I
AE
S
I
nte
rna
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
3
,
No
.
2
,
J
u
n
e
201
4
,
p
p
.
64
~
72
I
SS
N:
2252
-
8938
64
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
jo
u
r
n
a
l.c
o
m/o
n
lin
e/in
d
ex
.
p
h
p
/I
J
AI
Yo
ruba
La
ng
ua
g
e and Nu
m
e
ra
ls’
O
f
fline
In
terpr
ete
r Using
M
o
rpho
lo
g
ica
l an
d Te
m
pla
te
M
a
t
c
hing
O
la
k
a
n
m
i O
.
O
la
da
y
o
El
e
c
tri
c
a
l
a
n
d
El
e
c
tro
n
ic E
n
g
in
e
e
rin
g
,
Un
iv
e
rsity
o
f
Ib
a
d
a
n
,
Ib
a
d
a
n
Nig
e
ria
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Dec
2
1
,
2
0
1
3
R
ev
i
s
ed
Ma
r
2
1
,
2
0
1
4
A
cc
ep
ted
A
p
r
27
,
2
0
1
4
Yo
ru
b
a
a
s
a
la
n
g
u
a
g
e
h
a
s
p
a
ss
e
d
t
h
ro
u
g
h
g
e
n
e
ra
ti
o
n
re
f
o
r
m
a
ti
o
n
s
m
a
k
in
g
so
m
e
o
f
th
e
o
l
d
d
o
c
u
m
e
n
ts
in
t
h
e
a
rc
h
iv
e
to
b
e
u
n
re
a
d
a
b
le
b
y
th
e
p
re
se
n
t
g
e
n
e
ra
ti
o
n
re
a
d
e
rs.
A
p
a
rt
f
ro
m
th
is,
so
m
e
Yo
ru
b
a
w
rit
e
rs
u
su
a
ll
y
m
i
x
e
d
En
g
li
sh
n
u
m
e
ra
ls
w
h
il
e
w
rit
in
g
d
u
e
to
b
re
v
it
y
a
n
d
c
o
n
c
ise
n
e
ss
o
f
En
g
li
sh
n
u
m
e
ra
l
c
o
m
p
a
re
to
Yo
r
u
b
a
n
u
m
e
ra
ls
w
h
ich
a
re
c
o
m
b
in
a
ti
o
n
o
f
se
v
e
ra
l
c
h
a
ra
c
ters
.
Re
-
t
y
p
in
g
su
c
h
h
ist
o
rica
l
d
o
c
u
m
e
n
ts
m
a
y
b
e
ti
m
e
c
o
n
su
m
in
g
,
th
e
re
f
o
re
a
n
e
e
d
f
o
r
a
n
e
f
f
i
c
ien
t
Op
ti
c
a
l
Ch
a
ra
c
ter
Re
a
d
e
r
(OCR)
w
h
ich
w
il
l
n
o
t
o
n
ly
e
ff
e
c
ti
v
e
l
y
re
c
o
g
n
ize
Y
o
ru
b
a
tex
ts
b
u
t
a
ls
o
c
o
n
v
e
rts
a
ll
t
h
e
En
g
li
s
h
n
u
m
e
ra
ls
in
th
e
d
o
c
u
m
e
n
t
to
Yo
ru
b
a
n
u
m
e
ra
ls.S
e
v
e
r
a
l
Op
ti
c
a
l
Ch
a
ra
c
ter
Re
a
d
e
r
(OCR)
s
y
ste
m
s
h
a
d
b
e
e
n
d
e
v
e
lo
p
e
d
to
re
c
o
g
n
ize
c
h
a
ra
c
ters
o
r
tex
ts
o
f
so
m
e
lan
g
u
a
g
e
s
su
c
h
a
s
E
n
g
li
s
h
,
A
ra
b
ic,
Ja
p
a
n
e
se
,
C
h
i
n
e
se
,
a
n
d
Ko
re
a
n
,
h
o
w
e
v
e
r,
d
e
sp
it
e
th
e
sig
n
if
ic
a
n
t
c
o
n
tri
b
u
t
io
n
o
f
Yo
ru
b
a
lan
g
u
a
g
e
to
h
ist
o
rica
l
d
o
c
u
m
e
n
tatio
n
a
n
d
c
o
m
m
u
n
ic
a
ti
o
n
,
it
w
a
s o
b
se
rv
e
d
th
a
t
t
h
e
re
is n
o
p
a
rti
c
u
lar
OCR
sy
ste
m
f
o
r
th
e
la
n
g
u
a
g
e
.
In
t
h
is
p
a
p
e
r
c
o
rre
latio
n
a
n
d
tem
p
late
m
a
tch
in
g
tec
h
n
iq
u
e
s w
e
re
u
se
d
to
d
e
v
e
lo
p
a
n
OCR f
o
r
th
e
re
c
o
g
n
it
io
n
o
f
Yo
ru
b
a
b
a
se
d
tex
ts
a
n
d
c
o
n
v
e
rt
En
g
li
sh
n
u
m
e
ra
ls
in
th
e
d
o
c
u
m
e
n
t
to
Yo
ru
b
a
n
u
m
e
ra
ls.
Ex
p
e
rime
n
tal
re
su
lt
s
sh
o
w
th
e
re
l
a
ti
v
e
l
y
h
ig
h
a
c
c
u
ra
c
y
o
f
th
e
d
e
v
e
l
o
p
e
d
OCR
w
h
e
n
it
w
a
s tes
ted
o
n
a
ll
siz
e
Yo
r
u
b
a
a
lp
h
a
b
e
ts an
d
n
u
m
e
ra
ls.
K
ey
w
o
r
d
:
I
m
ag
e
OC
R
P
atter
n
R
ec
o
g
n
itio
n
T
em
p
late
Ma
tch
in
g
Yo
r
u
b
a
Co
p
y
rig
h
t
©
201
4
In
s
t
it
u
te o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
Al
l
rig
h
ts re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Olak
a
n
m
i O
.
Olad
a
y
o
,
E
lectr
ical
an
d
E
lectr
o
n
ic
E
n
g
i
n
ee
r
in
g
,
Un
i
v
er
s
it
y
o
f
I
b
ad
an
,
I
b
ad
an
Nig
er
ia
T
ec
h
n
o
lo
g
y
Dr
i
v
e,
Of
f
ice
6
,
Ne
w
Facu
l
t
y
o
f
E
n
g
in
ee
r
i
n
g
B
u
ild
in
g
.
E
m
ail:
o
la
k
an
m
i.o
lad
a
y
o
@
u
i.
ed
u
.
n
g
1.
I
NT
RO
D
UCT
I
O
N
Yo
r
u
b
a
is
th
e
m
o
s
t
d
o
cu
m
e
n
t
ed
W
est
Af
r
ica
n
lan
g
u
ag
e.
Y
o
r
u
b
a
is
s
p
o
k
e
n
b
y
1
8
,
8
5
0
,
0
0
0
p
eo
p
le
in
Nig
er
ia.
T
h
e
to
tal
p
o
p
u
latio
n
o
f
n
ativ
e
s
p
ea
k
er
s
in
all
co
u
n
tr
ies
is
ab
o
u
t
2
0
,
0
0
0
,
0
0
0
.
T
h
e
n
u
m
b
er
r
is
es
to
2
2
,
0
0
0
,
0
0
0
if
w
e
also
i
n
clu
d
e
s
ec
o
n
d
-
la
n
g
u
a
g
e
s
p
ea
k
er
s
.
T
h
e
lan
g
u
ag
e
h
as
n
u
m
er
o
u
s
d
ialec
ts
s
p
o
k
en
i
n
d
if
f
er
e
n
t
ar
ea
s
o
f
Ni
g
er
ia.
W
it
h
in
Nig
er
ia
t
h
e
la
n
g
u
a
g
e
is
s
p
o
k
en
i
n
t
h
e
ar
ea
s
o
f
O
y
o
,
O
g
u
n
,
On
d
o
Osu
n
,
K
w
ar
a,
L
a
g
o
s
an
d
th
e
w
e
s
ter
n
p
ar
t
o
f
Ko
g
i
State.
I
t
is
also
s
p
o
k
en
in
B
en
in
,
T
o
g
o
,
an
d
b
y
i
m
m
i
g
r
an
t
s
in
th
e
Un
ited
Kin
g
d
o
m
an
d
th
e
US
A
.
Yo
r
u
b
a
is
o
n
e
o
f
th
e
1
2
E
d
ek
ir
i
la
n
g
u
a
g
es
o
f
th
e
Yo
r
u
b
o
id
g
r
o
u
p
th
at
also
in
clu
d
es
I
g
ala.
T
h
e
Yo
r
u
b
o
id
g
r
o
u
p
b
elo
n
g
s
to
t
h
e
De
f
o
id
lan
g
u
ag
e
s
o
f
t
h
e
B
en
u
e
-
C
o
n
g
o
g
r
o
u
p
a
n
d
u
l
ti
m
a
tel
y
to
th
e
Vo
lta
-
C
o
n
g
o
,
an
d
A
tlan
tic
-
C
o
n
g
o
g
r
o
u
p
s
o
f
th
e
Ni
g
er
-
C
o
n
g
o
Fa
m
il
y
o
f
1
4
1
9
lan
g
u
a
g
e
s
m
o
s
tl
y
s
p
o
k
en
i
n
C
en
tr
al
a
n
d
So
u
t
h
Af
r
ica
(
I
s
aa
c,
2
0
0
3
)
.
I
m
ag
e
r
ec
o
g
n
itio
n
i
s
th
e
p
r
o
ce
s
s
o
f
id
en
ti
f
y
i
n
g
a
n
d
d
etec
tin
g
an
o
b
j
ec
t
o
r
a
f
ea
tu
r
e
in
a
d
ig
ital
i
m
a
g
e
o
r
v
id
eo
.
T
h
is
co
n
ce
p
t
is
u
s
ed
in
m
a
n
y
ap
p
licatio
n
s
lik
e
s
y
s
te
m
s
f
o
r
f
ac
to
r
y
a
u
to
m
a
tio
n
,
to
ll
b
o
o
th
m
o
n
ito
r
in
g
,
an
d
s
ec
u
r
it
y
s
u
r
v
ei
llan
ce
.
T
y
p
ical
i
m
a
g
e
r
ec
o
g
n
i
tio
n
al
g
o
r
it
h
m
s
i
n
clu
d
e:
a.
Op
tical
ch
ar
ac
ter
r
ec
o
g
n
itio
n
b.
P
atter
n
an
d
g
r
ad
ien
t
m
atc
h
in
g
c.
Face
r
ec
o
g
n
itio
n
d.
L
ice
n
s
e
p
late
m
atch
in
g
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
Yo
r
u
b
a
La
n
g
u
a
g
e
a
n
d
N
u
mera
ls
’
Offlin
e
I
n
terp
r
eter U
s
in
g
Mo
r
p
h
o
lo
g
ica
l
a
n
d
…
(
Ola
ka
n
mi
O.
Ola
d
a
yo
)
65
e.
