I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
pu
t
er
Science
Vo
l.
22
,
No
.
3
,
J
u
n
e
2
0
2
1
,
p
p
.
1404
~
1
4
1
0
I
SS
N:
2
5
0
2
-
4
7
5
2
,
DOI
: 1
0
.
1
1
5
9
1
/ijeecs.v
2
2
.i
3
.
pp
1
4
0
4
-
1
4
1
0
1404
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs.ia
esco
r
e.
co
m
H
a
lf
-
face ba
sed
r
ecog
nition us
ing
principa
l com
po
n
ent
a
na
ly
sis
Ahm
ed
M
.
Alk
a
ba
bji
1
,
Sa
ra
Ra
ed
Abd
2
1
De
p
a
rtme
n
t
o
f
Co
m
p
u
ter E
n
g
in
e
e
rin
g
,
U
n
iv
e
rsit
y
o
f
M
o
su
l,
Ira
q
2
Tec
h
n
ica
l
Co
m
p
u
ter E
n
g
i
n
e
e
rin
g
De
p
a
rtme
n
t,
Al
h
a
d
b
a
a
Un
iv
e
rsi
ty
Co
ll
e
g
e
,
M
o
su
l
,
Ira
q
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Oct
8
,
2
0
2
0
R
ev
is
ed
Ma
y
2
6
,
2
0
2
1
Acc
ep
ted
J
u
n
1
,
2
0
2
1
F
a
c
e
re
c
o
g
n
it
io
n
is
a
c
o
n
sid
e
ra
b
le
p
ro
b
lem
in
th
e
field
o
f
ima
g
e
p
r
o
c
e
ss
in
g
.
It
is
u
se
d
d
a
il
y
i
n
v
a
rio
u
s
a
p
p
li
c
a
ti
o
n
s
fr
o
m
p
e
rso
n
a
l
c
a
m
e
ra
s
to
fo
re
n
sic
in
v
e
stig
a
t
io
n
s.
M
o
st
o
f
t
h
e
p
r
o
v
id
e
s
so
l
u
ti
o
n
s
p
ro
p
o
se
d
b
a
se
d
o
n
fu
ll
-
fa
c
e
ima
g
e
s,
a
re
slo
w
to
c
o
m
p
u
te
a
n
d
n
e
e
d
m
o
re
sto
ra
g
e
.
In
th
is
p
a
p
e
r,
we
p
ro
p
o
se
a
n
e
ffe
c
ti
v
e
wa
y
to
re
d
u
c
e
t
h
e
f
e
a
tu
re
s
a
n
d
siz
e
o
f
t
h
e
d
a
tab
a
se
in
t
h
e
fa
c
e
re
c
o
g
n
it
i
o
n
m
e
th
o
d
a
n
d
t
h
u
s
we
g
e
t
a
n
in
c
re
a
se
in
th
e
sp
e
e
d
o
f
d
isc
rimin
a
ti
o
n
b
y
u
sin
g
h
a
lf
o
f
t
h
e
fa
c
e
.
Tak
i
n
g
a
d
v
a
n
tag
e
o
f
fa
c
e
sy
m
m
e
try
,
th
e
first
ste
p
is
to
d
iv
i
d
e
t
h
e
fa
c
e
ima
g
e
i
n
to
tw
o
h
a
lv
e
s,
th
e
n
t
h
e
left
h
a
lf
is
p
r
o
c
e
ss
e
d
u
sin
g
th
e
p
ri
n
c
ip
a
l
c
o
m
p
o
n
e
n
t
a
n
a
ly
s
is
(P
CA)
a
lg
o
rit
h
m
,
a
n
d
th
e
r
e
su
lt
s
a
re
c
o
m
p
a
re
d
b
y
u
si
n
g
E
u
c
li
d
ia
n
d
is
tan
c
e
to
d
isti
n
g
u
is
h
th
e
p
e
rso
n
.
