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.
11
,
No
.
3
,
Sep
tem
b
er
201
8
,
p
p
.
1
1
2
9
~
1
1
3
5
I
SS
N:
2502
-
4752
,
DOI
:
1
0
.
1
1
5
9
1
/
i
j
ee
cs
.
v
11
.i
3
.
p
p
1
1
2
9
-
1
1
3
5
1129
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
co
r
e.
co
m/jo
u
r
n
a
ls
/in
d
ex
.
p
h
p
/
ijeec
s
C
o
ntact Lens
C
la
ss
ificatio
n by
U
si
ng
S
eg
m
e
nted
L
e
ns
B
o
unda
ry
F
ea
tur
es
Nur
Arif
f
in M
o
hd
Z
in
1
,
H
is
ha
mm
u
dd
in A
s
m
u
ni
2
,
H
a
za
Nuzly
Abdu
l H
a
m
e
d
3
,
Ra
zib
M
.
O
t
h
m
a
n
4
,
Sh
a
hree
n
K
a
s
i
m
5
,
Ro
ha
y
a
nti
H
a
s
s
a
n
6
,
Z
a
l
m
iy
a
h
Z
a
k
a
ria
7
,
Ro
s
f
uza
h Ro
s
la
n
8
1
,
5,
8
S
o
f
t
Co
m
p
u
ti
n
g
a
n
d
Da
ta M
in
in
g
Ce
n
tre
,
F
a
c
u
lt
y
o
f
Co
m
p
u
ter S
c
ien
c
e
a
n
d
In
f
o
rm
a
ti
o
n
T
e
c
h
n
o
l
o
g
y
,
Un
iv
e
rsiti
T
u
n
Hu
ss
e
in
O
n
n
M
a
lay
sia
2
,
3,
4,
6
,
7
L
a
b
o
ra
to
ry
o
f
Co
m
p
u
tatio
n
a
l
In
telli
g
e
n
c
e
a
n
d
Im
a
g
e
P
ro
c
e
ss
in
g
,
F
a
c
u
lt
y
o
f
Co
m
p
u
ti
n
g
,
Un
iv
e
rsiti
T
e
k
n
o
lo
g
i
M
a
lay
sia
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ma
y
2
,
2
0
1
8
R
ev
i
s
ed
J
u
n
9
,
201
8
A
cc
ep
ted
J
u
n
2
1
,
2
0
1
8
A
R
e
c
e
n
t
stu
d
ies
h
a
v
e
sh
o
w
n
t
h
a
t
th
e
w
e
a
rin
g
o
f
so
f
t
len
s
m
a
y
lea
d
to
p
e
rf
o
r
m
a
n
c
e
d
e
g
ra
d
a
ti
o
n
w
it
h
th
e
in
c
re
a
se
o
f
f
a
lse
r
e
jec
t
ra
te.
Ho
w
e
v
e
r,
d
e
tec
ti
n
g
th
e
p
re
se
n
c
e
o
f
so
f
t
l
e
n
s
is
a
n
o
n
-
tri
v
ial
tas
k
a
s
it
s
t
e
x
tu
re
th
a
t
a
lm
o
st
in
d
isc
e
rn
ib
le.
I
n
t
h
is
w
o
rk
,
w
e
p
ro
p
o
se
d
a
c
las
sif
ica
ti
o
n
m
e
th
o
d
t
o
id
e
n
ti
f
y
th
e
e
x
isten
c
e
o
f
so
f
t
len
s
in
iri
s
im
a
g
e
.
Ou
r
p
ro
p
o
se
d
m
e
th
o
d
sta
rts
w
it
h
se
g
m
e
n
ti
n
g
th
e
len
s
b
o
u
n
d
a
r
y
o
n
to
p
o
f
th
e
sc
lera
re
g
io
n
.
T
h
e
n
,
th
e
se
g
m
e
n
ted
b
o
u
n
d
a
ry
is
u
se
d
a
s
fe
a
tu
re
s
a
n
d
e
x
trac
ted
b
y
lo
c
a
l
d
e
sc
rip
to
rs.
T
h
e
se
f
e
a
tu
re
s
a
re
th
e
n
train
e
d
a
n
d
c
las
sif
ied
u
sin
g
S
u
p
p
o
rt
V
e
c
to
r
M
a
c
h
in
e
s.
T
h
is
m
e
th
o
d
w
a
s
tes
t
e
d
o
n
No
tre
Da
m
e
Co
s
m
e
ti
c
Co
n
tac
t
L
e
n
s
2
0
1
3
d
a
tab
a
se
.
Ex
p
e
rim
e
n
t
sh
o
w
e
d
th
a
t
th
e
p
r
o
p
o
se
d
m
e
th
o
d
p
e
rf
o
r
m
e
d
b
e
tt
e
r
th
a
n
sta
te o
f
th
e
a
rt
m
e
th
o
d
s
K
ey
w
o
r
d
s
:
C
o
n
tact
le
n
s
cla
s
s
i
f
icatio
n
L
o
ca
l d
escr
ip
to
r
Su
p
p
o
r
t V
ec
to
r
Ma
ch
in
es
Co
p
y
rig
h
t
©
2
0
1
8
In
stit
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
:
Nu
r
A
r
if
f
i
n
Mo
h
d
Z
in
,
So
f
t
C
o
m
p
u
ti
n
g
an
d
Data
Mi
n
in
g
C
e
n
tr
e
,
Facu
lt
y
o
f
C
o
m
p
u
ter
Scien
ce
an
d
I
n
f
o
r
m
atio
n
T
ec
h
n
o
lo
g
y
,
Un
i
v
er
s
iti T
u
n
H
u
s
s
ei
n
On
n
Ma
la
y
s
ia
.
E
m
ail:
ar
i
f
f
in
@
u
th
m
.
ed
u
.
m
y
1.
I
NT
RO
D
UCT
I
O
N
Sin
ce
t
h
e
f
o
u
n
d
atio
n
o
f
ir
is
r
ec
o
g
n
itio
n
d
ev
i
s
ed
b
y
Da
u
g
m
an
[
1
]
,
it
is
t
h
en
ca
te
g
o
r
ized
u
n
d
er
th
e
m
o
s
t
r
eliab
le
b
io
m
etr
ics
e
x
is
t
[
2
-
4
]
.
T
h
is
is
s
u
p
p
o
r
ted
b
y
W
an
g
an
d
Ha
n
[
5
]
w
h
o
s
e
s
t
ated
3
tr
aits
o
f
ir
is
b
io
m
etr
ic
s
;
Fir
s
tl
y
,
ir
is
i
s
an
i
n
ter
n
al
b
o
d
y
w
h
ic
h
p
r
o
tecte
d
,
s
ec
o
n
d
l
y
,
it
ca
n
n
o
t
ea
s
il
y
b
ei
n
g
a
lte
r
ed
w
i
th
o
u
t
d
am
a
g
i
n
g
it
s
v
is
io
n
a
n
d
la
s
tl
y
t
h
e
d
ilat
io
n
d
u
e
to
il
lu
m
i
n
a
tio
n
r
es
p
o
n
s
e
m
a
k
es
i
t
h
ar
d
t
o
i
m
itate.
A
lt
h
o
u
g
h
b
ein
g
co
n
s
id
er
ed
as
r
eliab
le,
th
er
e
ar
e
s
till
o
p
en
is
s
u
e
s
in
i
r
is
r
ec
o
g
n
itio
n
.
Am
o
n
g
th
e
m
,
o
n
e
th
at
co
u
ld
b
e
h
o
w
th
e
r
ec
o
g
n
itio
n
r
ea
cts
to
w
ar
d
s
t
h
e
w
ea
r
in
g
o
f
co
n
tact
l
en
s
,
w
h
et
h
er
it
is
s
o
f
t
o
r
co
s
m
etic
len
s
.
T
h
is
i
s
s
u
e
h
as
b
ee
n
th
o
r
o
u
g
h
l
y
s
t
u
d
ied
in
[
6
]
an
d
[
7
]
,
th
at
ap
ar
t
f
r
o
m
co
s
m
etic
le
n
s
,
s
o
f
t
len
s
ca
n
l
ea
v
e
s
i
g
n
if
ican
t
d
eg
r
ad
atio
n
d
u
r
in
g
v
er
if
icatio
n
.
Fro
m
th
eir
f
in
d
i
n
g
s
,
th
e
f
a
l
s
e
r
ej
ec
t
r
ate
f
o
r
m
atc
h
in
g
th
e
s
a
m
e
s
u
b
j
ec
t,
b
y
co
n
s
id
er
in
g
w
it
h
s
o
f
t
le
n
s
as
p
r
o
b
e
im
ag
e
a
n
d
w
it
h
o
u
t
s
o
f
t
len
s
a
s
g
al
ler
y
i
m
ag
e
h
a
s
r
esu
lted
in
5
.
6
6
%
FR
R
.
As
a
co
m
p
ar
is
o
n
,
m
atc
h
i
n
g
w
it
h
o
u
t
len
s
an
d
s
o
f
t
le
n
s
f
o
r
b
o
th
g
aller
y
an
d
s
e
n
s
o
r
i
m
a
g
es
o
n
l
y
r
esu
lted
in
1
.
1
7
% a
n
d
1
.
6
7
% r
esp
ec
tiv
el
y
.
T
o
h
an
d
le
th
e
e
m
er
g
in
g
o
f
1
2
5
m
il
lio
n
co
n
tac
t
len
s
w
ea
r
e
r
s
[
8
]
,
it
is
cr
u
cial
th
at
o
n
e
r
ec
o
g
n
itio
n
s
y
s
te
m
s
h
o
u
ld
h
a
v
e
an
ea
r
l
y
s
tag
e
m
ec
h
a
n
is
m
to
d
etec
t
th
e
p
r
esen
ce
o
f
co
n
tact
len
s
[
9
]
,
an
d
w
h
a
t
t
y
p
e
th
e
len
s
i
s
.
I
t
is
also
s
tatis
t
icall
y
p
r
o
v
en
th
at
a
r
ec
o
g
n
itio
n
s
y
s
te
m
p
er
f
o
r
m
a
n
ce
d
eg
r
ad
es
w
h
ile
r
ec
o
g
n
izi
n
g
co
n
tact
len
s
s
u
b
j
ec
ts
[
1
0
]
,
[
1
1
]
.
Un
lik
e
co
s
m
etic
len
s
,
s
o
f
t
l
en
s
i
s
w
o
r
n
to
c
o
r
r
ec
t
ey
e
v
is
io
n
r
ath
er
th
a
n
f
o
r
ap
p
ea
r
an
ce
p
u
r
p
o
s
e
.
So
f
t
len
s
is
u
s
u
all
y
co
lo
u
r
les
s
,
w
h
ile
co
s
m
etic
len
s
m
a
y
ap
p
ea
r
in
w
id
e
v
ar
iet
y
o
f
co
lo
u
r
s
.