Scen
e
ch
a
n
g
e
d
etec
tio
n
I
t
h
as
b
ec
o
m
e
a
tr
e
n
d
to
d
o
cu
m
en
t
m
o
s
t
o
f
t
h
e
d
o
cu
m
e
n
ts
i
n
t
h
e
ar
ch
iv
e
s
u
s
i
n
g
s
ca
n
n
er
,
h
o
w
ev
er
,
th
ese
d
o
cu
m
en
ts
ca
n
n
o
t
b
e
ed
ited
o
r
r
ea
d
th
er
ea
f
ter
b
y
co
m
p
u
ter
s
y
s
te
m
s
.
Du
e
to
th
e
f
ac
t
th
at
s
ca
n
n
er
s
ca
n
s
d
o
cu
m
en
ts
a
s
a
n
i
m
a
g
e
n
o
t
as
en
co
d
ed
s
et
o
f
c
h
ar
ac
ter
s
.
Op
tical
C
h
ar
ac
ter
R
ea
d
er
(
OC
R
)
s
y
s
te
m
d
o
es
elec
tr
o
n
ic
tr
an
s
latio
n
o
f
h
an
d
w
r
itte
n
o
r
p
r
in
ted
tex
t in
to
m
a
ch
in
e
e
n
co
d
ed
tex
t.
OC
R
is
w
i
d
ely
u
s
ed
to
co
n
v
er
t
b
o
o
k
s
an
d
d
o
cu
m
e
n
ts
in
to
elec
tr
o
n
ic
f
iles
an
d
to
co
m
p
u
ter
ize
a
r
ec
o
r
d
k
ee
p
in
g
s
y
s
te
m
in
an
o
f
f
ice.
OC
R
m
a
k
es
it p
o
s
s
ib
le
to
ed
it
s
u
ch
d
o
cu
m
en
t,
s
ea
r
ch
f
o
r
a
w
o
r
d
o
r
p
h
r
ase,
s
to
r
e
it m
o
r
e
co
m
p
ac
tl
y
,
d
is
p
lay
o
r
p
r
in
t a
co
p
y
an
d
ap
p
ly
tec
h
n
iq
u
e
s
s
u
c
h
as
m
ac
h
in
e
tr
a
n
s
lat
io
n
,
tex
t
-
to
-
s
p
ee
ch
an
d
tex
t
m
i
n
i
n
g
to
it.
Op
tical
C
h
ar
a
cter
R
ec
o
g
n
itio
n
s
tu
d
y
w
as
s
tar
ted
b
y
T
y
u
r
i
n
a
R
u
s
s
ia
n
s
cie
n
tis
t
(
A
.
J
ain
&
Ka
r
u
,
2
0
0
6
)
.
T
h
e
f
ir
s
t
m
o
d
er
n
c
h
ar
ac
ter
r
ec
o
g
n
izer
s
ap
p
ea
r
ed
in
th
e
m
id
d
l
e
o
f
th
e
1
9
4
0
s
w
it
h
t
h
e
d
ev
elo
p
m
e
n
t
o
f
th
e
d
i
g
ital
co
m
p
u
ter
.
T
h
e
ea
r
l
y
w
o
r
k
o
n
th
e
a
u
to
m
a
tic
r
ec
o
g
n
itio
n
o
f
c
h
ar
ac
ter
s
h
as
b
ee
n
co
n
ce
n
tr
at
ed
eith
er
u
p
o
n
w
ell
p
r
in
ted
tex
t
o
r
u
p
o
n
s
m
all
s
et
o
f
w
ell
d
is
t
in
g
u
is
h
ed
h
an
d
w
r
it
ten
tex
t
o
r
s
y
m
b
o
ls
,
alth
o
u
g
h
,
s
u
cc
e
s
s
f
u
l
b
u
t
h
ad
b
ee
n
i
m
p
le
m
e
n
ted
m
o
s
t
l
y
f
o
r
L
ati
n
c
h
ar
ac
ter
s
an
d
n
u
m
er
a
ls
.
B
esid
es
s
o
m
e
s
t
u
d
ies
o
n
J
ap
an
ese,
C
h
i
n
ese
,
Heb
r
e
w
,
I
n
d
ian
a
n
d
A
r
ab
ic
ch
ar
ad
es
an
d
n
u
m
er
al
s
in
b
o
t
h
p
r
in
ted
an
d
h
an
d
w
r
itte
n
ca
s
e
s
wer
e
also
co
n
s
id
er
ed
b
y
s
o
m
e
O
C
R
s
y
s
te
m
s
.
T
h
e
d
ev
elo
p
m
en
t
s
i
n
OC
R
u
n
ti
l
1
9
8
0
s
s
u
f
f
er
ed
f
r
o
m
lac
k
o
f
a
d
v
an
ce
d
alg
o
r
it
h
m
,
p
o
w
er
f
u
l
co
m
p
u
ti
n
g
h
ar
d
w
ar
e
an
d
o
p
tical
d
ev
ices.
W
it
h
th
e
o
u
t
w
ar
d
ex
p
lo
s
io
n
o
n
th
e
co
m
p
u
ti
n
g
tec
h
n
o
lo
g
y
d
ev
elo
p
m
en
t,
th
e
p
r
ev
io
u
s
l
y
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
ie
s
f
o
u
n
d
a
f
er
tile
e
n
v
ir
o
n
m
e
n
t
f
o
r
r
ap
id
g
r
o
w
t
h
i
n
m
a
n
y
ap
p
licatio
n
ar
ea
s
.
P
r
esen
tl
y
,
r
en
e
w
ed
v
ig
o
u
r
s
ar
e
b
ein
g
p
u
t
in
th
e
o
p
tical
ch
ar
ac
ter
r
ec
o
g
n
itio
n
r
esear
ch
.
On
e
o
f
th
ese
is
r
ec
o
g
n
itio
n
o
f
p
r
in
ted
an
d
h
an
d
w
r
itte
n
d
o
cu
m
e
n
ts
.
Mo
r
e
s
o
p
h
is
ticated
alg
o
r
i
th
m
s
w
h
ich
u
tili
ze
ad
v
an
ce
d
m
et
h
o
d
o
lo
g
ies ar
e
b
ein
g
d
e
v
elo
p
ed
.
I
n
th
is
w
o
r
k
t
w
o
m
et
h
o
d
o
lo
g
ie
s
ar
e
co
m
b
in
ed
to
ac
h
iev
e
a
n
e
f
f
icien
t
Yo
r
u
b
a
OC
R
s
y
s
te
m
w
h
ic
h
w
ill
b
e
ab
le
to
r
ec
o
g
n
ize
o
f
f
-
li
n
e
t
y
p
ed
an
d
h
an
d
w
r
itte
n
Yo
r
u
b
a
d
o
cu
m
e
n
ts
an
d
co
n
v
er
t
E
n
g
li
s
h
n
u
m
er
als
to
Yo
r
u
b
a
n
u
m
er
als.
T
h
e
r
e
m
ain
in
g
p
ar
t
o
f
th
is
p
ap
er
is
ar
r
an
g
ed
as
f
o
llo
w
s
:
s
ec
tio
n
2
is
th
e
r
ev
ie
w
o
f
r
elate
d
w
o
r
k
s
on
OC
R
s
y
s
te
m
s
a
n
d
m
eth
o
d
o
lo
g
ies.
T
h
e
d
esig
n
m
e
th
o
d
o
lo
g
y
a
n
d
w
o
r
k
i
n
g
p
r
in
cip
le
o
f
th
e
s
y
s
te
m
ar
e
ex
p
lain
ed
in
s
ec
tio
n
3
.
Sectio
n
4
co
n
tai
n
s
th
e
tes
t r
esu
l
ts
a
n
d
co
n
clu
s
io
n
.
2
.
RE
L
AT
E
D
WO
RK
S
(
Ku
n
d
u
,
MI
T
R
E
C
o
r
p
.
,
Hin
es
,
P
h
illi
p
s
,
&
Hu
y
c
k
,
2
0
0
7
)
r
ef
er
en
ce
d
escr
ib
ed
a
co
m
p
lete
s
y
s
te
m
f
o
r
th
e
r
ec
o
g
n
itio
n
o
f
u
n
co
n
s
tr
ain
ed
h
an
d
w
r
itte
n
A
r
ab
ic
w
o
r
d
s
u
s
in
g
o
v
er
-
s
e
g
m
e
n
tatio
n
o
f
c
h
a
r
ac
ter
s
an
d
v
ar
iab
le
d
u
r
atio
n
h
id
d
en
Ma
r
k
o
v
m
o
d
el
(
VDHM
M)
.
I
n
th
i
s
,
a
s
eg
m
en
tatio
n
al
g
o
r
it
h
m
w
a
s
u
s
ed
t
o
t
r
an
s
late
t
h
e
2
-
D
i
m
a
g
e
in
to
1
-
D
s
eq
u
en
ce
o
f
s
u
b
-
ch
ar
ac
ter
s
y
m
b
o
ls
.
T
h
is
s
eq
u
en
ce
o
f
s
y
m
b
o
ls
w
as
m
o
d
ele
d
b
y
th
e
VDHM
M.
T
h
e
s
h
ap
e
in
f
o
r
m
a
tio
n
o
f
c
h
ar
ac
ter
an
d
s
u
b
-
ch
ar
ac
ter
s
y
m
b
o
l
s
w
as
co
m
p
ac
tl
y
r
ep
r
es
en
ted
b
y
f
o
r
t
y
-
f
iv
e
f
ea
t
u
r
es
in
t
h
e
f
ea
t
u
r
e
s
p
ac
e.
T
h
e
f
e
atu
r
e
v
ec
to
r
w
a
s
m
o
d
e
led
as
an
in
d
ep
en
d
en
tl
y
d
is
tr
i
b
u
ted
m
u
l
tiv
ar
iate
d
is
cr
ete
d
is
tr
ib
u
tio
n
.
An
d
th
e
v
ar
iab
le
d
u
r
atio
n
s
tate
i
s
u
s
ed
to
r
eso
lv
e
th
e
s
e
g
m
en
ta
tio
n
a
m
b
ig
u
it
y
a
m
o
n
g
th
e
co
n
s
ec
u
tiv
e
c
h
ar
ac
ter
s
.
Dif
f
er
en
t
m
et
h
o
d
o
lo
g
ies
o
n
h
o
w
th
e
q
u
ali
t
y
o
f
t
h
e
c
ap
tu
r
ed
ca
m
er
a
i
m
a
g
e
co
u
ld
b
e
im
p
r
o
v
ed
h
ad
b
ee
n
th
o
r
o
u
g
h
l
y
co
n
s
id
er
ed
b
y
v
ar
io
u
s
r
esear
ch
es.
Fo
r
ex
a
m
p
le,
(
Du
eir
e
L
i
n
s
,
P
er
eir
a
Sil
v
a,
&
Go
m
e
s
e
Sil
v
a,
2
0
0
7
)
an
al
y
ze
d
th
e
q
u
alit
y
o
f
s
u
ch
ca
p
t
u
r
ed
i
m
ag
e
f
o
r
o
p
tical
ch
ar
ac
ter
r
ec
o
g
n
itio
n
.