T
h
e
sy
ste
m
wa
s tr
a
in
e
d
a
n
d
tes
ted
o
n
ORL
d
a
tab
a
se
.
It
wa
s fo
u
n
d
t
h
a
t
t
h
e
a
c
c
u
r
a
c
y
o
f
th
e
sy
ste
m
re
a
c
h
e
d
u
p
t
o
9
6
%
,
a
n
d
t
h
e
d
a
tab
a
se
wa
s
m
in
imiz
e
d
b
y
4
6
%
a
n
d
th
e
ru
n
n
in
g
ti
m
e
wa
s
d
e
c
re
a
se
d
fro
m
1
2
0
m
se
c
to
7
0
m
se
c
with
a
4
1
.
6
%
re
d
u
c
ti
o
n
.
K
ey
w
o
r
d
s
:
E
u
clid
an
d
is
tan
ce
Face
r
ec
o
g
n
itio
n
Half
f
ac
e
PC
A
R
ea
l tim
e
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
:
Ah
m
ed
M.
Alk
ab
a
b
jir
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
E
n
g
i
n
ee
r
in
g
Un
iv
er
s
ity
o
f
Mo
s
u
l
Mo
s
u
l,
I
r
aq
E
m
ail:
ah
m
ed
alk
ab
a
b
ji7
2
@
u
o
m
o
s
u
l.e
d
u
.
iq
1.
I
NT
RO
D
UCT
I
O
N
On
e
o
f
th
e
em
i
n
en
t
a
p
p
licatio
n
s
in
an
aly
zin
g
im
a
g
es
in
p
at
ter
n
r
ec
o
g
n
itio
n
an
d
co
m
p
u
te
r
v
is
io
n
is
f
ac
e
r
ec
o
g
n
itio
n
d
is
tin
g
u
is
h
es
its
elf
to
b
e.
I
ts
wo
r
k
is
b
ase
d
o
n
th
e
v
ar
iatio
n
o
f
f
ac
ial
c
h
ar
ac
ter
is
tics
.
Face
r
ec
o
g
n
itio
n
ca
n
b
e
f
o
u
n
d
in
v
ar
io
u
s
f
ield
s
in
o
u
r
life
s
tar
tin
g
f
r
o
m
h
ig
h
s
ec
u
r
ity
ap
p
licati
o
n
s
to
o
p
e
n
in
g
o
u
r
p
h
o
n
e.
T
wo
k
e
y
im
p
o
r
tan
t
te
ch
n
o
lo
g
ies
to
s
o
lv
e
th
e
p
r
o
b
l
em
s
o
f
f
ac
e
r
ec
o
g
n
itio
n
a
r
e
t
h
e
ex
tr
ac
tio
n
o
f
t
h
e
ch
ar
ac
ter
is
tics
o
f
f
ac
e
im
ag
es
an
d
th
e
s
elec
tio
n
o
f
a
s
u
itab
le
class
if
ier
ar
e.
Facial
f
ea
tu
r
es
ca
n
d
if
f
e
r
en
tiate
p
er
s
o
n
s
d
ep
e
n
d
in
g
o
n
th
eir
d
i
s
tin
ct
ch
ar
ac
ter
is
tics
.
Sev
er
al
f
ac
to
r
s
ca
n
af
f
ec
t
th
e
ap
p
ea
r
an
ce
o
f
a
n
in
d
iv
id
u
al,
f
o
r
ex
am
p
le
lig
h
tin
g
co
n
d
itio
n
s
,
b
r
ig
h
tn
ess
v
ar
iatio
n
s
,
d
if
f
er
en
t
em
o
tio
n
s
,
an
d
ag
in
g
f
a
cto
r
s
d
u
r
in
g
im
a
g
e
ac
q
u
is
itio
n
,
all
o
f
th
ese
asp
ec
t
s
af
f
ec
t th
e
ap
p
ea
r
an
ce
o
f
a
p
e
r
s
o
n
an
d
m
ak
e
it h
a
r
d
to
b
e
r
e
co
g
n
ized
[
1
]
-
[
5
]
.