B
esid
es,
s
o
f
t
len
s
is
i
m
p
er
ce
p
tib
le
u
n
les
s
b
ee
n
i
n
s
p
ec
ted
ca
r
ef
u
ll
y
.
He
n
ce
,
d
etec
tio
n
o
f
s
o
f
t
len
s
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
.
11
,
No
.
3
,
Sep
tem
b
er
201
8
:
1
1
2
9
–
1
1
3
5
1130
w
o
u
ld
b
e
a
n
o
n
-
tr
i
v
ial
ta
s
k
t
h
at
r
eq
u
ir
es
a
m
a
th
e
m
atica
l
a
n
d
s
tatis
tical
ap
p
r
o
ac
h
to
a
u
to
m
at
i
n
g
t
h
e
ta
s
k
.
T
h
e
p
r
esen
ce
o
f
s
o
f
t
le
n
s
ca
n
b
e
n
o
ticed
f
o
r
o
n
e
d
is
tin
ct
f
ea
t
u
r
e
,
w
h
ich
i
s
th
e
le
n
s
b
o
u
n
d
ar
y
lo
ca
ted
o
n
to
p
o
f
th
e
s
cler
a
r
eg
io
n
.
W
e
m
a
y
a
s
s
u
m
e
th
is
b
o
u
n
d
ar
y
ca
n
b
e
ea
s
il
y
d
etec
ted
,
h
o
w
e
v
er
,
it
co
m
p
r
is
es
o
f
v
er
y
t
h
i
n
li
n
e
an
d
ca
n
b
e
ea
s
il
y
co
n
f
u
s
ed
b
y
n
o
is
e
r
es
u
lted
f
r
o
m
i
n
co
n
s
i
s
te
n
t
ill
u
m
in
at
io
n
.
I
n
t
h
is
w
o
r
k
,
w
e
ta
k
e
th
e
ch
alle
n
g
e
to
s
eg
m
e
n
t
th
e
le
n
s
b
o
u
n
d
ar
y
an
d
u
s
e
it
as
f
ea
tu
r
e
s
,
th
en
,
tr
ain
it
u
s
i
n
g
S
u
p
p
o
r
t
Vec
to
r
Ma
ch
in
e,
in
o
r
d
er
to
class
if
y
b
et
w
ee
n
w
ith
o
r
w
it
h
o
u
t
s
o
f
t
len
s
.
T
h
e
n
o
v
e
lt
y
o
f
o
u
r
w
o
r
k
i
s
t
h
e
f
u
s
io
n
o
f
ex
tr
ac
ted
f
ea
t
u
r
e
s
p
r
o
d
u
ce
s
f
r
o
m
t
w
o
p
r
o
m
i
n
e
n
t
d
escr
ip
to
r
s
,
Hi
s
to
g
r
a
m
o
f
Gr
ad
ien
t
(
HO
G)
an
d
Scale
I
n
v
ar
ia
n
t
Feat
u
r
e
T
r
an
s
f
o
r
m
(
SIFT
)
.
T
h
e
d
eta
ils
o
f
o
u
r
w
o
r
k
ar
e
r
ep
r
esen
t
ed
in
t
h
e
f
o
llo
w
in
g
s
ec
tio
n
s
.
I
n
Sectio
n
2
,
an
y
p
r
ev
io
u
s
w
o
r
k
s
r
eg
ar
d
i
n
g
co
n
tact
len
s
d
etec
tio
n
ar
e
d
is
cu
s
s
ed
.
I
n
Sectio
n
3
,
o
u
r
p
r
o
p
o
s
ed
m
eth
o
d
is
d
escr
ib
ed
in
d
etail
s
.
T
h
e
ex
p
e
r
i
m
en
tal
r
esu
lt
s
an
d
d
is
cu
s
s
io
n
s
ar
e
r
ep
o
r
te
d
in
Sectio
n
4
a
n
d
las
tl
y
,
Sectio
n
5
d
r
a
w
s
t
h
e
co
n
cl
u
s
io
n
.
2.
RE
L
AT
E
D
WO
RK
S
Du
r
in
g
d
ec
ad
es,
co
s
m
etic
le
n
s
h
as
g
ain
ed
a
lo
t
o
f
atten
tio
n
in
ir
is
r
ec
o
g
n
i
tio
n
co
m
m
u
n
it
y
.
A
lo
t
o
f
r
esear
ch
h
a
s
b
ee
n
co
n
d
u
c
ted
ex
ten
s
i
v
el
y
m
ain
l
y
u
n
d
er
th
e
s
u
b
j
ec
t
of
f
ak
e
ir
is
d
etec
tio
n
an
d
ir
is
s
p
o
o
f
in
g
.
I
t
w
a
s
p
io
n
ee
r
ed
b
y
Dau
g
m
an
[
12
]
w
h
o
m
a
n
ag
ed
to
d
ete
ct
d
o
t
m
atr
i
x
co
s
m
etic
len
s
u
s
i
n
g
Fo
u
r
ier
tr
an
s
f
o
r
m
.
T
h
en
,
L
ee
et
a
l.
[
1
3
]
in
tr
o
d
u
c
ed
th
e
u
s
e
o
f
P
u
r
k
i
n
j
e
i
m
a
g
e
to
d
etec
t
f
a
k
e
ir
is
.
T
h
e
r
esear
ch
co
n
tin
u
es
w
h
er
e
He
et
al.
[
12
]
u
tili
ze
d
g
r
a
y
-
le
v
el
co
-
o
cc
u
r
r
en
ce
m
atr
i
x
(
GL
C
M)
as
f
ea
t
u
r
e
d
escr
ip
to
r
an
d
SVM
as
class
i
f
ier
.
T
h
ey
r
ep
o
r
ted
1
0
0
%
ac
cu
r
ac
y
test
ed
o
n
s
elf
-
d
atab
ase.
Me
an
w
h
ile,
W
ei
et
al.
[
15
]
p
r
o
p
o
s
e
d
th
r
ee
m
et
h
o
d
s
f
o
r
co
n
tact
le
n
s
d
etec
tio
n
;
ir
is
ed
g
e
s
h
ar
p
n
es
s
,
ir
is
-
te
x
to
n
s
an
d
GL
C
M
w
it
h
SVM.
E
v
a
lu
at
io
n
is
d
o
n
e
b
y
u
s
i
n
g
C
A
SI
A
a
n
d
B
A
T
H
d
atab
ase
w
it
h
ac
c
u
r
ac
y
ac
h
iev
ed
ab
o
v
e
7
6
.
8
%.
Z
h
a
n
g
et
al.
[
16
]
u
s
ed
SIFT
-
w
ei
g
h
te
d
L
o
ca
l
B
in
ar
y
P
atter
n
w
i
th
S
VM
as
cla
s
s
i
f
ier
.
T
h
e
y
ac
h
i
ev
ed
9
9
%
ac
cu
r
ac
y
u
s
i
n
g
s
e
lf
-
d
atab
ase.
Un
lik
e
co
s
m
etic
len
s
,
s
o
f
t
le
n
s
h
a
s
g
ain
ed
les
s
atte
n
tio
n
i
n
t
h
e
co
m
m
u
n
it
y
.
T
h
is
is
d
u
e
to
t
h
e
b
eliev
e
t
h
at
s
o
f
t
le
n
s
w
ea
r
i
n
g
d
o
es
n
o
t
ca
u
s
e
s
ig
n
i
f
ican
t
i
m
p
ac
t
o
f
d
eg
r
ad
atio
n
d
u
r
in
g
w
ea
r
i
n
g
as
s
u
p
p
o
r
ted
in
[
1
7
-
19]
.
Ho
w
ev
er
,
th
e
a
w
ar
en
e
s
s
o
f
t
h
e
s
o
f
t le
n
s
’
s
w
ea
r
i
n
g
i
m
p
ac
t i
n
[
6
]
,
[
20
]
h
as ig
n
ited
m
o
r
e
r
esear
ch
e
s
b
ein
g
co
n
d
u
cted
.
T
h
er
e
ar
e
th
r
ee
ap
p
r
o
ac
h
es
to
s
o
f
t
le
n
s
d
etec
tio
n
,
w
h
et
h
er
it
is
h
ar
d
w
ar
e,
m
ac
h
in
e
lear
n
i
n
g
or
i
m
a
g
e
s
eg
m
e
n
tatio
n
ap
p
r
o
ac
h
.
Har
d
w
ar
e
ap
p
r
o
ac
h
r
eq
u
ir
es
th
e
u
s
e
o
f
s
o
p
h
is
ticated
ca
m
er
a.
Su
c
h
ex
a
m
p
les
ar
e
f
r
o
m
K
y
w
e
et
al.
[
21
]
w
h
er
e
th
e
y
u
s
ed
a
th
er
m
al
ca
m
er
a
to
m
ea
s
u
r
e
th
e
d
ec
r
e
m
e
n
t
o
f
te
m
p
er
atu
r
e
o
n
th
e
e
y
e
s
u
r
f
ac
e
d
u
r
in
g
th
e
b
li
n
k
in
g
o
f
th
e
e
y
e.
T
h
e
y
o
b
s
er
v
ed
th
at
a
ce
r
tain
d
eg
r
ee
o
f
d
ec
r
e
m
en
t
in
d
icate
s
th
e
w
ea
r
i
n
g
o
f
s
o
f
t
len
s
.
An
o
th
er
w
o
r
k
b
y
L
ee
et
a
l.
[
22
]
claim
e
d
th
at
P
u
r
k
i
n
j
e
i
m
ag
e
s
b
et
w
ee
n
o
r
ig
i
n
al
an
d
le
n
s
w
o
r
n
ir
i
s
ar
e
d
if
f
er
en
ce
.
T
h
es
e
i
m
a
g
es
ar
e
ca
p
tu
r
ed
u
s
i
n
g
t
w
o
co
lli
m
ated
I
R
-
L
E
D
ca
m
er
as.
R
ec
en
t
w
o
r
k
b
y
Hu
g
h
e
s
a
n
d
B
o
w
y
er
[
23
]
d
etec
t
th
e
p
r
esen
ce
o
f
len
s
b
y
u
s
in
g
s
ter
eo
v
is
io
n
f
r
o
m
t
w
o
ca
m
er
as.
So
f
t
len
s
w
ea
r
i
n
g
i
s
d
etec
ted
if
t
h
e
ca
p
tu
r
ed
i
m
a
g
e
s
ee
n
a
s
cu
r
v
ed
s
u
r
f
ac
e
r
ath
er
t
h
an
f
lat
(
w
it
h
o
u
t l
en
s
)
.