I
n
t
h
eir
w
o
r
k
d
if
f
er
en
t
m
ea
n
s
o
f
i
m
p
r
o
v
i
n
g
tr
an
s
cr
ip
tio
n
an
d
r
ec
o
g
n
itio
n
w
as
p
r
o
p
o
s
ed
.
A
ls
o
,
(
Yin
,
Su
n
,
Nao
i,
&
Fu
j
i
m
o
to
,
2
0
0
7
)
p
r
o
p
o
s
ed
a
n
e
w
p
er
s
p
ec
ti
v
e
r
ec
tif
icatio
n
s
y
s
te
m
b
ased
o
n
v
an
i
s
h
in
g
p
o
in
t
d
etec
t
io
n
.
T
h
eir
s
y
s
te
m
ac
h
ie
v
ed
b
o
th
th
e
d
esire
d
ef
f
icie
n
c
y
an
d
ac
cu
r
ac
y
u
s
i
n
g
a
m
u
lti
-
s
tag
e
s
tr
ateg
y
:
at
t
h
e
f
ir
s
t
s
tag
e,
d
o
cu
m
e
n
t
b
o
u
n
d
ar
ies
an
d
s
tr
aig
h
t
li
n
es
ar
e
u
s
ed
to
co
m
p
u
te
v
a
n
i
s
h
i
n
g
p
o
in
ts
;
at
t
h
e
s
ec
o
n
d
s
tag
e,
tex
t
b
aselin
e
s
an
d
b
lo
ck
alig
n
s
ar
e
u
tili
ze
d
;
a
n
d
at
th
e
last
s
tag
e,
ch
ar
ac
ter
tilt
o
r
ien
tat
io
n
s
ar
e
v
o
ted
f
o
r
th
e
v
er
tical
v
an
is
h
in
g
p
o
i
n
t.
A
p
r
o
f
it
f
u
n
ctio
n
w
as
i
n
tr
o
d
u
ce
d
to
ev
alu
ate
th
e
r
eliab
ilit
y
o
f
d
etec
ted
v
an
i
s
h
in
g
p
o
in
t
s
at
ea
ch
s
ta
g
e.
I
f
v
a
n
is
h
i
n
g
p
o
in
t
s
at
o
n
e
s
tag
e
ar
e
r
eliab
le,
th
e
n
r
ec
tif
ica
t
io
n
i
s
en
d
ed
at
th
at
s
tag
e.
Ot
h
er
w
is
e,
m
u
lti
-
s
ta
g
e
s
tr
ate
g
y
m
et
h
o
d
co
n
tin
u
es to
o
b
tain
m
o
r
e
r
eliab
le
v
an
i
s
h
i
n
g
p
o
in
ts
i
n
t
h
e
n
e
x
t stag
e.
R
esear
ch
h
a
s
s
h
o
w
n
t
h
at
C
h
a
r
ac
ter
d
eg
r
ad
atio
n
af
f
ec
ts
m
ac
h
in
e
p
r
in
ted
ch
ar
ac
ter
r
ec
o
g
n
i
tio
n
.
T
w
o
m
ai
n
r
ea
s
o
n
s
f
o
r
d
eg
r
a
d
atio
n
w
er
e
ex
tr
i
n
s
ic
i
m
a
g
e
d
eg
r
ad
ati
o
n
s
u
c
h
as
b
lu
r
r
i
n
g
a
n
d
lo
w
i
m
ag
e
d
i
m
e
n
s
io
n
,
an
d
in
tr
i
n
s
ic
d
eg
r
ad
atio
n
ca
u
s
ed
b
y
f
o
n
t
v
ar
iatio
n
s
.
A
r
ec
o
g
n
i
tio
n
m
et
h
o
d
th
at
co
m
b
i
n
es
t
w
o
co
m
p
le
m
e
n
tar
y
class
i
f
ier
s
i
s
p
r
o
p
o
s
ed
in
.
(
Su
n
,
Hu
a
n
g
,
Ho
t
ta,
&
F
u
j
i
m
o
to
,
2
0
0
7
)
.
T
h
e
lo
ca
l
f
ea
t
u
r
e
b
ased
class
i
f
ier
ex
tr
ac
t
s
th
e
lo
ca
l
co
n
to
u
r
d
ir
ec
tio
n
ch
an
g
e
s
,
w
h
ic
h
is
ef
f
ec
ti
v
e
f
o
r
ch
ar
ac
ter
p
atter
n
s
w
ith
le
s
s
s
tr
u
ctu
r
e
d
eter
io
r
atio
n
.
T
h
e
g
lo
b
al
f
ea
tu
r
e
b
ased
clas
s
if
ier
ex
tr
ac
ts
t
h
e
tex
tu
r
e
d
is
tr
i
b
u
tio
n
o
f
t
h
e
c
h
ar
ac
ter
i
m
a
g
e,
w
h
ic
h
is
e
f
f
ec
tiv
e
w
h
e
n
th
e
c
h
ar
ac
ter
s
tr
u
ctu
r
e
is
h
ar
d
to
d
is
cr
i
m
in
ate.
T
h
e
t
w
o
co
m
p
le
m
e
n
tar
y
cla
s
s
i
f
ier
s
ar
e
co
m
b
i
n
ed
b
y
ca
n
d
id
ate
f
u
s
io
n
i
n
a
co
ar
s
e
-
to
-
f
i
n
e
s
t
y
le.
E
x
p
er
i
m
en
t
s
ar
e
ca
r
r
ied
o
n
d
eg
r
ad
e
d
C
h
in
e
s
e
ch
ar
ac
ter
r
ec
o
g
n
itio
n
(
P
r
atap
,
Sat
y
ap
r
asad
,
&
Sas
tr
y
)
w
o
r
k
ed
o
n
C
h
ar
ac
ter
r
ec
o
g
n
i
tio
n
s
y
s
te
m
T
elu
g
u
;
o
n
e
o
f
t
h
e
an
cie
n
t
l
a
n
g
u
a
g
es
o
f
So
u
t
h
I
n
d
ia.
I
t
h
as
a
co
m
p
lex
o
r
th
o
g
r
ap
h
y
w
it
h
a
lar
g
e
n
u
m
b
er
o
f
d
is
ti
n
ct
c
h
ar
ac
ter
s
h
ap
es
co
m
p
o
s
ed
o
f
s
i
m
p
le
a
n
d
co
m
p
o
u
n
d
c
h
ar
ac
ter
s
.
I
n
th
is
w
o
r
k
,
s
tr
u
ct
u
r
al
f
e
atu
r
es
o
f
t
h
e
s
y
llab
le
an
d
t
h
e
co
m
p
o
n
e
n
t
m
o
d
el
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
3
,
No
.
2
,
J
u
n
e
201
4
:
64
–
72
66
w
er
e
co
m
b
i
n
ed
to
ex
tr
ac
t
m
id
d
le
zo
n
e
co
m
p
o
n
e
n
t
s
.
T
h
e
s
h
ap
e
o
f
th
e
m
id
d
le
zo
n
e
co
m
p
o
n
e
n
ts
i
s
clo
s
el
y
r
elate
d
to
a
cir
cle
w
h
er
ea
s
o
t
h
er
co
m
p
o
n
en
t
s
ar
e
f
o
u
n
d
w
ith
d
i
f
f
er
en
t to
p
o
lo
g
ical
f
ea
tu
r
e
s
.
A
s
i
m
p
le
an
d
ef
f
ec
ti
v
e
te
m
p
late
m
atch
in
g
m
et
h
o
d
f
o
r
id
en
ti
f
icatio
n
o
f
Mu
s
n
ad
ch
ar
ac
ter
s
w
a
s
in
tr
o
d
u
ce
d
in
(
Mo
h
a
m
m
ed
,
2
0
1
1
)
.
T
h
e
ch
ar
ac
ter
s
w
er
e
ex
t
r
ac
ted
f
r
o
m
in
p
u
t
i
m
ag
e
a
n
d
n
o
r
m
alize
d
.
D
u
r
in
g
r
ec
o
g
n
itio
n
,
th
e
ex
tr
ac
ted
ch
ar
ac
ter
w
a
s
co
m
p
ar
ed
to
ea
c
h
te
m
p
late
in
th
e
d
atab
ase
to
f
in
d
th
e
clo
s
est
r
ep
r
esen
tatio
n
o
f
th
e
i
n
p
u
t
ch
ar
ac
ter
.
T
h
e
m
atc
h
in
g
m
etr
ic
w
a
s
co
m
p
u
ted
u
s
i
n
g
2
-
D
co
r
r
elatio
n
co
ef
f
icie
n
t
s
ap
p
r
o
ac
h
to
id
en
tify
s
i
m
ilar
p
atter
n
s
b
et
w
ee
n
th
e
te
s
t i
m
ag
e
an
d
th
e
d
atab
ase
i
m
ag
e
s
.
I
n
(
Hu
a
n
g
,
S
u
n
,
Ho
tta,
&
F
u
j
im
o
to
)
,
a
n
o
v
el
ap
p
r
o
ac
h
to
ef
f
icien
tl
y
r
ec
o
g
n
ize
h
a
n
d
w
r
itte
n
n
u
m
er
als
w
a
s
p
r
o
p
o
s
ed
.
T
h
is
ap
p
r
o
ac
h
ex
p
lo
its
a
t
w
o
-
s
ta
g
e
f
r
a
m
e
w
o
r
k
b
y
u
s
i
n
g
d
if
f
er
en
ce
f
ea
tu
r
es.
I
n
t
h
e
f
ir
s
t
s
ta
g
e,
a
r
eg
u
lar
SVM
i
s
tr
ai
n
ed
o
n
all
th
e
tr
ai
n
i
n
g
d
ata;
i
n
t
h
e
s
ec
o
n
d
s
tag
e,
o
n
l
y
t
h
e
s
a
m
p
le
s
m
is
c
lass
i
f
ied
i
n
t
h
e
f
ir
s
t
s
tag
e
ar
e
s
p
ec
iall
y
co
n
s
id
er
ed
.
T
h
e
n
u
m
b
er
o
f
m
i
s
clas
s
i
f
icatio
n
s
is
o
f
ten
s
m
all
b
ec
au
s
e
o
f
th
e
g
o
o
d
p
e
r
f
o
r
m
a
n
ce
o
f
SVM.
T
h
is
w
i
l
l
p
r
esen
t
d
i
f
f
i
cu
ltie
s
in
tr
ai
n
i
n
g
an
ac
c
u
r
ate
SVM
en
g
i
n
e
o
n
l
y
f
o
r
th
ese
m
i
s
class
if
ied
s
a
m
p
le
s
.
W
e
th
en
f
u
r
t
h
er
p
r
o
p
o
s
e
a
m
u
l
ti
-
w
a
y
to
b
in
ar
y
ap
p
r
o
ac
h
u
s
i
n
g
d
if
f
er
e
n
ce
f
ea
t
u
r
es.