Prin
cip
al
co
m
p
o
n
e
n
t
an
al
y
s
is
(
PC
A)
is
co
m
m
o
n
ly
u
s
ed
f
o
r
th
e
p
r
o
ce
s
s
in
g
o
f
f
ac
ial
im
a
g
es,
wh
ich
tak
es
ad
v
an
ta
g
e
o
f
c
h
ar
ac
ter
is
tic
f
ea
tu
r
es
ca
lled
“
E
ig
en
f
ac
es
”
.
Usi
n
g
E
ig
e
n
f
ac
es
it
tr
ies
to
cr
ea
te
a
“
p
r
in
cip
a
l
co
m
p
o
n
en
t
”
im
a
g
e
f
r
o
m
th
e
tr
ain
in
g
s
et
b
y
alig
n
in
g
th
e
m
o
u
th
,
ey
es
a
n
d
o
th
er
f
ac
ial
f
ea
tu
r
es
o
f
th
e
s
u
b
jects
i
n
th
e
s
ca
n
n
ed
im
ag
es,
k
ee
p
in
g
i
n
m
id
e
th
at
th
e
g
aller
y
an
d
p
r
o
b
e
im
ag
es
m
u
s
t
b
e
n
o
r
m
aliz
ed
in
th
e
s
am
e
s
ize.
T
h
en
,
a
m
eth
o
d
is
u
s
ed
in
r
ed
u
cin
g
th
e
d
im
en
s
io
n
ality
o
f
d
ata
b
y
im
ag
e
co
m
p
r
ess
io
n
m
ea
n
s
an
d
p
r
o
v
id
in
g
a
s
tr
u
ctu
r
e
o
f
t
h
e
m
o
s
t e
f
f
ec
tiv
e
lo
w
d
im
en
s
io
n
o
f
f
ac
ial
p
atter
n
.
Af
ter
th
is
r
ed
u
ctio
n
a
n
y
ir
r
elev
an
t in
f
o
r
m
atio
n
is
d
r
o
p
p
e
d
a
n
d
th
e
f
ac
ial
s
tr
u
ctu
r
e
is
d
ec
o
m
p
o
s
ed
in
to
u
n
co
r
r
elate
d
(
o
r
t
h
o
g
o
n
al)
co
m
p
o
n
en
ts
th
ese
co
m
p
o
n
en
ts
ar
e
id
en
tifie
d
as
E
ig
en
f
ac
es.
A
we
ig
h
ted
s
u
m
f
ea
tu
r
e
v
ec
to
r
o
f
eig
en
f
ac
es
ch
ar
ac
ter
i
ze
ea
ch
f
ac
e
im
ag
e.
T
h
ese
eig
e
n
f
ac
es
ar
e
k
e
p
t
in
a
o
n
e
-
d
im
en
s
io
n
al
ar
r
ay
,
wh
ich
f
o
r
a
s
et
o
f
g
aller
y
im
a
g
es
will
r
esu
lt
in
a
two
-
d
im
en
s
io
n
al
ar
r
ay
.
T
h
e
im
a
g
e
u
n
d
er
in
v
esti
g
atio
n
is
co
m
p
ar
ed
with
th
e
g
aller
y
im
ag
e.
T
h
e
m
atch
in
g
r
esu
lt is
d
ec
lar
ed
af
ter
a
co
m
p
u
tatio
n
o
f
th
e
d
is
tan
ce
b
etwe
en
th
e
im
ag
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
Ha
lf
-
fa
ce
b
a
s
ed
r
ec
o
g
n
itio
n
u
s
in
g
p
r
in
cip
a
l c
o
mp
o
n
en
t
a
n
a
l
ysis
(
A
h
med
M.
A
lka
b
a
b
ji
)
1405
f
ea
tu
r
e
v
ec
to
r
an
d
th
e
g
aller
y
.