Me
an
w
h
ile
,
m
ac
h
i
n
e
lear
n
i
n
g
r
eq
u
ir
es
f
ea
tu
r
es
d
escr
ip
to
r
an
d
class
i
f
ier
to
p
er
f
o
r
m
.
Do
y
l
e
et
al.
[
9
]
u
s
ed
a
m
o
d
i
f
ied
L
o
ca
l
B
in
ar
y
P
atter
n
(
L
B
P
)
as
f
ea
t
u
r
es
d
escr
ip
to
r
an
d
ex
p
er
i
m
e
n
ted
w
it
h
1
4
d
if
f
er
en
t
class
i
f
ier
s
.
T
h
e
y
ac
h
iev
ed
9
6
.
5
%
o
f
co
r
r
ec
t
class
if
icatio
n
f
o
r
co
s
m
etic
le
n
s
.
Ho
w
e
v
er
,
o
n
l
y
5
0
.
2
5
%
co
r
r
ec
t
class
i
f
icatio
n
f
o
r
s
o
f
t
le
n
s
.
Ko
h
li
et
al.
i
n
[
24
]
ex
p
er
i
m
e
n
ted
w
i
th
f
o
u
r
m
et
h
o
d
s
;
ir
is
ed
g
e
s
h
ar
p
n
e
s
s
,
te
x
t
u
r
al
f
ea
t
u
r
es
b
ased
o
n
co
-
o
cc
u
r
r
en
ce
m
atr
i
x
,
g
r
a
y
le
v
el
co
-
o
cc
u
r
r
en
ce
m
atr
ix
a
n
d
L
B
P
w
ith
S
VM
.
T
h
ey
r
ep
o
r
ted
th
at
L
B
P
w
it
h
SVM
h
as
i
n
f
er
r
ed
th
e
b
est
r
esu
l
t
in
o
v
er
all.
H
o
w
e
v
er
,
o
n
l
y
5
4
.
8
%
C
C
R
ac
h
iev
ed
f
o
r
s
o
f
t
len
s
.
L
ater
,
Yad
av
et
al.
[
7
]
ex
ten
d
th
e
w
o
r
k
i
n
[
9
]
an
d
[
24
]
w
it
h
ad
d
itio
n
al
d
atab
ase
an
d
r
ev
is
ed
alg
o
r
ith
m
.
T
h
e
y
ac
h
iev
ed
C
C
R
ab
o
v
e
4
5
.
3
5
%.
Gr
ag
n
a
n
iello
et
al.
in
[
25
],
[
26
]
u
s
ed
s
cler
a
a
n
d
ir
is
r
eg
i
o
n
as
f
ea
t
u
r
es
a
n
d
ap
p
lied
Scale
I
n
v
ar
ian
t
Descr
ip
to
r
as
f
ea
tu
r
e
d
escr
ip
to
r
w
it
h
SVM
as
clas
s
i
f
ier
.
T
h
ey
r
e
p
o
r
ted
C
C
R
ab
o
v
e
7
6
.
2
9
%.
R
ag
h
a
v
en
d
r
a
e
t
al.
[
27
]
p
r
o
p
o
s
ed
th
e
u
s
e
o
f
B
in
ar
iz
ed
Statis
tical
I
m
a
g
e
Fea
tu
r
es
(
B
SIF)
w
ith
SV
M.
T
h
ey
ac
h
ie
v
ed
6
2
%
f
o
r
in
tr
a
-
s
en
s
o
r
an
d
5
4
%
ac
cu
r
ac
y
f
o
r
in
ter
-
s
e
n
s
o
r
.
Sil
v
a
et
al.
[
29
]
u
s
ed
C
o
n
v
o
l
u
tio
n
a
l
Neu
r
al
Net
w
o
r
k
.
T
h
e
y
r
es
u
lte
d
in
6
5
% f
o
r
in
tr
a
-
s
en
s
o
r
an
d
4
2
.
2
5
% f
o
r
in
ter
-
s
en
s
o
r
.
On
t
h
e
o
th
er
h
a
n
d
,
i
m
a
g
e
s
e
g
m
en
tatio
n
ap
p
r
o
ac
h
u
s
es
ed
g
e
d
etec
tio
n
tech
n
iq
u
e
to
s
e
g
m
e
n
t
t
h
e
t
h
i
n
len
s
b
o
u
n
d
ar
y
lo
ca
ted
o
n
to
p
o
f
th
e
s
c
ler
a
r
eg
io
n
.
A
s
to
d
ate,
E
r
d
o
g
an
an
d
R
o
s
s
[
29
]
ar
e
th
e
p
io
n
ee
r
to
i
m
p
le
m
en
t
th
i
s
ap
p
r
o
ac
h
.
T
h
e
y
p
r
o
p
o
s
ed
a
clu
s
ter
in
g
b
ased
ed
g
e
d
etec
tio
n
to
s
eg
m
en
t
th
e
o
u
ter
len
s
b
o
u
n
d
ar
y
w
h
ich
h
a
s
ac
h
ie
v
e
d
an
ac
cu
r
ac
y
o
f
7
0
%.
Fro
m
t
h
e
liter
at
u
r
e,
i
t
ca
n
b
e
s
u
m
m
ar
ized
t
h
at
t
h
e
h
ar
d
w
ar
e
ap
p
r
o
ac
h
r
elies
o
n
th
e
p
h
y
s
ical
c
h
ar
ac
ter
is
t
ic
o
f
th
e
e
y
e
.
Ma
ch
i
n
e
lear
n
in
g
ap
p
r
o
ac
h
in
s
tea
d
in
v
o
l
v
es
t
h
e
d
is
cr
i
m
in
at
io
n
b
et
w
ee
n
t
w
o
te
m
p
lates
[
30
]
w
h
ile
i
m
a
g
e
s
e
g
m
en
ta
tio
n
d
ea
ls
w
it
h
p
ix
els
m
an
ip
u
latio
n
.
3.
P
RO
P
O
SE
D
M
E
T
H
O
D
I
n
t
h
is
w
o
r
k
,
w
e
p
r
o
p
o
s
ed
a
f
u
s
io
n
o
f
i
m
a
g
e
s
e
g
m
en
tatio
n
a
n
d
m
ac
h
in
e
lear
n
i
n
g
ap
p
r
o
ac
h
t
o
class
i
f
y
t
w
o
clas
s
p
r
o
b
lem
s
o
f
v
alid
at
in
g
t
h
e
p
r
esen
ce
o
f
w
it
h
o
r
w
ith
o
u
t
s
o
f
t
le
n
s
.
T
h
ese
ap
p
r
o
a
ch
es
ar
e
ch
o
s
e
n
a
s
m
ac
h
in
e
lear
n
i
n
g
h
as t
h
e
ab
ilit
y
to
ex
tr
ac
t a
n
d
lear
n
f
r
o
m
tr
ain
in
g
d
ata
to
in
f
er
d
ec
is
io
n
f
o
r
a
n
e
w
d
ata
as
w
el
l
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
o
n
ta
ct
Len
s
C
la
s
s
i
fica
tio
n
b
y
U
s
in
g
S
eg
men
ted
L
en
s
B
o
u
n
d
a
r
y
F
ea
tu
r
es
(
N
u
r
A
r
i
ffin
Mo
h
d
Zi
n
)
1131
as
i
m
a
g
e
s
eg
m
e
n
tatio
n
ap
p
r
o
ac
h
th
at
r
o
b
u
s
t
to
len
s
t
y
p
e
an
d
s
en
s
o
r
s
.
C
ar
e
f
u
l
in
s
p
ec
tio
n
s
h
o
w
s
t
h
at
th
e
w
ea
r
i
n
g
o
f
s
o
f
t
len
s
w
ill
lea
v
e
an
o
b
s
cu
r
e
o
r
d
ar
k
er
ed
g
e
lin
e
o
n
to
p
o
f
th
e
s
c
ler
a
r
eg
io
n
as
s
h
o
w
n
i
n
Fig
u
r
e
1
.
I
t
h
as
b
ee
n
th
o
r
o
u
g
h
l
y
s
tu
d
ied
in
[
25
]
th
at
th
e
s
cler
a
r
eg
io
n
d
ef
in
e
t
h
e
b
est
s
a
m
p
le
to
p
r
o
v
e
th
e
p
r
esen
ce
o
f
s
o
f
t
len
s
.
T
h
er
ef
o
r
e,
o
u
r
f
o
cu
s
is
to
ex
tr
ac
t
i
n
f
o
r
m
at
io
n
o
n
t
h
is
r
eg
io
n
b
y
lo
ca
tin
g
th
e
p
i
x
els
o
f
len
s
b
o
u
n
d
ar
y
an
d
tr
a
n
s
f
o
r
m
i
t
in
to
f
ea
t
u
r
e
s
p
ac
e
b
y
m
ea
n
o
f
a
f
ea
t
u
r
e
d
escr
ip
to
r
.
T
h
ese
f
ea
t
u
r
es
ar
e
th
e
n
to
b
e
tr
ain
ed
b
y
a
s
tati
s
tical
lear
n
in
g
ap
p
r
o
ac
h
in
o
r
d
er
to
b
e
class
i
f
ied
.
T
h
e
n
ex
t
s
u
b
-
s
ec
tio
n
p
r
esen
t
s
t
h
e
m
ai
n
th
r
ee
s
ta
g
es
o
f
t
h
e
p
r
o
p
o
s
ed
m
et
h
o
d
w
h
ic
h
ar
e
s
e
g
m
e
n
tati
o
n
,
f
ea
t
u
r
es
e
x
tr
ac
tio
n
an
d
cla
s
s
i
f
icatio
n
.
T
h
e
s
tep
b
y
s
tep
p
r
o
ce
s
s
es o
f
t
h
is
m
et
h
o
d
is
d
ep
icted
in
Fig
u
r
e
2
.
Fig
u
r
e
1
.
E
x
a
m
p
les o
f
s
o
f
t le
n
s
b
o
u
n
d
ar
y
lo
ca
tio
n
a
s
p
o
in
t
ed
b
y
ar
r
o
w
3
.
1
Seg
m
e
nta
t
io
n
I
n
th
e
s
eg
m
e
n
tatio
n
s
ta
g
e,
th
e
m
ai
n
o
b
j
ec
tiv
e
is
to
lo
ca
te
th
e
s
o
f
t
le
n
s
b
o
u
n
d
ar
y
w
h
ich
is
d
ee
m
ed
to
r
esid
e
o
n
to
p
o
f
t
h
e
s
cler
a.
I
t
s
tar
ts
w
i
th
s
e
g
m
en
tat
io
n
o
f
t
h
e
lef
t
an
d
r
i
g
h
t
s
cler
a
r
e
g
io
n
.