T
h
is
ap
p
r
o
ac
h
s
u
cc
ess
f
u
ll
y
tr
an
s
f
o
r
m
s
m
u
lti
-
ca
te
g
o
r
y
clas
s
if
ica
tio
n
to
b
in
ar
y
c
lass
if
ica
ti
o
n
an
d
ex
p
an
d
s
t
h
e
tr
ai
n
in
g
s
a
m
p
les
g
r
ea
tl
y
.
2
.
1
.
O
v
er
v
ie
w
o
f
Y
o
ruba
O
r
t
ho
g
ra
ph
y
I
n
its
w
r
itte
n
f
o
r
m
,
Yo
r
u
b
a
u
s
es
th
e
R
o
m
an
alp
h
ab
et.
I
t
h
as
2
5
letter
s
as
s
h
o
w
n
i
n
F
ig
u
r
e
2
.
T
h
e
letter
'
p
'
is
al
w
a
y
s
p
r
o
n
o
u
n
ce
d
as
'k
'
an
d
'
p
'
co
m
b
in
ed
.
Yo
r
u
b
a
o
r
th
o
g
r
ap
h
y
d
o
es
n
o
t
u
s
e
t
h
e
letter
s
c,
q
,
v
,
x
,
z.
Yo
r
u
b
a
h
as
th
r
ee
b
asic
to
n
e
s
,
h
ig
h
,
m
id
,
an
d
lo
w
,
w
h
ich
ar
e
in
d
icate
d
in
t
h
e
o
r
th
o
g
r
ap
h
y
.
T
h
e
h
ig
h
is
m
ar
k
ed
w
it
h
an
ac
u
te
ac
ce
n
t
(
e.
g
.
á)
,
t
h
e
lo
w
w
it
h
a
g
r
av
e
ac
ce
n
t
(
à)
,
an
d
th
e
m
id
to
n
e
u
s
u
all
y
lef
t
u
n
m
ar
k
ed
.
T
h
ese
m
ar
k
s
ar
e
u
s
u
al
l
y
p
lace
d
o
n
th
e
v
o
w
el
s
.
I
n
s
o
m
e
cir
cu
m
s
ta
n
ce
s
th
e
m
id
to
n
e
is
in
d
icate
d
w
it
h
a
'm
ac
r
o
n
'
.
T
h
e
lan
g
u
a
g
e
h
a
s
b
ee
n
w
r
itte
n
s
in
c
e
th
e
ea
r
l
y
1
9
th
ce
n
t
u
r
y
,
al
th
o
u
g
h
t
h
er
e
h
a
v
e
b
ee
n
m
a
n
y
c
h
a
n
g
es
i
n
asp
ec
ts
o
f
its
o
r
th
o
g
r
ap
h
ic
r
ep
r
ese
n
tatio
n
.
I
n
th
e
1
9
6
0
s
,
th
e
th
e
n
Mi
n
is
tr
y
o
f
E
d
u
ca
tio
n
w
it
h
i
n
th
e
W
e
s
ter
n
R
eg
io
n
o
f
Ni
g
er
ia,
w
h
ic
h
w
as
w
h
er
e
m
o
s
t
o
f
t
h
e
Yo
r
u
b
a
s
p
ea
k
in
g
co
m
m
u
n
it
y
is
lo
ca
ted
,
f
o
r
m
ed
t
w
o
co
m
m
ittees
to
co
n
s
id
er
a
s
tan
d
ar
d
o
r
th
o
g
r
ap
h
y
f
o
r
th
e
lan
g
u
a
g
e.
T
h
e
m
o
r
e
in
f
lu
e
n
tial o
f
th
ese
t
w
o
,
th
e
Yo
r
u
b
a
Or
th
o
g
r
ap
h
y
C
o
m
m
ittee
w
a
s
s
et
u
p
i
n
1
9
6
6
.
T
h
e
r
ep
o
r
t
w
h
ic
h
t
h
i
s
s
ec
o
n
d
o
r
t
h
o
g
r
ap
h
y
co
m
m
i
ttee
s
u
b
m
itted
i
n
1
9
6
6
b
ec
am
e
th
e
b
asis
f
o
r
th
e
cr
ea
tio
n
a
n
d
in
tr
o
d
u
cti
o
n
in
to
s
c
h
o
o
ls
o
f
t
h
e
s
ta
n
d
ar
d
Yo
r
u
b
a
o
r
th
o
g
r
ap
h
y
(
I
s
aa
c,
2
0
0
3
)
.
Fig
u
r
e
2
(
a
).
I
n
p
u
t i
m
a
g
e
ch
ar
a
cter
b
ef
o
r
e
n
o
r
m
aliza
t
io
n
Fig
u
r
e
2
(
b
).
I
n
p
u
t i
m
ag
e
te
x
t a
f
ter
n
o
r
m
a
lizatio
n
Fig
u
r
e
2
(
c
).
I
n
p
u
t i
m
a
g
e
tex
t a
f
ter
f
i
lter
ed
T
ab
le
1
.
E
n
g
lis
h
n
u
m
er
als an
d
th
eir
eq
u
i
v
alen
t Y
o
r
u
b
a
n
u
m
e
r
als
En
g
l
i
sh
1
2
3
4
5
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
Yo
r
u
b
a
La
n
g
u
a
g
e
a
n
d
N
u
mera
ls
’
Offlin
e
I
n
terp
r
eter U
s
in
g
Mo
r
p
h
o
lo
g
ica
l
a
n
d
…
(
Ola
ka
n
mi
O.
Ola
d
a
yo
)
67
Y
o
r
u
b
a
En
i
Ej
i
Et
a
Er
i
n
A
r
u
n
En
g
l
i
sh
6
7
8
9
10
0
Y
o
r
u
b
a
Ef
a
Ej
e
Ej
o
Esa
n
Ew
a
O
d
o
T
ab
le
2
.
Yo
r
u
b
a
u
p
p
er
an
d
lo
w
er
alp
h
ab
ets
2
.
2
.
Yo
ruba
O
CR
Sy
s
t
e
m
M
et
ho
do
lo
g
y
OC
R
a
s
ea
r
lier
s
tated
is
th
e
s
cien
ce
t
h
at
en
tai
ls
t
h
e
d
escr
ip
tio
n
o
r
class
i
f
icatio
n
o
f
ch
ar
ac
ter
m
ea
s
u
r
e
m
e
n
t
s
th
at
u
s
u
al
l
y
b
as
ed
o
n
s
o
m
e
m
o
d
els.
OC
R
is
o
n
e
o
f
th
e
ca
teg
o
r
ies
o
f
i
m
a
g
e
r
ec
o
g
n
itio
n
.
T
h
er
e
is
v
ar
io
u
s
c
h
ar
ac
ter
r
ec
o
g
n
itio
n
m
et
h
o
d
s
u
s
ed
i
n
d
ev
elo
p
in
g
ch
ar
ac
ter
r
ec
o
g
n
izer
.
T
h
ese
m
et
h
o
d
s
ar
e:
n
eu
r
a
l
n
et
w
o
r
k
,
m
o
m
en
t
b
ased
ap
p
r
o
ac
h
,
co
n
to
u
r
b
ased
ap
p
r
o
ac
h
,
te
m
p
late
m
atc
h
i
n
g
an
d
m
o
r
p
h
o
lo
g
ical
ap
p
r
o
ac
h
.
I
n
t
h
is
w
o
r
k
te
m
p
late
m
atc
h
i
n
g
an
d
m
o
r
p
h
o
lo
g
ical
tech
n
iq
u
es
ar
e
u
s
ed
to
r
ec
o
g
n
ize
Yo
r
u
b
a
tex
ts
.
T
e
m
p
late
m
atc
h
in
g
r
ef
er
s
to
th
e
p
r
o
ce
s
s
o
f
d
etec
tin
g
an
o
b
j
ec
t
h
av
i
n
g
a
ce
r
tain
s
ize,
s
h
ap
e
an
d
o
r
ien
tatio
n
i
n
a
n
i
m
a
g
e
b
y
ap
p
l
y
i
n
g
an
o
p
er
ato
r
co
n
tain
i
n
g
p
o
s
iti
v
e
w
ei
g
h
t
s
in
a
r
eg
io
n
r
ese
m
b
li
n
g
t
h
e
o
b
j
ec
ts
to
b
e
d
etec
ted
an
d
co
n
tain
i
n
g
n
eg
at
iv
e
w
eig
h
t
s
i
n
a
r
eg
io
n
s
u
r
r
o
u
n
d
in
g
t
h
e
p
o
s
it
iv
e
w
ei
g
h
t
(
R
.
M.
K
S
i
n
h
a,
1
9
9
7
)
.
Mo
r
p
h
o
lo
g
y
a
s
d
er
iv
ed
f
r
o
m
b
io
lo
g
y
i
s
a
b
r
an
ch
o
f
b
io
lo
g
y
w
h
ic
h
d
ea
ls
w
it
h
th
e
f
o
r
m
a
n
d
an
i
m
a
ls
an
d
p
lan
ts
.
I
t is ad
o
p
ted
in
th
is
co
n
te
x
t
a
s
a
to
o
l
f
o
r
ex
tr
a
ctin
g
i
m
ag
e
co
m
p
o
n
e
n
ts
th
a
t
ar
e
u
s
e
f
u
l
i
n
t
h
e
r
ep
r
esen
tat
io
n
a
n
d
d
escr
ip
tio
n
o
f
th
e
r
eg
io
n
s
h
ap
e.
T
h
er
e
ar
e
s
ev
er
al
p
r
o
ce
d
u
r
al
s
tep
s
e
n
g
a
g
ed
in
ac
h
iev
i
n
g
m
o
r
p
h
o
lo
g
ica
l
tech
n
iq
u
e
s
.
T
h
ese
in
cl
u
d
e
f
ilter
i
n
g
,
th
i
n
n
in
g
,
p
r
u
n
in
g
,
er
o
s
io
n
a
n
d
d
ilatio
n
,
o
p
en
in
g
a
n
d
clo
s
in
g
.
3.
YO
RUB
A
O
CR
I
M
P
L
E
M
E
NT
A
T
I
O
N
USI
N
G
T
E
M
P
L
AT
E
M
AT
CH
I
NG
AND
M
O
RP
H
O
L
O
G
I
CA
L
T
E
C
H
NIQ
U
E
T
em
p
late
m
atc
h
i
n
g
an
d
m
o
r
p
h
o
lo
g
ical
tech
n
iq
u
es
as
s
tated
ea
r
lier
,
ar
e
OC
R
r
ec
o
g
n
itio
n
tech
n
iq
u
es
.
T
h
ese
alg
o
r
ith
m
s
in
v
o
lv
e
f
ea
t
u
r
es
ex
tr
ac
tio
n
an
d
class
i
f
ier
.