T
h
is
lin
ea
r
m
et
h
o
d
is
wid
ely
u
s
ed
in
f
ac
e
r
ec
o
g
n
itio
n
,
wh
ich
b
y
lo
wer
d
i
m
e
n
s
io
n
r
ep
r
esen
tatio
n
aim
s
in
s
o
lv
in
g
th
e
p
r
o
b
lem
o
f
r
ec
o
g
n
itio
n
[
6
]
,
[
7
]
.
Ma
n
y
r
esear
ch
er
s
h
av
e
c
o
n
tr
i
b
u
ted
v
ar
i
o
u
s
id
ea
s
an
d
m
eth
o
d
s
to
d
is
tin
g
u
is
h
p
eo
p
le
b
y
f
ac
e.
T
h
is
is
b
ec
au
s
e
f
ac
e
r
ec
o
g
n
itio
n
is
an
ea
s
y
way
to
d
is
tin
g
u
is
h
p
eo
p
le
i
n
a
s
h
o
r
t
tim
e
an
d
with
g
o
o
d
ac
c
u
r
ac
y
.
Halv
i
et
a
l.
[
8
]
,
t
h
e
au
th
o
r
s
s
u
g
g
ested
a
way
to
d
is
tin
g
u
is
h
f
ac
es
th
r
o
u
g
h
o
n
e
-
d
i
m
en
s
io
n
al
(
1
D)
tr
an
s
f
o
r
m
d
o
m
ain
,
wh
ich
is
o
p
tio
n
e
d
b
y
tr
an
s
f
o
r
m
i
n
g
t
h
e
two
-
d
im
en
s
io
n
al
im
ag
e
f
ac
e
in
to
1
D.
T
h
e
f
ea
tu
r
es
ex
tr
ac
tio
n
f
r
o
m
d
is
cr
ete
wav
elet
tr
a
n
s
f
o
r
m
(
DW
T
)
a
n
d
f
ast
Fo
u
r
ier
tr
an
s
f
o
r
m
(
FF
T
)
ar
e
co
m
p
a
r
ed
with
d
ataset
u
s
in
g
E
u
clid
ian
d
is
tan
ce
(
E
D)
to
d
i
s
tin
g
u
is
h
f
ac
es.
Var
io
u
s
g
r
ad
i
en
t
an
d
L
ap
lacia
n
m
o
d
el
f
o
r
f
ac
e
r
ec
o
g
n
itio
n
is
p
r
o
p
o
s
ed
b
y
th
e
au
th
o
r
s
in
[
9
]
,
f
ea
tu
r
e
ex
tr
ac
ti
o
n
is
ac
h
iev
e
d
u
s
in
g
lin
ea
r
r
eg
r
ess
io
n
.
An
ed
g
e
d
etec
tio
n
f
ilter
co
n
v
er
ts
th
e
f
ac
e
im
ag
es
to
b
in
ar
y
as
p
r
ep
r
o
ce
s
s
in
g
th
en
ea
ch
im
ag
e
is
d
iv
id
ed
in
to
s
eg
m
en
ts
.
Nex
t
PC
A
i
s
u
s
ed
f
o
r
f
ea
tu
r
e
ex
tr
ac
tio
n
.
Fin
ally
,
a
tr
ain
ed
ar
tific
ial
n
eu
r
al
n
etwo
r
k
o
n
th
ese
f
ea
tu
r
es
is
th
e
to
o
l
f
o
r
class
if
icatio
n
p
u
r
p
o
s
es.
C
h
u
et
a
l
.
[
1
0
]
t
h
e
au
th
o
r
b
en
e
f
its
th
e
f
ac
t o
f
f
ac
ial
s
y
m
m
etr
y
b
y
p
r
o
p
o
s
in
g
a
m
u
ltip
le
f
ea
tu
r
e
s
u
b
s
p
ac
e
an
aly
s
is
(
MFSA)
ap
p
r
o
ac
h
.