W
e
u
s
ed
t
h
e
r
ad
iu
s
in
f
o
r
m
atio
n
p
r
o
v
id
ed
w
it
h
N
o
tr
e
Da
m
e
C
o
s
m
et
ic
C
o
n
tact
L
e
n
s
2
0
1
3
(
NDCC
L
1
3
)
[
31
]
to
s
eg
m
e
n
t
t
h
e
ir
i
s
b
o
u
n
d
ar
y
b
y
t
h
e
m
ea
n
o
f
C
i
r
cu
lar
Ho
u
g
h
T
r
an
s
f
o
r
m
.
T
h
en
,
w
e
ex
te
n
d
t
h
e
ir
is
r
ad
i
u
s
w
it
h
ad
d
itio
n
al
3
0
p
ix
els
to
in
c
lu
d
e
th
e
s
cler
a
r
eg
io
n
co
n
ce
n
tr
ical
l
y
.
T
h
e
r
ig
h
t
-
s
id
e
r
eg
io
n
is
s
e
g
m
en
ted
b
et
w
e
en
th
e
u
p
p
er
v
alu
e
o
f
1
5
0
to
lo
w
er
v
al
u
e
o
f
2
1
0
d
eg
r
ee
r
o
tatio
n
a
n
d
3
0
to
-
30
d
eg
r
ee
r
o
tatio
n
r
esp
ec
tiv
el
y
at
t
h
e
le
f
t
-
s
id
e
.
T
h
er
ef
o
r
e,
t
h
e
s
e
g
m
en
ta
tio
n
p
r
o
ce
s
s
w
ill
p
r
o
d
u
ce
t
w
o
s
e
g
m
e
n
ted
i
m
ag
e
s
o
f
le
f
t
an
d
r
ig
h
t
s
cler
a
.
T
h
ese
i
m
a
g
es
ar
e
th
e
n
to
b
e
n
o
r
m
a
lized
to
en
s
u
r
e
eq
u
al
d
i
m
en
s
io
n
f
o
r
ev
er
y
i
ter
atio
n
.
I
n
t
h
i
s
ca
s
e,
w
e
ap
p
lied
Dau
g
m
a
n
’
s
r
u
b
b
er
s
h
ee
t
n
o
r
m
aliza
tio
n
m
et
h
o
d
[
1
]
.
Th
e
n
o
r
m
al
iz
ed
i
m
a
g
es
w
ill
h
av
e
t
h
e
s
ize
o
f
3
0
2
4
0
p
ix
els.
A
ll
t
h
ese
s
t
ep
s
ar
e
illu
s
tr
ated
in
Fi
g
u
r
e
3
.
Nex
t,
a
n
y
p
r
esen
ce
o
f
e
y
elas
h
is
r
e
m
o
v
ed
u
s
in
g
i
n
p
ain
ti
n
g
alg
o
r
ith
m
[
32
]
.
T
h
is
i
s
d
o
n
e
b
y
r
ep
lacin
g
th
e
e
y
e
las
h
p
i
x
els
w
it
h
t
h
e
m
ea
n
o
f
s
u
r
r
o
u
n
d
i
n
g
p
ix
e
ls
.
W
e
o
b
s
er
v
ed
th
at
eit
h
er
o
n
e
o
f
t
h
e
le
f
t
o
r
r
ig
h
t
n
o
r
m
al
ized
i
m
ag
e
w
il
l
h
av
e
a
s
li
g
h
t
ly
b
r
ig
h
ter
o
r
d
ar
k
er
le
n
s
b
o
u
n
d
ar
y
.
T
h
is
i
s
d
u
e
to
t
h
e
i
n
co
n
s
is
ten
t
ill
u
m
in
at
io
n
ca
u
s
ed
b
y
p
o
o
r
f
las
h
li
g
h
tin
g
t
h
r
o
u
g
h
o
u
t th
e
en
r
o
l
l
m
en
t
p
r
o
ce
s
s
.
I
n
o
r
d
er
to
o
v
er
co
m
e
th
i
s
,
w
e
p
r
o
p
o
s
ed
a
m
e
th
o
d
ca
lled
s
u
m
m
ed
-
h
i
s
to
g
r
a
m
,
w
h
er
e
th
e
f
r
eq
u
e
n
cies
f
o
r
ea
ch
p
ix
el
’
s
in
ten
s
it
y
v
al
u
e
in
th
e
w
h
o
le
i
m
a
g
e
is
s
u
m
m
ed
.
W
h
en
e
v
er
th
e
s
u
m
v
alu
e
is
g
r
ea
ter
,
th
e
n
o
r
m
alize
d
i
m
a
g
e
is
co
n
s
id
er
ed
to
h
av
e
b
r
ig
h
ter
le
n
s
b
o
u
n
d
ar
y
co
m
p
ar
ed
to
t
h
e
o
t
h
er
.
T
h
e
s
u
m
m
ed
-
h
is
to
g
r
a
m
o
f
in
d
i
v
id
u
al
n
o
r
m
a
lized
i
m
ag
e
,
is
r
ep
r
esen
ted
as
f
o
llo
w
s
:
∫
∑
(
E
q
.
1
)
W
h
er
e
is
th
e
s
u
m
m
ed
-
h
is
to
g
r
a
m
o
f
n
o
r
m
alize
d
i
m
ag
e
an
d
is
th
e
p
ix
e
l’
s
in
ten
s
it
y
in
t
h
e
co
r
r
esp
o
n
d
in
g
n
o
r
m
ali
s
ed
p
o
lar
co
o
r
d
in
ates
.
T
h
e
v
alu
e
o
f
w
ill
b
e
th
e
p
r
e
li
m
i
n
ar
y
i
n
p
u
t
f
o
r
th
e
n
e
x
t
s
e
g
m
e
n
tat
io
n
pr
o
ce
s
s
.
I
n
t
h
is
p
r
o
ce
s
s
,
t
h
e
r
id
g
e
d
etec
tio
n
al
g
o
r
it
h
m
is
e
m
p
lo
y
ed
to
s
eg
m
e
n
t
t
h
e
le
n
s
b
o
u
n
d
ar
y
o
n
th
e
n
o
r
m
alize
d
i
m
a
g
e.
Am
o
n
g
o
th
er
s
s
eg
m
e
n
tatio
n
al
g
o
r
ith
m
,
r
i
d
g
e
d
etec
tio
n
h
as
th
e
ab
ilit
y
t
o
d
etec
t
ed
g
e
w
h
ile
s
h
ar
i
n
g
t
h
e
s
a
m
e
b
ac
k
g
r
o
u
n
d
co
lo
u
r
b
u
t
d
if
f
er
in
in
ten
s
itie
s
,
w
h
ic
h
r
ese
m
b
le
s
th
e
len
s
b
o
u
n
d
ar
y
[
33
]
.
T
h
e
o
u
tp
u
t
f
r
o
m
t
h
e
r
id
g
e
d
etec
tio
n
alg
o
r
ith
m
is
t
h
e
s
eg
m
e
n
ted
len
s
b
o
u
n
d
ar
y
i
n
a
f
o
r
m
o
f
b
in
ar
y
i
m
ag
e
w
h
er
e
len
s
b
o
u
n
d
ar
y
w
ill
b
e
r
eg
ar
d
e
d
as
w
h
ite
(
1
)
an
d
th
e
b
ac
k
g
r
o
u
n
d
as
b
lack
(
0
)
.
Fo
r
w
it
h
o
u
t
s
o
f
t
le
n
s
i
m
a
g
es,
a
b
lan
k
i
m
a
g
e
o
f
b
lack
co
lo
u
r
ed
w
ill b
e
g
e
n
er
ated
.
S
o
f
t
len
s
b
o
u
n
d
a
r
y
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
.
11
,
No
.
3
,
Sep
tem
b
er
201
8
:
1
1
2
9
–
1
1
3
5
1132
Fig
u
r
e
2
.
T
h
e
p
r
o
p
o
s
ed
tw
o
-
cl
ass
co
n
tact
le
n
s
cla
s
s
i
f
icat
io
n
p
r
o
ce
s
s
Fig
u
r
e
3
.
T
h
e
s
cler
a
s
eg
m
e
n
ta
tio
n
an
d
i
m
ag
e
n
o
r
m
aliza
t
io
n
p
r
o
ce
s
s
3
.
2
F
ea
t
ure
E
x
t
ra
ct
io
n
Un
d
er
f
ea
t
u
r
e
ex
tr
ac
tio
n
s
ta
g
e
,
t
w
o
i
n
p
u
t
i
m
a
g
e
s
ar
e
r
eq
u
ir
e
d
;
s
eg
m
en
ted
len
s
b
o
u
n
d
ar
y
i
m
ag
e
a
n
d
n
o
r
m
alize
d
i
m
a
g
e
o
f
s
c
ler
a
r
eg
io
n
.
T
h
e
s
eg
m
e
n
ted
le
n
s
b
o
u
n
d
ar
y
i
m
ag
e
is
ex
tr
ac
ted
u
s
i
n
g
a
w
i
n
d
o
w
ba
s
ed
f
ea
t
u
r
e
d
escr
ip
to
r
;
His
to
g
r
a
m
o
f
Gr
ad
ien
t
[
34
]
.
HOG
is
c
h
o
s
en
as
th
e
s
e
g
m
en
ted
le
n
s
b
o
u
n
d
ar
y
h
as
t
h
e
p
r
o
p
er
ties
o
f
th
e
ed
g
e
o
r
ien
ta
t
io
n
.
B
y
t
h
i
s
m
ea
n
,
f
o
r
e
v
er
y
n
n
p
ix
el
i
n
th
e
i
m
ag
e,
t
h
e
f
r
eq
u
en
c
y
h
is
to
g
r
a
m
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
o
n
ta
ct
Len
s
C
la
s
s
i
fica
tio
n
b
y
U
s
in
g
S
eg
men
ted
L
en
s
B
o
u
n
d
a
r
y
F
ea
tu
r
es
(
N
u
r
A
r
i
ffin
Mo
h
d
Zi
n
)
1133
o
f
ed
g
e
o
r
ien
tatio
n
is
co
m
p
u
ted
.
T
h
en
,
th
e
r
es
u
lted
ed
g
e
o
r
ien
tatio
n
is
q
u
an
t
ized
in
to
b
b
in
s
.
I
n
o
u
r
ex
p
er
i
m
e
n
t,
n
an
d
b
ar
e
s
et
t
o
4
an
d
9
r
esp
ec
tiv
el
y
.
Fro
m
o
u
r
o
b
s
er
v
atio
n
,
th
e
g
r
ad
ie
n
t
o
f
len
s
b
o
u
n
d
ar
y
is
p
r
o
n
e
to
ap
p
ea
r
in
h
o
r
izo
n
tal
s
ca
le,
w
h
ic
h
1
7
0
to
1
9
0
d
eg
r
ee
r
o
tatio
n
.