I
n
te
m
p
late
m
atc
h
i
n
g
i
m
a
g
e
p
ix
els
ar
e
u
s
ed
as
th
e
f
ea
t
u
r
es
b
ein
g
ex
tr
ac
ted
f
r
o
m
b
o
th
th
e
i
n
p
u
t
c
h
ar
ac
ter
an
d
t
h
e
clas
s
i
f
ied
ch
ar
ac
ter
s
.
T
h
e
class
i
f
ier
co
m
p
ar
es
th
e
i
n
p
u
t
ch
ar
ac
ter
f
ea
t
u
r
es
w
i
th
a
s
et
o
f
c
h
ar
ac
ter
te
m
p
la
te
i
n
t
h
e
c
h
ar
ac
ter
clas
s
.
I
n
th
is
c
o
n
tex
t
th
e
ch
ar
ac
ter
class
co
n
tai
n
s
n
u
m
er
al
s
,
u
p
p
er
an
d
lo
w
er
ca
s
es
o
f
Yo
r
u
b
a
ch
ar
ac
ter
s
as
s
h
o
w
n
i
n
f
i
g
1
an
d
f
i
g
.
2
.
T
h
e
ab
s
o
lu
te
v
alu
e
o
f
th
e
clas
s
i
f
ier
p
r
o
ce
d
u
r
e
w
h
ic
h
is
th
e
co
r
r
elatio
n
co
ef
f
icie
n
t
b
et
w
ee
n
th
e
i
n
p
u
t
ch
ar
ac
ter
an
d
th
e
co
n
s
id
er
ed
ch
ar
ac
ter
te
m
p
late
is
u
s
ed
to
m
o
r
p
h
o
lo
g
icall
y
d
eter
m
i
n
e
th
e
te
m
p
late
w
i
th
a
clo
s
est
co
r
r
elatio
n
m
atc
h
.
Fo
r
m
all
y
,
=
(
,
,
,
)
(
1
)
=
{
∈
(
,
)
}
(
2
)
W
he
r
e
:
U
is
the
s
e
t
of
uppe
r
c
a
s
e
Y
or
uba
c
ha
r
a
c
te
r
s
L
is
the
s
e
t
of
l
ow
e
r
c
a
s
e
Y
or
ub
a
c
ha
r
a
c
te
r
s
N
is
the
s
e
t
of
Y
or
uba
n
umb
e
r
s
P
is
the
s
e
t
of
Y
or
ub
a
pun
c
tua
tion
ma
r
k
s
T
h
e
tr
an
s
f
o
r
m
a
tio
n
f
u
n
ctio
n
o
n
ch
ar
ac
ter
is
:
:
→
ℎ
ℎ
I
n
th
e
c
h
ar
ac
ter
class
s
o
m
e
o
f
th
e
ch
ar
ac
ter
s
w
er
e
w
r
itte
n
i
n
d
if
f
er
e
n
t
w
a
y
s
i
n
o
r
d
er
to
ac
c
o
m
m
o
d
ate
d
if
f
er
e
n
t
w
a
y
s
o
f
w
r
iti
n
g
.
T
h
e
p
r
o
p
o
s
ed
Yo
r
u
b
a
OC
R
s
y
s
t
e
m
,
as
s
h
o
w
n
i
n
F
i
g
u
r
e
3
,
is
g
r
o
u
p
ed
in
to
th
r
ee
p
r
o
ce
s
s
in
g
le
v
els
w
h
ich
ar
e
lo
w
lev
el
p
r
o
ce
s
s
i
n
g
,
in
ter
m
e
d
iate
lev
el
an
d
h
ig
h
lev
el
p
r
o
ce
s
s
in
g
.
T
h
ese
ar
e
i
m
p
le
m
en
ted
u
s
i
n
g
6
4
-
b
it M
at
lab
v
er
s
io
n
7
.
8
.
0
.
3
8
7
an
d
th
e
in
p
u
t te
x
ts
ar
e
b
u
il
t
w
it
h
p
ain
t
b
r
u
s
h
a
n
d
tex
t.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
3
,
No
.
2
,
J
u
n
e
201
4
:
64
–
72
68
Fig
u
r
e
3
.
Sch
e
m
atic
o
f
t
h
e
o
f
f
-
li
n
e
Yo
r
u
b
a
Op
tical
C
h
ar
ac
te
r
R
ea
d
er
Fig
u
r
e
3
(
a
).
OC
R
h
a
n
d
w
r
itte
n
Yo
r
u
b
a
ch
ar
ac
ter
k
n
o
w
led
g
e
b
ase
Fig
u
r
e
3
(
b
).
OC
R
t
y
p
ed
Yo
r
u
b
a
ch
ar
ac
ter
k
n
o
w
led
g
e
b
ase
3
.
1
.
L
o
w
L
ev
el
P
ro
ce
s
s
ing
As
s
h
o
w
n
i
n
th
e
F
ig
u
r
e
3
,
lo
w
lev
el
p
r
o
ce
s
s
in
g
i
n
v
o
l
v
es
i
m
a
g
e
ac
q
u
is
itio
n
a
n
d
p
r
e
-
p
r
o
ce
s
s
in
g
o
f
th
e
ac
q
u
ir
ed
i
m
ag
e
s
.
I
m
a
g
e
ac
q
u
i
s
itio
n
s
ta
g
e
ac
q
u
ir
es
i
m
a
g
e
o
f
t
h
e
d
o
cu
m
e
n
t
o
r
ch
ar
ac
ter
s
to
b
e
r
ec
o
g
n
ized
.
Mo
s
t
ti
m
e
in
p
u
t
ch
ar
ac
ter
i
m
a
g
e
is
o
f
f
in
ite
r
eso
l
u
tio
n
w
h
ich
u
lt
i
m
at
el
y
a
f
f
ec
t
s
th
e
q
u
alit
y
o
f
its
tr
an
s
f
o
r
m
at
io
n
,
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
Yo
r
u
b
a
La
n
g
u
a
g
e
a
n
d
N
u
mera
ls
’
Offlin
e
I
n
terp
r
eter U
s
in
g
Mo
r
p
h
o
lo
g
ica
l
a
n
d
…
(
Ola
ka
n
mi
O.
Ola
d
a
yo
)
69
th
er
ef
o
r
e,
p
r
e
-
p
r
o
ce
s
s
i
n
g
b
ec
o
m
es
n
ec
es
s
ar
y
.
T
h
e
p
r
e
-
p
r
o
ce
s
s
i
n
g
s
ta
g
e
i
n
cl
u
d
es c
o
lo
u
r
n
o
r
m
aliza
tio
n
,
s
ca
li
n
g
f
ilter
i
n
g
a
n
d
t
h
i
n
n
i
n
g
.
C
o
lo
u
r
n
o
r
m
aliza
t
io
n
is
u
s
ed
to
ch
a
n
g
e
i
n
p
u
t
c
h
ar
ac
ter
f
o
r
eg
r
o
u
n
d
co
lo
u
r
to
b
lack
an
d
b
ac
k
g
r
o
u
n
d
co
lo
u
r
to
w
h
ite.
T
o
ac
h
iev
e
t
h
is
,
h
is
to
g
r
a
m
tec
h
n
iq
u
e
w
as
u
s
ed
.
T
h
e
in
p
u
t
ch
ar
ac
ter
w
a
s
u
s
ed
to
f
o
r
m
h
is
to
g
r
a
m
o
f
s
i
n
g
le
cla
s
s
w
h
ich
w
as
g
r
o
u
p
ed
in
to
i
n
ter
v
al
s
.
Ov
er
ea
c
h
o
f
t
h
ese
in
ter
v
a
ls
a
v
er
t
ical
r
ec
tan
g
le
i
s
d
r
a
w
n
w
i
th
i
ts
ar
e
a
p
r
o
p
o
r
tio
n
al
to
th
e
n
u
m
b
er
o
f
p
o
in
t f
alli
n
g
in
to
th
at
i
n
ter
v
a
l.
T
h
e
lu
m
i
n
a
n
ce
o
f
th
e
i
m
ag
e
w
a
s
d
eter
m
i
n
ed
u
s
i
n
g
E
q
u
atio
n
(
3
)
.
Fi
g
u
r
e
2
(
a
)
s
h
o
w
s
in
p
u
t
i
m
ag
e
b
e
f
o
r
e
n
o
r
m
ali
za
tio
n
w
h
ile
F
i
g
u
r
e
2
(
b
)
an
d
Fig
u
r
e
2
(
c
)
d
ep
ict
th
e
in
p
u
t i
m
a
g
e
af
ter
n
o
r
m
aliza
ti
o
n
an
d
f
il
ter
in
g
r
esp
ec
tiv
e
l
y
.
=
0
.
3
+
0
.
59
+
0
.
11
(
3
)
No
r
m
a
lizatio
n
al
g
o
r
ith
m
:
1.
Select
th
e
r
ele
v
an
t p
ar
t o
f
t
h
e
ch
ar
ac
ter
.
2.
Dete
r
m
i
n
e
th
e
t
h
r
es
h
o
ld
f
o
r
th
e
co
lo
u
r
n
o
r
m
a
lizatio
n
3.
P
r
o
ce
s
s
th
e
i
m
a
g
e
f
r
o
m
to
p
co
r
n
er
lin
e
b
y
li
n
e
4.
Sto
r
e
th
e
R
,
G,
B
v
al
u
e
o
f
ea
ch
p
ix
el
5.
Dete
r
m
i
n
e
u
s
in
g
eq
u
atio
n
1
6.
I
f
<
th
r
es
h
o
ld
v
al
u
e
th
e
n
t
u
r
n
t
h
e
p
ix
el
b
lack
o
t
h
er
w
is
e
w
h
ite
.
7.
R
ep
ea
t f
o
r
th
e
w
h
o
le
in
p
u
t i
m
ag
e
T
h
e
i
m
ag
e
s
ca
li
n
g
s
ca
les
t
h
e
i
n
p
u
t
c
h
ar
ac
ter
i
m
a
g
e
u
p
o
r
d
o
w
n
d
ep
en
d
in
g
o
n
t
h
e
o
r
ig
i
n
al
s
ize.
T
h
is
w
a
s
d
o
n
e
to
r
ed
u
ce
th
e
r
ec
o
g
n
itio
n
ti
m
e
an
d
er
r
o
r
r
ate
as
l
ar
g
e
ch
ar
ac
ter
i
m
ag
e
s
w
o
u
ld
tak
e
lo
n
g
er
ti
m
e
to
p
r
o
ce
s
s
w
h
ile
s
m
all
i
m
a
g
e
m
a
y
b
e
d
if
f
ic
u
lt
to
r
ec
o
g
n
ize.
Af
t
er
s
ca
lin
g
t
h
e
c
h
ar
ac
ter
b
ec
o
m
es
b
lo
ck
y
an
d
h
en
ce
th
e
s
m
o
o
t
h
en
i
n
g
f
ilter
i
n
g
s
tag
e
r
em
o
v
e
s
th
e
s
p
ik
e
ed
g
es.
T
h
is
s
tag
e
also
co
n
tai
n
s
s
m
o
o
t
h
e
n
in
g
f
i
lter
,
lo
w
p
ass
f
ilter
.
