
g
a
n
d
C
o
mp
u
ter
S
c
ien
c
e
(
IJ
EE
CS
)
,
v
ol
.
13
,
n
o
.
2
,
p
p
:
5
9
8
-
6
0
5
,
2
0
1
6
,
d
o
i:
1
0
.
1
1
5
9
1
/i
jee
c
s.v
1
3
.
i
2
.
p
p
5
9
8
-
6
0
5
.
[2
5
]
Y
.
Ou
a
d
id
,
M
.
Bo
u
tao
u
n
te,
M
.
F
a
k
ir
,
a
n
d
B
.
M
in
a
o
u
i
,
“
T
if
in
a
g
h
Ch
a
ra
c
ter
Re
c
o
g
n
it
io
n
u
sin
g
Ha
rris
Co
rn
e
r
De
tec
to
r
a
n
d
G
ra
p
h
Re
p
re
se
n
tatio
n
,”
I
n
ter
n
a
t
io
n
a
l
J
o
u
rn
a
l
o
f
Co
mp
u
ter
A
p
p
l
ica
ti
o
n
s
(
IJ
CA)
,
v
ol
.
1
4
9
,
n
o
.
3
,
p
p
.
1
7
-
2
3
,
S
e
p
.
2
0
1
6
,
d
o
i:
1
0
.
5
1
2
0
/i
jca
2
0
1
6
9
1
1
3
6
6
.
Evaluation Warning : The document was created with Spire.PDF for Python.