Me
an
w
h
i
le,
th
e
n
o
r
m
al
ized
i
m
a
g
e
o
f
th
e
s
cler
a
r
eg
io
n
i
s
d
escr
ip
ted
u
s
i
n
g
Scale
I
n
v
ar
ian
t
Feat
u
r
e
T
r
an
s
f
o
r
m
(
SIFT
)
[
35
]
.
T
h
e
u
t
ilizatio
n
o
f
SI
FT
is
to
ad
ap
t
th
e
p
r
o
p
er
ties
o
f
h
av
in
g
d
if
f
er
e
n
t
s
ca
le
s
,
r
o
tatio
n
an
d
in
h
o
m
o
g
en
eo
u
s
s
h
ap
e
o
f
len
s
b
o
u
n
d
ar
y
,
w
it
h
o
u
t
d
eg
r
ad
in
g
th
e
p
er
f
o
r
m
a
n
ce
o
f
d
etec
tio
n
.
T
h
er
e
ar
e
f
o
u
r
m
ai
n
s
tep
s
o
f
SIFT
;
s
ca
le
-
s
p
ac
e
r
ep
r
esen
tatio
n
,
k
e
y
p
o
i
n
ts
d
et
ec
tio
n
,
k
e
y
p
o
in
ts
o
r
ien
tatio
n
an
d
g
e
n
er
ate
k
e
y
p
o
in
t
d
escr
i
p
to
r
.
Du
r
in
g
s
ca
le
-
s
p
ac
e
r
ep
r
esen
tatio
n
,
th
e
Gau
s
s
ian
b
lu
r
r
i
n
g
is
ap
p
lied
to
th
e
w
h
o
le
i
m
a
g
e
w
it
h
t
h
e
n
u
m
b
er
o
f
o
ctav
e
is
4
an
d
lev
el
p
er
o
ctav
e
ar
e
s
et
to
3
.
T
h
en
,
Dif
f
er
en
ce
o
f
Ga
u
s
s
ian
(
Do
G)
o
f
th
e
b
lu
r
r
ed
im
ag
e
is
o
b
tain
ed
b
y
s
u
b
tr
ac
ti
n
g
s
u
b
s
eq
u
e
n
t
s
ca
les
in
ea
c
h
o
ctav
e,
p
r
o
d
u
ci
n
g
m
u
ltip
le
i
m
a
g
e
p
o
in
t
o
f
.
B
y
t
h
is
m
ea
n
,
a
k
e
y
p
o
in
t
is
d
etec
ted
w
h
en
i
ts
v
al
u
e
i
s
s
m
aller
(
lo
ca
l
m
in
i
m
u
m
)
o
r
lar
g
er
(
lo
ca
l
m
ax
i
m
u
m
)
t
h
an
th
e
s
u
r
r
o
u
n
d
in
g
p
o
in
t.
P
o
o
r
ly
lo
ca
lized
p
o
in
ts
ar
e
ex
c
lu
d
ed
d
u
r
in
g
k
e
y
p
o
in
t
s
d
etec
tio
n
.
T
h
e
n
e
x
t
p
r
o
ce
s
s
is
to
ass
i
g
n
a
n
o
r
ien
tatio
n
to
ea
ch
k
e
y
p
o
in
t
b
y
ca
lcu
la
tin
g
it
s
g
r
a
d
ien
t
d
ir
ec
tio
n
s
an
d
m
ag
n
it
u
d
e
s
in
a
r
eg
io
n
o
f
1
6
16
.
Fin
all
y
,
t
h
ese
r
e
g
io
n
s
ar
e
b
r
o
k
en
in
to
s
ix
teen
4
4
w
i
n
d
o
w
a
n
d
ac
cu
m
u
lates
th
e
m
i
n
to
8
b
in
s
h
is
to
g
r
a
m
w
it
h
a
w
ei
g
h
ted
v
a
lu
e
o
f
g
r
ad
ien
t
m
a
g
n
itu
d
e.
T
h
er
ef
o
r
e,
t
h
er
e
is
4
4
8
=
1
2
8
f
ea
tu
r
e
v
ec
to
r
s
f
o
r
ea
ch
k
e
y
p
o
in
t
.
I
n
th
is
w
o
r
k
,
o
u
r
in
ter
est
is
to
u
s
e
t
h
e
d
escr
ip
tio
n
o
f
ed
g
e
o
r
ien
tatio
n
o
f
H
OG
an
d
th
e
i
n
v
ar
ia
n
ce
ed
g
e
p
r
o
p
er
ties
th
r
o
u
g
h
SIFT
to
f
o
r
m
a
d
is
ti
n
g
u
i
s
h
ab
le
f
ea
t
u
r
e
o
f
w
it
h
o
u
t
a
n
d
w
it
h
s
o
f
t
l
en
s
.
T
h
is
is
d
o
n
e
b
y
co
n
ca
ten
ati
n
g
b
o
th
lef
t a
n
d
r
ig
h
t f
ea
tu
r
es o
f
HOG
a
n
d
SIFT
f
o
r
th
e
r
esp
ec
tiv
e
ir
is
i
m
a
g
e
in
t
o
a
f
ea
tu
r
e
v
ec
to
r
.
3
.
3
Cla
s
s
if
ica
t
io
n
T
h
e
co
n
ca
ten
ated
f
ea
t
u
r
es
ar
e
tr
ain
ed
u
s
in
g
a
n
o
n
-
lin
ea
r
S
VM
w
it
h
r
ad
i
u
s
b
a
s
is
f
u
n
ctio
n
k
er
n
el
.
1
0
f
o
ld
cr
o
s
s
v
a
lid
atio
n
w
er
e
u
t
ilized
t
o
o
b
tain
th
e
f
i
ttes
t
p
ar
a
m
eter
s
o
f
an
d
g
a
m
m
a
in
o
r
d
er
to
r
e
d
u
ce
o
v
er
f
itti
n
g
.
I
n
o
r
d
er
to
class
if
y
b
et
w
ee
n
w
it
h
o
u
t
o
r
w
i
th
s
o
f
t
len
s
,
th
e
i
m
ag
e
s
o
f
w
it
h
o
u
t
l
en
s
ar
e
al
s
o
ap
p
lied
to
th
e
w
h
o
le
p
r
o
ce
s
s
f
r
o
m
s
eg
m
e
n
tat
io
n
to
f
ea
tu
r
e
e
x
tr
ac
tio
n
.
I
m
a
g
es
w
i
th
p
r
io
r
k
n
o
w
led
g
e
o
f
s
o
f
t
le
n
s
p
r
esen
ce
w
ill b
e
lab
elled
as p
o
s
iti
v
e
s
a
m
p
le
s
an
d
w
it
h
o
u
t le
n
s
i
m
a
g
es a
s
n
e
g
ati
v
e
s
a
m
p
les.
4.
RE
SU
L
T
S AN
D
AN
AL
Y
SI
S
I
n
th
is
w
o
r
k
,
th
e
ir
is
i
m
a
g
e
s
ar
e
r
etr
iev
ed
f
r
o
m
No
tr
e
Da
m
e
C
o
s
m
etic
C
o
n
tact
L
en
s
2
0
1
3
(
NDCC
L
1
3
)
[
31
]
th
at
co
n
tai
n
s
g
r
a
y
s
ca
le
ir
is
i
m
a
g
e
s
o
f
w
i
t
h
o
u
t
len
s
,
w
it
h
s
o
f
t
le
n
s
an
d
w
it
h
co
s
m
et
ic
le
n
s
.
A
ll
i
m
ag
e
s
ar
e
ca
p
tu
r
ed
u
s
i
n
g
eith
er
L
G4
0
0
0
o
r
A
D1
0
0
NI
R
ca
m
er
a.
F
o
r
L
G4
0
0
0
,
all
im
ag
e
s
ar
e
s
p
lit
i
n
to
3
0
0
0
f
o
r
tr
ain
in
g
an
d
1
2
0
0
f
o
r
test
in
g
w
h
ile
f
o
r
A
D1
0
0
,
6
0
0
im
a
g
es
ar
e
f
o
r
tr
ain
in
g
a
n
d
3
0
0
f
o
r
test
i
n
g
.
T
ab
le
1
s
h
o
w
s
t
h
e
i
m
a
g
es d
is
t
r
ib
u
tio
n
f
o
r
ND
C
C
L
1
3
d
atab
ase.
T
ab
le
1
.
NDCC
L
1
3
i
m
a
g
e
s
cl
ass
d
is
tr
ib
u
tio
n
C
a
me
r
a
C
l
a
ss
L
a
b
e
l
T
r
a
i
n
i
n
g
T
e
st
i
n
g
L
G
4
0
0
0
No
N
1
0
0
0
4
0
0
S
o
f
t
S
1
0
0
0
4
0
0
C
o
sme
t
i
c
T
1
0
0
0
4
0
0
A
D
1
0
0
No
N
2
0
0
1
0
0
S
o
f
t
S
2
0
0
1
0
0
C
o
sme
t
i
c
T
2
0
0
1
0
0
A
ll
e
x
p
er
i
m
e
n
ts
ar
e
ex
ec
u
ted
u
s
i
n
g
Ma
tlab
R
2
0
1
7
b
o
n
a
m
ac
h
in
e
w
i
th
2
.
3
GHz
an
d
6
GB
m
e
m
o
r
y
.
W
e
ca
lcu
lated
co
n
f
u
s
io
n
m
a
tr
i
x
f
o
r
ea
ch
cla
s
s
o
f
d
if
f
er
e
n
t c
a
m
er
as.
Fo
r
u
n
i
f
o
r
m
it
y
w
it
h
[
7
]
,
w
e
o
n
l
y
r
ep
o
r
ted
co
r
r
ec
t
class
if
icat
io
n
r
ate
(
C
C
R
)
w
ith
i
ts
a
v
er
ag
e.
Ho
w
ev
er
,
th
e
ev
al
u
atio
n
o
f
co
s
m
etic
le
n
s
i
n
n
o
t
w
i
th
i
n
o
u
r
s
co
p
e
as
t
h
i
s
w
o
r
k
o
n
l
y
f
o
cu
s
in
g
o
n
t
w
o
cla
s
s
clas
s
i
f
icatio
n
o
f
w
it
h
o
r
w
it
h
o
u
t
s
o
f
t
len
s
.
W
e
an
n
o
tated
t
h
e
class
o
f
w
i
th
o
u
t
len
s
as
N
a
n
d
s
o
f
t
le
n
s
a
s
S.
N
-
N
r
ef
er
s
to
th
e
p
r
o
b
ab
ilit
y
o
f
w
it
h
o
u
t
len
s
s
a
m
p
les
ar
e
class
i
f
ied
b
elo
n
g
s
to
w
it
h
o
u
t
len
s
w
h
i
le
S
-
S
r
e
f
er
s
to
t
h
e
p
r
o
b
a
b
ilit
y
o
f
s
o
f
t
len
s
s
a
m
p
les
ar
e
class
i
f
ied
b
elo
n
g
s
to
s
o
f
t
le
n
s
.