T
h
ese
f
ilter
s
ar
e
u
s
ed
to
r
ed
u
ce
b
lu
r
r
in
g
an
d
n
o
is
e.
A
ls
o
,
im
p
le
m
e
n
ted
in
th
e
lo
w
le
v
el
p
r
o
ce
s
s
in
g
is
t
h
e
th
i
n
n
i
n
g
w
h
ic
h
co
n
v
er
ts
a
n
y
elo
n
g
ated
p
ar
ts
o
r
s
tr
ip
s
in
th
e
i
m
a
g
e
r
eg
ar
d
les
s
o
f
th
e
ir
b
its
in
to
n
ar
r
o
w
s
tr
ip
s
th
at
ar
e
o
n
l
y
ab
o
u
t o
n
e
p
ix
el
w
id
e.
3
.
2
.
I
nte
r
m
edia
t
e
L
ev
el
P
ro
ce
s
s
i
ng
I
n
ter
m
ed
iate
L
e
v
el
P
r
o
ce
s
s
in
g
(
I
L
P
)
in
th
e
i
n
F
i
g
u
r
e
3
in
v
o
lv
e
s
i
m
a
g
e
r
o
tatio
n
a
n
d
s
e
g
m
e
n
tat
io
n
.
So
m
eti
m
es
in
p
u
t c
h
ar
ac
ter
i
m
ag
e
m
a
y
n
o
t b
e
p
r
o
p
er
ly
ali
g
n
ed
in
a
n
g
u
lar
f
as
h
io
n
w
it
h
r
es
p
ec
t to
th
e
c
h
ar
ac
ter
te
m
p
late
s
e
t.
A
n
i
n
s
tan
ce
o
f
t
h
is
w
i
ll
b
e
co
r
r
ec
ted
b
y
r
ea
li
g
n
th
e
i
m
a
g
e
O
C
R
.
Seg
m
e
n
tati
o
n
w
h
ic
h
f
o
r
m
s
th
e
co
r
e
o
f
I
L
p
r
o
ce
s
s
in
g
s
ta
g
e
p
ar
titi
o
n
s
t
h
e
in
p
u
t
i
m
ag
e
i
n
t
o
its
co
n
s
tit
u
e
n
t
ch
ar
ac
ter
s
.
S
h
o
w
n
b
elo
w
is
t
h
e
alg
o
r
ith
m
u
s
ed
f
o
r
s
e
g
m
e
n
tat
i
o
n
:
Seg
m
en
tatio
n
al
g
o
r
ith
m
:
1.
Scan
t
h
e
i
m
a
g
e
f
r
o
m
r
ig
h
t to
l
ef
t r
o
w
w
i
s
e
2.
A
d
d
an
d
co
u
n
t a
ll th
e
x
co
o
r
d
in
ates
3.
Dete
r
m
i
n
e
t
h
e
x
-
co
o
r
d
in
ate
o
f
t
h
e
ce
n
tr
o
id
u
s
i
n
g
=
∑
(
)
/
w
h
er
e
n
is
t
h
e
to
tal
n
u
m
b
er
o
f
th
e
ce
n
tr
o
id
.
4.
Dete
r
m
i
n
e
th
e
y
-
co
o
r
d
in
ate
o
f
th
e
ce
n
tr
o
id
u
s
i
n
g
=
∑
(
)
/
w
h
er
e
n
is
th
e
to
tal
n
u
m
b
er
o
f
th
e
ce
n
tr
o
id
.
3
.
3
.
Repre
s
ent
a
t
io
n a
nd
Descri
pt
io
n
R
ep
r
esen
tat
io
n
m
ap
s
t
h
e
s
c
an
n
ed
c
h
ar
ac
ter
i
m
a
g
e
to
f
o
r
m
s
u
i
tab
le
f
o
r
s
u
b
s
eq
u
e
n
t
co
m
p
u
ter
p
r
o
ce
s
s
in
g
w
h
ile
d
escr
ip
tio
n
i
s
a
f
ea
t
u
r
e
s
elec
tio
n
w
h
ic
h
d
ea
ls
w
it
h
ex
tr
ac
ti
n
g
f
ea
t
u
r
es
i
n
s
o
m
e
q
u
a
n
titati
v
e
m
an
n
er
o
r
d
if
f
er
en
tiati
n
g
o
n
e
class
o
f
o
b
j
ec
ts
f
r
o
m
a
n
o
th
er
.
T
h
is
w
as
ac
h
ie
v
ed
u
s
in
g
i
n
ter
n
al
c
h
ar
ac
ter
is
tic
s
,
th
at
is
,
t
h
e
p
ix
el
s
co
m
p
r
o
m
i
s
i
n
g
t
h
e
r
eg
io
n
.
3
.
4
.
K
no
w
ledg
e
B
a
s
e
T
h
e
k
n
o
w
led
g
e
b
ase
co
n
tai
n
s
th
e
n
u
m
b
er
s
,
p
u
n
ct
u
atio
n
,
u
p
p
er
an
d
lo
w
er
ca
s
e
s
o
f
Yo
r
u
b
a
alp
h
ab
ets
as
s
h
o
w
n
i
n
F
ig
u
r
e
1
an
d
2
.
I
t
i
s
b
asicall
y
a
d
atab
ase
o
f
ty
p
ed
an
d
h
an
d
w
r
itte
n
E
n
g
lis
h
alp
h
a
b
ets,
n
u
m
b
er
s
,
an
d
p
u
n
ct
u
at
io
n
s
.
I
n
d
iv
id
u
al
ch
ar
a
cter
im
a
g
es
i
n
th
e
k
n
o
w
led
g
e
b
ase
ar
e
u
s
ed
to
g
en
er
ate
th
e
co
r
r
elatio
n
v
alu
es
f
o
r
th
e
in
p
u
t c
h
ar
ac
ter
i
m
a
g
e
an
d
o
u
tp
u
t c
h
ar
ac
ter
te
x
t.
4.
T
E
ST
AN
D
CO
NCLU
SI
O
N
T
h
e
OC
R
s
y
s
te
m
w
a
s
s
u
b
j
e
cted
to
d
if
f
er
en
t
s
et
o
f
i
n
p
u
t
tex
t
i
m
ag
e
s
in
o
r
d
er
to
d
et
er
m
in
e
i
t
s
r
ec
o
g
n
itio
n
ef
f
icie
n
c
y
.
T
h
e
tes
t
w
a
s
ca
r
r
ied
o
u
t
o
n
b
o
t
h
t
y
p
ed
an
d
h
a
n
d
w
r
itte
n
i
n
p
u
t
te
x
t
s
.
T
h
e
i
n
p
u
t
i
m
a
g
es
a
s
s
h
o
w
n
i
n
F
ig
u
r
e
4
(
a
)
,
Fig
u
r
e
5
(
a
)
an
d
Fig
u
r
e
6
(
a
)
a
r
e
d
if
f
er
en
t
s
et
o
f
in
p
u
t
tex
t
s
cr
ea
ted
u
s
in
g
th
e
p
ain
t
b
r
u
s
h
as
p
en
an
d
p
ain
t
tex
t
w
h
ich
r
ep
r
esen
t
h
a
n
d
w
r
itte
n
an
d
t
y
p
ed
Yo
r
u
b
a
tex
ts
r
esp
ec
tiv
el
y
.
T
h
e
o
u
tp
u
ts
o
f
th
e
O
C
R
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
3
,
No
.
2
,
J
u
n
e
201
4
:
64
–
72
70
s
y
s
te
m
f
o
r
th
e
i
n
p
u
t
te
x
t
i
m
a
g
e
ar
e
s
h
o
w
n
i
n
F
i
g
u
r
e
4
(
b
)
,
Fig
u
r
e
5
(
b
)
an
d
Fig
u
r
e
6
(
b
)
.
T
h
e
test
r
esu
l
ts
w
er
e
q
u
ite
i
m
p
r
es
s
i
v
e.
I
t
w
as
o
b
s
er
v
ed
f
r
o
m
t
h
e
O
C
R
o
u
tp
u
t
in
F
ig
u
r
e
4
(
b
)
th
at
ch
ar
ac
ter
s
I
a
n
d
O
w
er
e
th
e
o
n
l
y
ch
ar
ac
ter
s
n
o
t
r
ec
o
g
n
ized
.
T
h
is
s
h
o
w
s
a
n
ac
cu
r
ac
y
o
f
8
6
%
f
o
r
th
e
t
y
p
ed
tex
t
w
it
h
ex
ec
u
tio
n
ti
m
e
o
f
1
1
2
ch
ar
/s
ec
r
ec
o
g
n
it
io
n
r
ate.
A
ls
o
,
f
o
r
in
p
u
t
tex
t
in
F
ig
u
r
e
5
(
b
)
it
w
as
o
b
s
er
v
ed
f
r
o
m
th
e
O
C
R
o
u
tp
u
t
in
F
ig
u
r
e
5
(
b
)
th
at
all
th
e
E
n
g
lis
h
n
u
m
b
e
r
s
w
er
e
co
r
r
ec
tl
y
r
ec
o
g
n
ized
an
d
co
n
v
er
ted
to
th
e
Yo
r
u
b
a
n
u
m
er
als.
T
h
is
s
h
o
w
ed
ac
cu
r
ac
y
o
f
1
0
0
%
f
o
r
th
e
n
u
m
er
als r
ec
o
g
n
itio
n
a
n
d
co
n
v
e
r
s
i
o
n
.
T
h
e
OC
R
s
y
s
te
m
o
u
tp
u
t i
n
F
ig
u
r
e
6
(
b
)
w
h
ic
h
r
ep
r
esen
ts
O
C
R
o
u
tp
u
t
f
o
r
th
e
h
an
d
w
r
itte
n
i
n
p
u
t
tex
t
in
Fi
g
u
r
e
6
(
a
)
,
also
r
ec
o
r
d
ed
an
ac
cu
r
ac
y
o
f
1
0
0
%.
I
t
w
as
o
b
s
er
v
ed
th
at
th
e
d
e
v
elo
p
ed
Yo
r
u
b
a
OC
R
s
y
s
te
m
’
s
p
er
f
o
r
m
a
n
ce
u
n
it
i
s
in
d
ep
en
d
e
n
t
an
d
co
n
s
ta
n
t
f
o
r
h
an
d
w
r
itte
n
a
n
d
t
y
p
ed
tex
t
i
m
ag
es
o
f
d
i
f
f
er
en
t
s
ize.
A
ls
o
,
th
e
r
esu
lt
s
h
o
w
ed
th
at
th
e
d
ev
e
l
o
p
ed
OC
R
s
y
s
te
m
m
o
r
e
e
f
f
ec
ti
v
el
y
r
ec
o
g
n
ized
n
u
m
er
al
s
th
a
n
alp
h
ab
et
s
.
Fig
u
r
e
4
(
a
).
OC
R
in
p
u
t o
f
a
s
c
an
n
ed
i
m
ag
e
te
x
t
d
o
cu
m
en
t
Fig
u
r
e
4
(
b
).
OC
R
o
u
tp
u
t o
f
t
h
e
s
ca
n
n
ed
i
m
a
g
e
tex
t
d
o
cu
m
en
t in
F
ig
u
r
e
4
(
a
)
Fig
u
r
e
5
(
a
).
OC
R
in
p
u
t o
f
a
s
c
an
n
ed
i
m
ag
e
te
x
t
d
o
cu
m
en
t
Fig
u
r
e
5
(
b
).
OC
R
o
u
tp
u
t o
f
t
h
e
s
ca
n
n
ed
i
m
a
g
e
tex
t
d
o
cu
m
en
t in
F
ig
u
r
e
5
(
a
)
Fig
u
r
e
6
(
a
).
OC
R
in
p
u
t o
f
a
s
c
an
n
ed
h
a
n
d
w
r
itte
n
i
m
a
g
e
tex
t d
o
cu
m
e
n
t
Fig
u
r
e
6
(
b
).
OC
R
o
u
tp
u
t o
f
a
s
ca
n
n
ed
h
a
n
d
w
r
it
ten
i
m
a
g
e
tex
t d
o
cu
m
e
n
t
RE
F
E
R
E
NC
E
S
[1
]
A
.
Ja
in
,
a
n
d
Ka
ru
,
K
.
P
a
g
e
S
e
g
m
e
n
tatio
n
Us
in
g
T
e
x
tu
re
A
n
a
l
y
sis,
Pa
tt
e
rn
Rec
o
g
n
it
io
n
.
v
o
l
29
,
p
a
g
e
7
4
3
-
7
7
0
.
2
0
0
6
[2
]
A
rd
it
i.
A
.
,
Ro
se
n
t
h
a
l,
B
.
De
v
e
lo
p
in
g
a
n
O
b
jec
ti
v
e
De
fi
n
it
i
o
n
o
f
V
i
su
a
l
Imp
a
irme
n
t.
P
ro
c
e
e
d
in
g
s
o
f
th
e
In
tern
a
ti
o
n
a
l
L
o
w
V
isio
n
,
p
p
3
3
1
-
3
3
4
,
1
9
9
8
[3
]
A
rt,
P
.
,
&
Rich
a
rd
,
R.
F
.
He
a
rin
g
b
y
Ba
ts H
a
n
d
b
o
o
k
o
f
A
u
d
i
to
ry
Re
se
a
rc
h
.
S
p
ri
n
g
e
r.
1
9
9
5
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
Yo
r
u
b
a
La
n
g
u
a
g
e
a
n
d
N
u
mera
ls
’
Offlin
e
I
n
terp
r
eter U
s
in
g
Mo
r
p
h
o
lo
g
ica
l
a
n
d
…
(
Ola
ka
n
mi
O.
Ola
d
a
yo
)
71
[4
]
Be
lo
te,
L
.
L
o
w
V
isio
n
Ed
u
c
a
ti
o
n
a
n
d
T
ra
in
in
g
:
De
f
in
in
g
t
h
e
Bo
u
n
d
a
ries
o
f
L
o
w
V
isio
n
P
a
ti
e
n
ts.
A
Per
so
n
a
l
G
u
id
e
to
t
h
e
VA
Vi
s
u
a
l
Imp
a
irme
n
t
S
e
rv
ice
s P
ro
g
ra
m
.
2
0
0
6
.
[5
]
Bru
n
o
,
P
.
P
o
we
r
Ultrso
u
n
d
i
n
El
e
c
tro
c
h
e
mistry
:
Fro
m
Ver
sa
ti
le
L
a
b
o
ra
t
o
ry
T
o
o
l
t
o
En
g
in
e
e
rin
g
S
o
l
u
ti
o
n
.
Jo
h
n
W
il
e
y
S
o
n
s.
2
0
1
2
.
[6
]
Co
rso
,
J.
B
o
n
e
-
Co
n
d
u
c
ti
o
n
T
h
re
sh
o
ld
s
f
o
r
S
o
n
ic
a
n
d
Ultras
o
n
ic
F
re
q
u
e
n
c
ies
.
J
o
u
rn
a
l
o
f
t
h
e
Aco
u
stica
l
S
o
c
iety
o
f
Ame
ric
a
,
3
5
.
1
9
6
3
.
[7
]
Da
m
b
h
a
re
,
S
.
,
&
A
.
S
a
k
h
a
re
.
S
m
a
rt
stick
f
o
r
Bli
n
d
:
Ob
sta
c
le De
tec
ti
o
n
,
A
rti
f
icia
l
v
isio
n
a
n
d
Re
a
l
-
ti
m
e
a
ss
istan
c
e
v
i
a
G
P
S
.
2
n
d
Na
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
In
f
o
rm
a
ti
o
n
a
n
d
C
o
mm
u
n
ica
ti
o
n
T
e
c
h
n
o
l
o
g
y
(
NCICT
)
,
p
p
.
3
1
-
3
3
,
2
0
1
1
.
[8
]
DIV
YA
,
S
.
,
NA
V
YA
,
B.
,
M
A
N
ASA
,
P
.
S
.
,
&
CHIT
RA
,
S
.
Ultra
s
o
n
ic
A
n
d
Vo
ice
B
a
se
d
W
a
lkin
g
S
ti
c
k
Fo
r
T
h
e
Bl
in
d
.
G
o
k
a
r
a
ju
Ra
n
g
a
ra
ju
In
st
it
u
te Of
En
g
in
e
e
rin
g
A
n
d
T
e
c
h
n
o
l
o
g
y
.
2
0
1
0
.
[9
]
DIV
YA
,
S
.
,
NA
V
YA
,
B.
,
M
A
N
ASA
,
P
.
S
.
,
&
CHIT
RA
,
S
.
Ultra
s
o
n
ic
A
n
d
Vo
ice
B
a
se
d
W
a
lkin
g
S
ti
c
k
Fo
r
T
h
e
Bl
in
d
.
G
o
k
a
r
a
ju
Ra
n
g
a
ra
ju
I
n
stit
u
te
Of
En
g
i
n
e
e
rin
g
A
n
d
T
e
c
h
n
o
lo
g
y
,
De
p
a
rtme
n
t
Of
El
e
c
tro
n
ics
A
n
d
Co
m
m
u
n
ica
ti
o
n
En
g
in
e
e
rin
g
.
2
0
1
0
.
[1
0
]
Du
e
ire
L
in
s,
R.
,
P
e
re
ira
S
il
v
a
,
G
.
,
&
G
o
m
e
s
e
S
il
v
a
,
A
.
A
ss
e
ss
in
g
a
n
d
Im
p
ro
v
in
g
t
h
e
Qu
a
li
ty
o
f
Do
c
u
m
e
n
t
Im
a
g
e
s
A
c
q
u
ired
w
it
h
P
o
rtab
le
Dig
it
a
l
C
a
m
e
ra
s.
ICDAR
2
0
0
7
.
N
in
t
h
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
.
Do
c
u
m
e
n
t
A
n
a
l
y
sis
a
n
d
Re
c
o
g
n
it
io
n
.
V
o
l
u
m
e
:
2
,
p
p
.
5
6
9
-
5
7
3
.
2
0
0
7
[1
1
]
F
u
k
u
n
a
g
a
,
K.
I
n
tro
d
u
c
ti
o
n
t
o
S
ta
t
isti
c
a
l
P
a
tt
e
rn
Rec
o
g
n
i
ti
o
n
.
1
9
9
0
.
[1
2
]
Hu
a
n
g
,
G
.
,
L
e
a
rn
e
d
-
M
il
ler,
E.
,
&
M
c
Ca
ll
u
m
,
A
.
(n
.
d
.
).
Cry
to
g
ra
m
De
c
o
d
in
g
f
o
r
Op
ti
c
a
l
Ch
a
ra
c
ter R
e
c
o
g
n
it
io
n
.
[1
3
]
Hu
a
n
g
,
K.,
S
u
n
,
J.,
H
o
tt
a
,
Y.,
&
F
u
ji
m
o
to
,
K.
A
n
S
VM
-
Ba
se
d
Hig
h
-
a
c
c
u
ra
te Rec
o
g
n
it
io
n
A
p
p
ro
a
c
h
f
o
r
Ha
n
d
w
rit
ten
Nu
m
e
r
a
ls
b
y
Us
in
g
Di
ff
e
re
n
c
e
F
e
a
tu
re
s.
ICDAR
Nin
t
h
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
Do
c
u
me
n
t
An
a
lys
is
a
n
d
Rec
o
g
n
it
io
n
,
v
o
l
2
,
p
p
.
5
8
9
-
5
9
3
,
2
0
0
7
.
[1
4
]
Isa
a
c
,
A
.
O.
T
h
e
S
e
a
rc
h
f
o
r
a
Yo
ru
b
a
Ort
h
o
g
ra
p
h
y
sin
c
e
t
h
e
1
8
4
0
s
Ob
sta
c
les
to
th
e
C
h
o
ice
o
f
th
e
A
ra
b
ic.
S
u
d
a
n
ic
Af
ric
a
,
7
7
-
1
0
2
,
2
0
0
3
.
[1
5
]
Ka
m
a
lj
it
,
K.,
&
Ba
lp
re
e
t,
K.
Ch
a
r
a
c
ter
Re
c
o
g
n
it
io
n
o
f
Hig
h
S
e
c
u
rit
y
Nu
m
b
e
r
P
late
s
Us
in
g
M
o
rp
h
o
l
o
g
ica
l
Op
e
ra
to
r.
In
ter
n
a
t
io
n
a
l
J
o
u
rn
a
l
o
f
C
o
mp
u
ter
S
c
ien
c
e
&
En
g
in
e
e
rin
g
T
e
c
h
n
o
l
o
g
y
(
IJ
C
S
ET
)
,
v
o
l
4
,
I
ss
u
e
5
,
M
a
y
2
0
1
3
.
[1
6
]
Ka
n
g
,
S
.
J.,
Yo
u
n
g
Ho
,
K.,
&
M
o
o
n
,
I.
H.
DEVEL
OP
M
ENT
OF
A
M
ECHA
T
R
ON
IC
BL
IN
D
S
T
IC
K.
Pro
c
e
e
d
in
g
s
o
f
t
h
e
2
0
0
1
IEE
E
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
Ro
b
o
ti
c
s
&
Au
to
ma
t
io
n
,
S
e
o
u
l,
K
o
re
a
.
pp
3
2
0
9
-
3
2
1
3
,
2
0
0
1
.
[1
7
]
Ku
n
d
u
,
A
.
,
M
IT
RE
Co
r
p
.
,
M
.
,
Hi
n
e
s,
T
.
,
P
h
i
ll
ip
s,
J.,
&
H
u
y
c
k
,
B.
A
ra
b
ic
Ha
n
d
w
rit
in
g
Re
c
o
g
n
it
io
n
Us
in
g
V
a
riab
le
Du
ra
ti
o
n
HMM
.
ICDAR
2
0
0
7
.
Ni
n
th
I
n
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
.