C
o
m
p
ar
i
s
o
n
s
ar
e
m
ad
e
w
it
h
e
x
is
t
in
g
m
et
h
o
d
s
p
r
o
p
o
s
ed
in
[
7
]
b
y
u
s
in
g
cla
s
s
ica
l
L
B
P
,
[
36
]
b
y
u
s
i
n
g
a
m
o
d
i
f
ied
v
e
r
s
io
n
o
f
L
B
P
,
[
28
]
b
y
u
s
i
n
g
C
o
n
v
o
lu
t
io
n
al
Ne
u
r
al
Ne
t
wo
r
k
,
[
27
]
b
y
u
s
i
n
g
B
in
ar
ized
Stati
s
tical
I
m
a
g
e
F
ea
tu
r
es
a
n
d
[
25
]
b
y
u
s
in
g
S
ca
le
I
n
v
ar
ian
t
Descr
ip
to
r
.
D
u
r
in
g
s
eg
m
e
n
tatio
n
s
tag
e,
ea
ch
ir
is
i
m
ag
e
w
il
l
p
r
o
d
u
ce
t
w
o
s
eg
m
e
n
ted
len
s
b
o
u
n
d
ar
y
i
m
a
g
es
an
d
t
w
o
n
o
r
m
al
ized
im
a
g
es
w
h
ic
h
d
er
iv
ed
f
r
o
m
t
h
e
le
f
t
an
d
r
i
g
h
t
s
cler
a
.
T
h
ese
i
m
ag
e
s
ar
e
u
s
ed
f
o
r
tr
ain
i
n
g
a
n
d
test
i
n
g
th
r
o
u
g
h
o
u
t
t
h
e
class
i
f
icatio
n
s
tag
e.
T
h
ese
s
a
m
p
les
o
f
i
m
ag
e
s
ar
e
s
h
o
w
n
i
n
T
ab
le
2
.
T
h
er
ef
o
r
e,
f
o
r
L
G4
0
0
0
,
th
er
e
ar
e
2
0
0
0
tr
ain
i
n
g
i
m
ag
e
s
f
o
r
w
it
h
o
u
t
l
en
s
a
n
d
a
n
o
th
er
2
0
0
0
f
o
r
s
o
f
t
len
s
.
Me
a
n
w
h
ile,
f
o
r
A
D1
0
0
,
b
o
th
w
it
h
o
u
t
le
n
s
an
d
s
o
f
t
le
n
s
co
n
s
tit
u
te
4
0
0
tr
ain
i
n
g
i
m
a
g
es
ea
c
h
.
T
h
e
n
u
m
b
er
s
o
f
test
in
g
i
m
a
g
es
ar
e
d
o
u
b
led
u
p
,
th
e
s
a
m
e
m
an
n
er
as tr
ai
n
i
n
g
i
m
a
g
es.
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
.
11
,
No
.
3
,
Sep
tem
b
er
201
8
:
1
1
2
9
–
1
1
3
5
1134
T
ab
le
2
.
NDCC
L
1
3
i
m
a
g
e
s
cl
ass
d
is
tr
ib
u
tio
n
I
mag
e
S
c
l
e
r
a
N
o
r
mal
i
z
e
d
i
mag
e
S
e
g
me
n
t
e
d
l
e
n
s b
o
u
n
d
a
r
y
0
4
2
6
1
d
1
0
1
6
L
e
f
t
R
i
g
h
t
0
6
0
0
8
d
5
8
L
e
f
t
R
i
g
h
t
T
ab
le
3
r
e
p
o
r
ts
th
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
w
it
h
th
e
a
f
o
r
e
m
e
n
tio
n
ed
s
tate
-
of
-
t
h
e
-
ar
t
m
et
h
o
d
s
.
T
h
e
b
est
r
es
u
lt
s
ar
e
w
r
i
tten
i
n
b
o
ld
.
I
n
s
u
m
m
ar
y
,
t
h
e
p
r
o
p
o
s
ed
m
eth
o
d
h
as
y
ield
ed
th
e
h
ig
h
es
t
av
er
ag
e
f
o
r
b
o
th
ca
m
er
a
s
.
H
o
w
e
v
er
,
th
e
L
G4
0
0
0
’s
C
C
R
f
o
r
N
-
N
cla
s
s
i
f
icatio
n
h
as
u
n
ab
le
to
s
u
r
p
as
s
t
h
e
m
et
h
o
d
f
r
o
m
[
25
]
.
I
n
t
h
e
m
ea
n
ti
m
e,
t
h
e
C
C
R
f
o
r
S
-
S
clas
s
i
f
icatio
n
r
es
u
lted
o
v
er
1
0
p
o
in
ts
f
r
o
m
t
h
e
n
ea
r
est
f
i
g
u
r
e
.
I
t
is
o
b
s
er
v
ed
th
at
th
e
r
esu
lts
f
o
r
A
D1
0
0
ar
e
v
er
y
m
ar
g
i
n
al
to
ac
h
ie
v
e
s
ig
n
i
f
ica
n
t
p
er
f
o
r
m
an
ce
.
W
e
b
eliev
e
t
h
e
le
s
s
n
u
m
b
er
o
f
s
u
b
j
ec
ts
is
n
o
t
s
u
f
f
icie
n
t
to
s
tati
s
ticall
y
g
en
er
alize
th
e
v
ar
iab
il
it
y
b
et
w
ee
n
cla
s
s
e
s
.
So
m
e
o
f
f
u
tu
r
e
w
o
r
k
s
[
3
7
,
3
8
]
m
a
y
b
e
ad
d
ed
f
o
r
b
etter
p
e
r
f
o
r
m
an
ce
i
n
t
h
is
r
esear
c
h
.
T
ab
le
3
.
T
h
e
co
r
r
ec
t c
lass
if
ica
tio
n
r
ate
an
d
av
er
ag
e
o
f
th
e
p
r
o
p
o
s
ed
an
d
s
tate
o
f
th
e
ar
t
m
et
h
o
d
s
C
a
me
r
a
s
C
l
a
ss
i
f
i
c
a
t
i
o
n
L
B
P
[
7
]
L
B
P
+
P
H
O
G
[
7
]
m
L
B
P
[
7
]
[
36
]
C
NN
[
28
]
B
S
I
F
[
27
]
S
I
D
[
25
]
P
r
o
p
o
se
d
M
e
t
h
o
d
L
G
4
0
0
0
N
-
N
7
0
.
0
0
8
1
.
2
5
8
5
.
5
0
8
4
.
5
0
7
6
.
5
0
9
5
.
7
5
8
5
.
6
3
S
-
S
6
0
.
1
5
6
5
.
4
1
4
5
.
2
5
7
3
.
7
5
8
4
.
5
0
8
4
.
0
0
9
5
.
7
5
A
v
e
r
a
g
e
6
5
.
0
8
7
3
.
3
3
6
5
.
3
8
7
9
.
1
3
8
0
.
5
0
8
9
.
8
8
9
0
.
6
9
A
D
1
0
0
N
-
N
4
2
.
0
0
4
2
.
0
0
8
1
.
0
0
7
3
.
0
0
7
9
.
5
0
7
9
.
0
0
8
0
.
2
0
S
-
S
5
4
.
0
0
6
0
.
0
0
5
2
.
0
0
6
5
.
0
0
6
2
.
0
0
7
8
.
0
0
7
9
.
1
2
A
v
e
r
a
g
e
48
51
6
6
.
5
69
7
0
.
7
5
7
8
.
5
7
9
.
6
6
5.
CO
NCLU
SI
O
N
T
h
is
w
o
r
k
p
r
o
p
o
s
ed
t
w
o
clas
s
co
n
tact
len
s
clas
s
i
f
icatio
n
o
f
w
it
h
o
r
w
it
h
o
u
t
s
o
f
t
le
n
s
.
I
n
o
r
d
er
t
o
d
is
tin
g
u
is
h
b
et
w
ee
n
t
h
ese
cla
s
s
es,
w
e
f
o
cu
s
ed
o
n
e
x
tr
ac
ti
n
g
lo
ca
l
in
f
o
r
m
atio
n
av
a
ilab
le
o
n
th
e
s
cler
a
r
eg
io
n
.
I
t
is
o
b
s
er
v
ed
th
at
t
h
e
w
ea
r
i
n
g
o
f
s
o
f
t
le
n
s
m
a
y
leav
e
t
h
in
li
n
es
w
h
ic
h
ap
p
ea
r
to
b
e
th
e
le
n
s
b
o
u
n
d
ar
y
.
T
h
er
ef
o
r
e,
w
e
p
r
o
p
o
s
ed
a
s
eg
m
e
n
tat
io
n
m
ec
h
an
i
s
m
to
s
e
g
m
e
n
t
th
i
s
b
o
u
n
d
ar
y
an
d
u
ti
lize
His
to
g
r
a
m
o
f
Gr
ad
ien
t
to
e
x
tr
ac
t
t
h
e
g
r
ad
ie
n
t
o
r
ien
tatio
n
o
f
th
e
b
o
u
n
d
ar
y
.
W
e
also
d
escr
ip
t
t
h
e
o
r
ig
in
al
s
cler
a
r
eg
io
n
w
it
h
Scale
I
n
v
ar
ian
t
Featu
r
e
T
r
an
s
f
o
r
m
in
o
r
d
er
to
tack
le
t
h
e
in
h
o
m
o
g
e
n
eo
u
s
s
h
ap
e
an
d
r
o
tatio
n
o
f
t
h
e
le
n
s
b
o
u
n
d
ar
y
.
T
h
ese
f
ea
t
u
r
es
ar
e
co
n
ca
ten
ated
b
ef
o
r
e
tr
ain
ed
an
d
class
i
f
ied
u
s
i
n
g
a
n
o
n
-
li
n
ea
r
SVM.
R
e
s
u
lts
s
h
o
w
ed
th
at
t
h
e
p
r
o
p
o
s
ed
m
e
th
o
d
ac
h
iev
ed
th
e
h
ig
h
es
t
av
e
r
ag
e
ac
cu
r
ac
y
co
m
p
ar
ed
to
o
t
h
er
s
tate
-
of
-
th
e
-
ar
t
m
et
h
o
d
s
.
ACK
NO
WL
E
D
G
E
M
E
NT
T
h
is
w
o
r
k
h
as
b
ee
n
f
u
n
d
ed
by
U
n
iv
er
s
iti
T
ek
n
o
l
o
g
i
Ma
la
y
s
ia
a
n
d
Mi
n
is
tr
y
o
f
H
ig
h
er
E
d
u
ca
t
io
n
Ma
la
y
s
ia
u
n
d
er
FR
GS V
o
t N
o
:
4
F9
7
3
.
RE
F
E
R
E
NC
E
S
[1
]
Da
u
g
m
a
n
J
G
.
Bi
o
me
tric P
e
rs
o
n
a
l
Id
e
n
ti
f
ica
ti
o
n
S
y
ste
m b
a
se
d
o
n
Ir
is
A
n
a
lys
is
.
G
o
o
g
le P
a
ten
ts.
1
9
9
4
.
[2
]
Dh
a
v
a
le S
V
.
Ro
b
u
st I
ris
Re
c
o
g
n
i
ti
o
n
b
a
se
d
o
n
S
tatisti
c
a
l
P
ro
p
e
rti
e
s o
f
W
a
lsh
H
a
d
a
m
a
rd
T
ra
n
s
f
o
r
m
D
o
m
a
in
.
IJ
CS
I
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
Iss
u
e
s
.
2
0
1
2
;
9
(
2
)
.
[3
]
Ho
ss
e
in
i
M
S
,
A
ra
a
b
i
BN
,
a
n
d
S
o
lt
a
n
ia
n
-
Zad
e
h
H.
P
ig
m
e
n
t
M
e
lan
in
:
P
a
tt
e
r
n
f
o
r
Iris
R
e
c
o
g
n
it
io
n
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
In
str
u
me
n
ta
t
io
n
a
n
d
M
e
a
su
re
me
n
t
.
2
0
1
0
;
5
9
(
4
):
7
9
2
-
8
0
4
.
[4
]
S
im
HM,
Hish
a
m
m
u
d
d
in
n
A
,
Ha
ss
a
n
R,
Oth
m
a
n
RM
.
M
u
l
ti
m
o
d
a
l
B
io
m
e
tri
c
s:
W
e
ig
h
ted
sc
o
re
lev
e
l
f
u
sio
n
b
a
se
d
on
n
o
n
-
id
e
a
l
ir
is
a
n
d
f
a
c
e
i
m
a
g
e
s.
Exp
e
rt S
y
ste
ms
wit
h
Ap
p
li
c
a
ti
o
n
s
.
2
0
1
4
;
4
1
(1
1
):
5
3
9
0
-
5
4
0
4
.
[5
]
W
a
n
g
Y,
Ha
n
J.
I
ris
Rec
o
g
n
it
io
n
u
sin
g
S
u
p
p
o
rt
Vec
to
r
M
a
c
h
in
e
s
.
In
tern
a
ti
o
n
a
l
S
y
m
p
o
siu
m
o
n
Ne
u
ra
l
Ne
tw
o
rk
s.
2
0
0
4
;
3
1
7
3
;
6
2
2
-
6
2
8
.
[6
]
Ba
k
e
r
S
E,
H
e
n
tz
A
,
Bo
wy
e
r
K
W
,
F
l
y
n
n
P
J.
De
g
ra
d
a
ti
o
n
o
f
Iris
Re
c
o
g
n
it
io
n
P
e
rf
o
rm
a
n
c
e
d
u
e
to
No
n
-
Co
s
m
e
ti
c
P
re
sc
rip
t
io
n
Co
n
tac
t
L
e
n
se
s.
Co
mp
u
ter
Vi
si
o
n
a
n
d
Im
a
g
e
Un
d
e
rs
ta
n
d
i
n
g
.
2
0
1
0
;
1
1
4
(9
):
1
0
3
0
-
1
0
4
4
.
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
o
n
ta
ct
Len
s
C
la
s
s
i
fica
tio
n
b
y
U
s
in
g
S
eg
men
ted
L
en
s
B
o
u
n
d
a
r
y
F
ea
tu
r
es
(
N
u
r
A
r
i
ffin
Mo
h
d
Zi
n
)
1135
[7
]
Ya
d
a
v
D
,
Ko
h
li
N,
Do
y
le
JS,
S
in
g
h
R,
V
a
tsa
M
,
Bo
wy
e
r
K
W
.
U
n
ra
v
e
li
n
g
th
e
Eff
e
c
t
o
f
T
e
x
tu
re
d
Co
n
tac
t
L
e
n
se
s
o
n
Iris
Re
c
o
g
n
it
io
n
.
I
EE
E
T
r
a
n
s
a
c
ti
o
n
s
o
n
I
n
f
o
rm
a
ti
o
n
F
o
re
n
sic
s a
n
d
S
e
c
u
rity
.
2
0
1
4
;
9
(5
)
:
8
5
1
-
8
6
2
.
[8
]
Ka
lso
o
m
S
,
Zi
a
u
d
d
in
S.
Iris R
e
c
o
g
n
it
i
o
n
:
Existin
g
M
e
th
o
d
s a
n
d
O
p
e
n
Is
su
e
s
.
T
h
e
F
o
u
rt
h
In
tern
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
s
o
n
P
e
rv
a
siv
e
P
a
tt
e
rn
s a
n
d
A
p
p
l
ica
ti
o
n
s.
2
0
1
2
:
23
-
2
8
.
[9
]
Do
y
le
JS
,
F
ly
n
n
P
J,
B
o
wy
e
r
KW
.
Au
to
ma
ted
Cl
a
ss
if
ica
t
io
n
o
f
C
o
n
t
a
c
t
L
e
n
s
T
y
p
e
in
Iris
Ima
g
e
s
,
2
0
1
3
In
tern
a
ti
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Bi
o
m
e
tri
c
s (
ICB).
2
0
1
3
.
[1
0
]
L
o
v
ish
,
Nig
a
m
A
,
Ku
m
a
r
B,
G
u
p
ta
P
.
R
o
b
u
st
C
o
n
t
a
c
t
L
e
n
s
De
tec
ti
o
n
u
si
n
g
L
o
c
a
l
P
h
a
se
Qu
a
n
ti
za
ti
o
n
a
n
d
Bi
n
a
ry
Ga
b
o
r
Pa
tt
e
rn
.
1
6
t
h
In
t
e
r
n
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
Co
m
p
u
ter A
n
a
l
y
sis o
f
I
m
a
g
e
s an
d
P
a
tt
e
rn
s.
2
0
1
5
:
7
0
2
-
7
1
4
.
[1
1
]
Ku
m
a
r
B,
Nig
a
m
A
,
G
u
p
ta
P
.
F
u
ll
y
Au
t
o
ma
ted
S
o
ft
Co
n
t
a
c
t
L
e
n
s
De
tec
ti
o
n
fro
m
NIR
Iris
Ima
g
e
s
.
In
tern
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
P
a
tt
e
rn
Re
c
o
g
n
it
io
n
A
p
p
li
c
a
ti
o
n
s an
d
M
e
th
o
d
s.
2
0
1
6
:
5
8
9
-
5
9
6
.
[1
2
]
Da
u
g
m
a
n
J.
D
e
m
o
d
u
latio
n
b
y
Co
m
p
lex
-
V
a
lu
e
d
W
a
v
e
lets
f
o
r
S
to
c
h
a
stic P
a
tt
e
rn
Re
c
o
g
n
it
i
o
n
.
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
W
a
v
e
lets,
M
u
lt
ire
so
l
u
ti
o
n
a
n
d
In
fo
rm
a
t
io
n
Pro
c
e
ss
in
g
.
2
0
0
3
:
1
(
0
1
);
1
-
1
7
.
[1
3
]
L
e
e
EC,
P
a
rk
KR,
Kim
J.
Fa
k
e
I
ris
De
tec
ti
o
n
b
y
Us
in
g
Pu
rk
i
n
je
Ima
g
e
.
I
n
tern
a
ti
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
B
io
m
e
tri
c
s.
2
0
0
6
:
3
9
7
-
4
0
3
.
[1
4
]
He
X
,
A
n
S
,
S
h
i
P
.
S
t
a
ti
stica
l
T
e
x
tu
re
An
a
lys
is
-
b
a
se
d
Ap
p
ro
a
c
h
f
o
r
F
a
k
e
Iris
De
tec
ti
o
n
u
sin
g
S
u
p
p
o
rt
Vec
to
r
M
a
c
h
in
e
s
.
In
tern
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
Bi
o
m
e
tri
c
s.
2
0
0
7
:
5
4
0
-
5
4
6
.
[1
5
]
W
e
i
Z,
Qiu
X
,
S
u
n
Z,
T
a
n
T
.
Co
u
n
ter
feit
Iris
De
tec
ti
o
n
b
a
se
d
o
n
T
e
x
tu
re
A
n
a
lys
is
.
1
9
th
In
ter
n
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
on
P
a
t
tern
Re
c
o
g
n
it
i
o
n
.
2
0
0
8
.
[1
6
]
Zh
a
n
g
H
,
S
u
n
Z
,
T
a
n
T
.
Co
n
ta
c
t
L
e
n
s
D
e
te
c
ti
o
n
b
a
se
d
o
n
W
e
ig
h
ted
L
BP
.
2
0
t
h
I
n
tern
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
.
P
a
tt
e
r
n
Re
c
o
g
n
it
i
o
n
(IC
P
R).
2
0
1
0
.
[1
7
]
W
il
li
a
m
s
G
O.
Iris R
e
c
o
g
n
it
i
o
n
T
e
c
h
n
o
l
o
g
y
.
3
0
th
A
n
n
u
a
l
In
tern
a
ti
o
n
a
l
Ca
rn
a
h
a
n
C
o
n
f
e
re
n
c
e
.
1
9
9
6
.
[1
8
]
Ne
g
in
M,
Ch
m
iel
e
w
sk
i
TA
,
S
a
lg
a
n
ico
f
f
M
,
v
o
n
S
e
e
len
UM,
V
e
n
e
tain
e
r
P
L
,
Zh
a
n
g
GG
.
A
n
Iris
Bi
o
m
e
tri
c
S
y
ste
m
f
o
r
P
u
b
li
c
a
n
d
P
e
rs
o
n
a
l
U
se
.
C
o
mp
u
ter
.
2
0
0
0
:
3
3
(2
);
70
-
7
5
.
[1
9
]
A
li
JM,
Ha
ss
a
n
ien
A
E.
A
n
Iris
Re
c
o
g
n
it
io
n
S
y
ste
m
to
En
h
a
n
c
e
e
-
S
e
c
u
rit
y
En
v
iro
n
m
e
n
t
b
a
se
d
o
n
W
a
v
e
let
T
h
e
o
r
y
.
Ad
v
a
n
c
e
d
M
o
d
e
li
n
g
a
n
d
Op
ti
miz
a
ti
o
n
.
2
0
0
3
:
5
(
2
)
;
9
3
-
1
0
4
.
[2
0
]
Ba
k
e
r
S
E,
He
n
tz A
,
Bo
wy
e
r
K
W
,
F
ly
n
n
P
J.