2.
Do
c
u
m
e
n
t
A
n
a
l
y
sis a
n
d
Re
c
o
g
n
it
io
n
.
2
0
0
7
.
[1
8
]
L
in
,
S
.
H
.
A
n
In
tr
o
d
u
c
ti
o
n
t
o
F
a
c
e
Re
c
o
g
n
it
io
n
T
e
c
h
n
o
lo
g
y
.
In
fo
rm
i
n
g
S
c
ien
c
e
sp
e
c
ia
l
issu
e
o
n
M
u
lt
i
me
d
ia
In
fo
rm
in
g
T
e
c
h
n
o
l
o
g
ies
,
V
o
l
.
3
(1
),
2
0
0
0
.
[1
9
]
M
o
h
a
m
m
e
d
,
A
.
Q.
(n
.
d
.
).
T
e
m
p
late
M
a
tch
in
g
M
e
th
o
d
f
o
r
Re
c
o
g
n
it
io
n
M
u
sn
a
d
Ch
a
ra
c
ters
b
a
se
d
o
n
Co
rre
lati
o
n
A
n
a
l
y
si
s.
[2
0
]
Na
d
e
e
m
,
D.,
&
Rizv
i,
S
.
Ch
a
ra
c
t
e
r
Rec
o
g
n
it
i
o
n
Us
in
g
T
e
mp
l
a
te M
a
tch
i
n
g
.
M
.
sc
P
r
o
jec
t.
[2
1
]
Na
w
a
z
,
T
.
,
Ha
ss
a
n
,
S
.
A
.
,
Na
q
v
i,
S
.
,
Re
h
m
a
n
,
H.
u
.
,
&
F
a
iz,
A
.
(n
.
d
.
).
Op
t
ica
l
Ch
a
ra
c
ter
Re
c
o
g
n
it
io
n
S
y
st
e
m
f
o
r
Urd
u
(Na
sk
h
F
o
n
t)
Us
i
n
g
P
a
tt
e
rn
M
a
tc
h
in
g
T
e
c
h
n
iq
u
e
.
[2
2
]
Nic
h
o
las
,
A
.
W
h
y
Us
e
th
e
lo
n
g
C
a
n
e
(1
st
E
d
it
io
n
e
d
.
).
De
x
ter.
1
9
9
5
.
[2
3
]
No
v
e
ll
in
e
,
R
.
S
q
u
ire
'
s
Fu
n
d
a
me
n
t
a
ls o
f
Ra
d
io
lo
g
y
(5
t
h
E
d
it
i
o
n
e
d
.
).
Ha
rv
a
rd
Un
iv
e
rsit
y
P
re
ss
.
1
9
9
7
.
[2
4
]
O
m
o
la
y
o
,
A
.
De
sig
n
a
n
d
Co
n
str
u
c
ti
o
n
o
f
a
M
u
lt
id
ime
n
sio
n
a
l
S
e
n
so
r
Bl
in
d
M
a
n
S
t
ick
.
B.
S
c
P
ro
jec
t,
Un
iv
e
rsity
o
f
Ib
a
d
a
n
,
El
e
c
tri
c
a
l
a
n
d
E
lec
tro
n
ic
En
g
in
e
e
rin
g
.
2
0
1
1
.
[2
5
]
P
ra
tap
,
R.
,
S
a
ty
a
p
ra
sa
d
,
L
.
,
&
S
a
str
y
,
A
.
M
id
d
le
Z
o
n
e
Co
m
p
o
n
e
n
t
Ex
trac
ti
o
n
a
n
d
Re
c
o
g
n
it
i
o
n
o
f
T
e
lu
g
u
.
ICDAR
2
0
0
7
,
N
in
t
h
I
n
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
Do
c
u
me
n
t
Ima
g
e
Do
c
u
m
e
n
t
An
a
lys
is a
n
d
Rec
o
g
n
it
i
o
n
.
vol
2
,
2
0
0
7
.
[2
6
]
Qin
g
,
C.
,
&
Em
i,
P
.
M
.
O
p
ti
c
a
l
C
h
a
ra
c
ter Rec
o
g
n
it
io
n
f
o
r
M
o
d
e
l
-
b
a
se
d
Ob
jec
t
Re
c
o
g
n
it
io
n
A
p
p
li
c
a
t
io
n
s.
[2
7
]
R.
M
.
K
S
i
n
h
a
,
e
.
Hy
b
rid
Co
n
tex
tu
a
l
T
e
x
t
Re
c
o
g
n
it
io
n
w
it
h
S
tri
n
g
m
a
tch
in
g
.
P
a
tt
e
rn
A
n
a
lys
is
a
n
d
M
a
c
h
in
e
In
telli
g
e
n
c
e
(
PA
M
I)
,
v
o
l
15
,
p
p
.
9
1
5
-
9
2
5
,
1
9
9
7
.
[2
8
]
S
a
q
ib
,
R.
,
A
sa
d
,
N.,
&
Om
e
r,
I.
Au
to
m
a
ted
N
u
mb
e
r
Pl
a
te
Rec
o
g
n
it
io
n
Us
in
g
Ho
u
g
h
L
i
n
e
s
a
n
d
T
e
mp
la
te
M
a
tc
h
in
g
.
P
r
o
c
e
e
d
in
g
s o
f
W
o
rld
C
o
n
g
re
ss
En
g
in
e
e
rin
g
a
n
d
C
o
m
p
u
ter S
c
ien
c
e
W
CECS
.
2
0
1
2
.
[2
9
]
S
h
riv
a
sta
v
a
,
K.,
V
e
rm
a
,
A
.
,
&
S
i
n
g
h
,
S
.
P
.
Dista
n
c
e
M
e
a
su
re
m
e
n
t
o
f
a
n
Ob
jec
t
o
r
Ob
sta
c
le
b
y
Ultr
a
so
u
n
d
S
e
n
so
rs
u
sin
g
P
8
9
C5
1
RD2
.
I
n
ter
n
a
t
io
n
a
l
J
o
u
rn
a
l
o
f
C
o
mp
u
ter
T
h
e
o
ry
a
n
d
En
g
in
e
e
rin
g
,
v
o
l
2
(
1
)
,
2
0
1
0
.
[3
0
]
S
u
n
,
J.,
Hu
a
n
g
,
K
.
,
H
o
tt
a
,
Y.
,
&
F
u
ji
m
o
to
,
K
.
De
g
r
a
d
e
d
Ch
a
r
a
c
ter
Rec
o
g
n
it
io
n
b
y
Co
m
p
lem
e
n
ta
ry
Cl
a
ss
if
ier
s
Co
mb
in
a
ti
o
n
.
ICDA
R
.
Nin
th
I
n
te
rn
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
.
2.
Do
c
u
m
e
n
t
A
n
a
l
y
sis a
n
d
Re
c
o
g
n
it
io
n
.
2
0
0
7
.
[3
1
]
T
a
k
e
d
a
,
S
.
,
M
o
r
io
k
a
,
I.
,
M
iy
a
sh
it
a
,
K.,
Ok
u
m
u
ra
,
A
.
,
Yo
sh
id
a
,
Y.,
&
M
a
tsu
m
o
to
,
K.
A
g
e
V
a
riatio
n
i
n
t
h
e
Up
p
e
r
L
im
it
o
f
He
a
rin
g
.
Eu
ro
p
e
a
n
J
o
u
rn
a
l
o
f
Ap
p
li
e
d
P
h
y
sio
lo
g
y
,
v
o
l
65
,
(
5
)
1
9
9
2
.
[3
2
]
Ullm
a
n
n
,
J.
Ap
p
li
c
a
ti
o
n
o
f
Pa
t
ter
n
Rec
o
g
n
it
i
o
n
.
CRC
P
re
ss
,
I
n
c
.
1
9
8
7
.
[3
3
]
Vi
rtu
a
l
W
a
lkin
g
S
ti
c
k
fo
r
t
h
e
Vi
su
a
ll
y
Imp
a
ire
d
.
Re
tri
e
v
e
d
Ja
n
u
a
ry
2
0
1
4
,
h
tt
p
:
//
c
o
n
tes
t.
tec
h
b
rief
s.c
o
m
/2
0
1
2
/en
tri
e
s/m
e
d
ica
l/
2
7
4
6
.
2
0
1
0
[3
4
]
W
a
h
a
b
,
M
.
H.,
T
a
li
b
,
A
.
A
.
,
Ka
d
i
r,
H.
A
.
,
J
o
h
a
ri,
A
.
,
A
.
No
ra
z
iah
,
S
id
e
k
,
R.
M
.
,
e
t
a
l.
S
m
a
rt
Ca
n
e
:
A
ss
isti
v
e
Ca
n
e
f
o
r
V
isu
a
ll
y
-
im
p
a
ired
P
e
o
p
le.
IJ
CS
I
I
n
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
C
o
mp
u
ter
S
c
ien
c
e
s,
8
(
4
)
,
2
0
1
1
.
[3
5
]
Yin
,
X
.
-
C.
,
S
u
n
,
J.,
Na
o
i,
S
.
,
&
F
u
ji
m
o
to
,
K.
A
M
u
lt
i
-
S
tag
e
S
trate
g
y
to
P
e
rsp
e
c
ti
v
e
Re
c
ti
f
i
c
a
ti
o
n
f
o
r
M
o
b
i
le
P
h
o
n
e
Ca
m
e
ra
-
B
a
se
d
Do
c
u
m
e
n
t
I
m
a
g
e
s
.
2
0
0
7
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
3
,
No
.
2
,
J
u
n
e
201
4
:
64
–
72
72
O.O
Ola
k
a
n
m
i
re
c
e
i
v
e
d
th
e
B.
T
e
c
h
i
n
C
o
m
p
u
ter
E
n
g
in
e
e
rin
g
f
ro
m
L
a
d
o
k
e
Ak
in
to
la
Un
iv
e
rsity
o
f
T
e
c
h
n
o
lo
g
y
,
Og
b
o
m
o
sh
o
2
0
0
0
a
n
d
M
.
sc
in
Co
m
p
u
ter
S
c
ien
c
e
f
ro
m
Un
iv
e
rsity
o
f
Ib
a
d
a
n
,
I
b
a
d
a
n
.
He
is
a
P
h
D
stu
d
e
n
t
in
t
h
e
De
p
a
rtme
n
t
o
f
El
e
c
tri
c
a
l
&
El
e
c
tro
n
ic
En
g
in
e
e
r
in
g
,
Un
iv
e
rsit
y
o
f
Ib
a
d
a
n
a
n
d
m
a
jo
r
in
P
a
ra
ll
e
l
&
Distrib
u
t
e
d
Co
m
p
u
ti
n
g
.
C
u
rre
n
tl
y
,
h
e
is
a
lec
tu
re
r
in
t
h
e
De
p
a
rtm
e
n
t
o
f
El
e
c
tri
c
a
l
&
El
e
c
tro
n
ic E
n
g
in
e
e
rin
g
.
Evaluation Warning : The document was created with Spire.PDF for Python.