Co
n
ta
c
t
L
e
n
se
s: Ha
n
d
l
e
wit
h
Ca
re
f
o
r Iris R
e
c
o
g
n
it
i
o
n
.
2
0
0
9
IE
EE
3
r
d
In
tern
a
ti
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Bi
o
m
e
tri
c
s: T
h
e
o
r
y
,
A
p
p
li
c
a
ti
o
n
s a
n
d
S
y
ste
m
s.
2
0
0
9
:
1
9
0
-
1
9
7
.
[2
1
]
K
y
w
e
W
,
Yo
sh
id
a
M
,
M
u
ra
k
a
m
i
K.
Co
n
ta
c
t
L
e
n
s
Extr
a
c
ti
o
n
b
y
Us
in
g
T
h
e
rm
o
-
Vi
si
o
n
.
1
8
th
In
tern
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
P
a
tt
e
rn
Re
c
o
g
n
it
io
n
.
2
0
0
6
.
[2
2
]
L
e
e
EC
,
P
a
rk
KR,
Kim
J.
F
a
k
e
Iri
s De
tec
ti
o
n
b
y
u
sin
g
P
u
rk
in
je Im
a
g
e
.
Ad
v
a
n
c
e
s in
B
io
me
trics
.
2
0
0
5
:
3
9
7
-
4
0
3
.
[2
3
]
Hu
g
h
e
s K,
Bo
wy
e
r
KW
.
De
tec
ti
o
n
o
f
Co
n
ta
c
t
L
e
n
s
-
b
a
se
d
Iris B
io
me
tric S
p
o
o
fs
u
si
n
g
S
ter
e
o
I
ma
g
i
n
g
.
4
6
t
h
Ha
wa
ii
In
tern
a
ti
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
S
y
ste
m
S
c
ien
c
e
s (HICS
S
).
2
0
1
3
.
[2
4
]
Ko
h
li
N,
Da
k
sh
a
Y
,
V
a
tsa
M
,
S
in
g
h
R.
Rev
isit
in
g
Iris
Rec
o
g
n
it
i
o
n
wit
h
Co
lo
r
Co
sm
e
ti
c
Co
n
t
a
c
t
L
e
n
se
s
.
In
tern
a
ti
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Bi
o
m
e
tri
c
s.
2
0
1
3
:
1
-
5
.
[2
5
]
G
ra
g
n
a
n
iello
D,
P
o
g
g
i
G
,
S
a
n
so
n
e
C,
V
e
rd
o
li
v
a
L
.
Co
n
ta
c
t
L
e
n
s
De
tec
ti
o
n
a
n
d
Cla
ss
if
ic
a
ti
o
n
in
Iris
Ima
g
e
s
th
ro
u
g
h
S
c
a
le
In
v
a
ria
n
t
D
e
sc
rip
to
r
.
1
0
th
I
n
tern
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
S
ig
n
a
l
-
Im
a
g
e
T
e
c
h
n
o
lo
g
y
a
n
d
In
ter
n
e
t
-
Ba
se
d
S
y
st
e
m
s.
2
0
1
4
:
5
6
0
-
5
6
5
.
[2
6
]
G
ra
g
n
a
n
iello
D
,
P
o
g
g
i
G
,
S
a
n
so
n
e
C,
V
e
rd
o
l
iv
a
L
.
Us
in
g
Iris
a
n
d
S
c
lera
f
o
r
De
tec
ti
o
n
a
n
d
Cl
a
ss
if
i
c
a
ti
o
n
o
f
Co
n
tac
t
L
e
n
se
s.
Pa
tt
e
rn
Rec
o
g
n
it
io
n
L
e
tt
e
rs
.
2
0
1
6
:
8
2
(2
)
;
2
5
1
-
2
5
7
.
[2
7
]
Ra
g
h
a
v
e
n
d
ra
R,
Ra
ja
KB
,
Bu
sc
h
C.
En
se
mb
le
o
f
S
ta
ti
stica
ll
y
In
d
e
p
e
n
d
e
n
t
Fi
lt
e
rs
f
o
r
Ro
b
u
st
Co
n
t
a
c
t
L
e
ns
De
tec
ti
o
n
in
Iris I
ma
g
e
s
.
In
d
ian
Co
n
f
e
re
n
c
e
o
n
Co
m
p
u
ter
V
isi
o
n
G
ra
p
h
ics
a
n
d
Im
a
g
e
P
ro
c
e
ss
in
g
.
2
0
1
4
.
[2
8
]
S
il
v
a
P
,
L
u
z
E,
Ba
e
t
a
R,
P
e
d
rin
i
H,
F
a
lca
o
A
X
,
M
e
n
o
tt
i
D.
An
Ap
p
ro
a
c
h
to
Iris
Co
n
ta
c
t
L
e
n
s
De
tec
ti
o
n
b
a
se
d
o
n
De
e
p
Ima
g
e
Rep
re
se
n
ta
ti
o
n
s
.
2
8
t
h
S
IBG
RA
P
I
Co
n
f
e
re
n
c
e
o
n
G
ra
p
h
ics
,
P
a
tt
e
r
n
s an
d
Im
a
g
e
s.
2
0
1
5
:
1
5
7
-
1
6
4
.
[2
9
]
Erd
o
g
a
n
G
,
Ro
ss
A.
A
u
to
m
a
ti
c
De
tec
ti
o
n
o
f
No
n
-
Co
sm
e
ti
c
S
o
f
t
Co
n
t
a
c
t
L
e
n
se
s
i
n
Oc
u
l
a
r
Im
a
g
e
s
.
Bio
m
e
tri
c
a
n
d
S
u
rv
e
il
lan
c
e
T
e
c
h
n
o
lo
g
y
f
o
r
Hu
m
a
n
a
n
d
A
c
ti
v
it
y
Id
e
n
ti
f
ica
ti
o
n
X,
2
0
1
3
:
8
7
1
2
.
[3
0
]
He
Z,
S
u
n
Z
,
T
a
n
T
,
W
e
i
Z.
Ef
fi
c
ien
t
Iris
S
p
o
o
f
De
tec
ti
o
n
v
ia
Bo
o
ste
d
L
o
c
a
l
Bi
n
a
ry
Pa
tt
e
rn
s
.
In
tern
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
Bio
m
e
tri
c
s.
2
0
0
9
:
1
0
8
0
-
1
0
9
0
.
[3
1
]
Do
y
le J,
Bo
wy
e
r
K.
No
tre
Da
me
Ima
g
e
Da
t
a
b
a
se
fo
r C
o
n
t
a
c
t
L
e
n
s De
tec
ti
o
n
In
Iris R
e
c
o
g
n
it
i
o
n
-
2
0
1
3
.
2
0
1
4.
[3
2
]
Be
rtal
m
io
M
,
S
a
p
iro
G
,
Ca
se
ll
e
s
V
,
Ba
ll
e
ste
r
C.
Ima
g
e
I
n
p
a
in
ti
n
g
.
2
7
th
a
n
n
u
a
l
c
o
n
f
e
re
n
c
e
o
n
Co
m
p
u
ter
G
ra
p
h
ics
a
n
d
In
tera
c
ti
v
e
Tec
h
n
iq
u
e
s.
2
0
0
0
:
4
1
7
-
4
2
4
.
[3
3
]
Da
m
o
n
J.
P
r
o
p
e
rti
e
s
o
f
R
id
g
e
s
a
n
d
Co
re
s
f
o
r
Tw
o
-
Di
m
e
n
sio
n
a
l
I
m
a
g
e
s,
J
o
u
rn
a
l
o
f
M
a
th
e
ma
t
ica
l
Ima
g
i
n
g
a
n
d
Vi
sio
n
.
1
9
9
9
:
1
0
(
2
);
1
6
3
-
1
7
4
.
[3
4
]
Da
lal
N,
T
rig
g
s
B.
Histo
g
ra
ms
o
f
Or
ien
ted
Gr
a
d
ien
ts
f
o
r
Hu
ma
n
D
e
tec
ti
o
n
.
IEE
E
Co
m
p
u
ter
S
o
c
iety
Co
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
ter V
isi
o
n
a
n
d
P
a
tt
e
rn
Re
c
o
g
n
it
io
n
.
2
0
0
5
:
1
;
8
8
6
-
8
9
3
.
[3
5
]
L
o
we
D
G
.
Distin
c
ti
v
e
I
m
a
g
e
F
e
a
tu
re
s
f
ro
m
S
c
a
le
-
In
v
a
ri
a
n
t
Ke
y
p
o
in
ts.
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
C
o
mp
u
ter
V
isio
n
.
2
0
0
4
:
6
0
(
2
);
91
-
1
1
0
.
[3
6
]
Do
y
le
JS
,
Bo
wy
e
r
KW
,
F
l
y
n
n
P
J
.
Va
ria
ti
o
n
i
n
A
cc
u
ra
c
y
o
f
T
e
x
tu
re
d
Co
n
ta
c
t
L
e
n
s
De
tec
ti
o
n
b
a
se
d
o
n
S
e
n
so
r
a
n
d
L
e
n
s P
a
t
ter
n
.
IEE
E
S
ix
th
In
ter
n
a
t
io
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Bi
o
m
e
tri
c
s:
T
h
e
o
r
y
,
A
p
p
li
c
a
ti
o
n
s an
d
S
y
ste
m
s
.
2
0
1
3
:
1
-
7
.
[
3
7
]
Gu
n
a
w
an
,
T
.
S.,
So
lih
i
n
,
N.
S
.
,
Mo
r
s
h
id
i,
M.
A
.
,
&
Kar
ti
w
i
,
M.
(
2
0
1
7
)
.
Dev
elo
p
m
e
n
t
o
f
E
f
f
icien
t
I
r
is
I
d
en
tif
icat
io
n
Alg
o
r
it
h
m
u
s
i
n
g
W
av
ele
t
P
ac
k
et
s
f
o
r
S
m
a
r
tp
h
o
n
e
A
p
p
licatio
n
.
I
n
d
o
n
esia
n
Jo
u
r
n
a
l
o
f
E
lectrica
l E
n
g
in
ee
r
in
g
a
n
d
C
o
mp
u
ter S
cien
ce
,
8
(
2
)
.
[
3
8
]
C
h
e
n
,
D.
,
&
Qin
,
G.
(
2
0
1
7
)
.
A
n
E
m
b
ed
d
ed
I
r
is
I
m
a
g
e
Acq
u
is
itio
n
R
esear
ch
.
I
n
d
o
n
esia
n
Jo
u
r
n
a
l
o
f
E
lectrica
l E
n
g
in
ee
r
in
g
a
n
d
C
o
mp
u
ter S
cien
ce
,
5
(
1
)
,
9
0
-
98.
Evaluation Warning : The document was created with Spire.PDF for Python.