T
h
e
f
ir
s
t
d
iv
is
io
n
o
f
th
e
f
ac
e
im
ag
e
is
ab
o
u
t
th
e
b
ilater
al
s
y
m
m
etr
y
ax
is
th
en
s
ev
er
al
lo
ca
l
f
ac
e
p
atch
es
ar
e
o
b
tain
ed
b
y
f
u
r
t
h
er
p
ar
titi
o
n
in
g
.
Patch
es
ar
e
th
en
la
b
eled
u
s
in
g
a
k
-
NN
class
if
ier
in
ea
ch
s
u
b
s
p
ac
e.
K
u
te
et
a
l
.
[
1
1
]
th
e
au
t
h
o
r
s
p
r
o
p
o
s
e
a
n
o
v
el
ap
p
r
o
ac
h
,
t
h
ey
c
laim
ed
th
at
p
ar
ts
o
f
th
e
f
ac
e
(
ea
r
,
lip
s
an
d
n
o
s
e)
h
a
v
e
co
m
m
o
n
in
f
o
r
m
ati
o
n
with
t
h
e
wh
o
le
f
ac
e
in
s
p
ite
th
e
f
ac
t th
at
t
h
ey
ar
e
f
o
r
m
d
if
f
er
en
t
d
o
m
ain
s
.
Hu
a
et
a
l
.
[
1
2
]
,
p
r
esen
ted
a
r
o
b
u
s
t
f
ac
e
r
e
co
g
n
itio
n
m
et
h
o
d
w
h
er
e
f
o
r
e
x
tr
ac
tin
g
th
e
v
ec
to
r
s
o
f
a
f
ea
tu
r
e
with
p
o
s
e
in
v
ar
ian
ce
an
d
s
ca
le
in
v
ar
ian
ce
f
r
o
m
f
ac
e
im
ag
es
th
e
s
p
ee
d
ed
-
u
p
r
o
b
u
s
t
f
ea
tu
r
e
(
SUR
F)
alg
o
r
ith
m
is
u
s
ed
.
PC
A
is
th
en
u
s
ed
in
p
r
o
d
u
ci
n
g
a
n
ew
f
e
atu
r
e
s
p
ac
e
as
PC
A
-
SUR
F
lo
ca
l
f
r
o
m
th
e
SUR
F
f
ea
tu
r
e
v
ec
to
r
s
.
Fin
ally
,
th
e
f
e
atu
r
e
p
o
in
ts
ar
e
clu
s
ter
b
y
th
e
K
-
m
ea
n
s
alg
o
r
ith
m
,
t
h
e
f
ac
e
i
m
ag
es a
r
e
class
if
ied
b
y
co
m
b
in
g
th
e
g
l
o
b
al
an
d
lo
c
al
s
im
ilar
ity
.
Gu
m
u
s
et
a
l
.
[
1
3
]
,
th
e
a
u
th
o
r
s
u
s
ed
an
E
i
g
en
f
ac
es
m
eth
o
d
wh
ic
h
is
PC
A
-
b
ased
an
d
wav
elet
d
ec
o
m
p
o
s
itio
n
to
e
x
tr
ac
t
f
ea
tu
r
es.
Af
ter
g
en
e
r
atin
g
f
ea
t
u
r
e
v
ec
t
o
r
s
,
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
(
SVM)
an
d
d
is
tan
ce
class
if
ier
s
ar
e
u
s
ed
f
o
r
th
e
clas
s
if
ic
atio
n
s
tag
e.
Face
r
ec
o
g
n
itio
n
u
s
in
g
h
alf
th
e
f
ac
e
is
n
o
t
a
n
ew
tech
n
iq
u
e,
o
v
er
th
e
last
two
d
ec
ad
es
s
tar
tin
g
at
2
0
0
3
a
p
aten
t
p
u
b
lis
h
ed
b
y
[
1
4
]
,
[
1
5
]
u
s
ed
lef
t
an
d
r
i
g
h
t
h
alf
im
a
g
es
o
f
t
h
e
f
ac
e
o
v
e
r
co
m
e
th
e
p
r
o
b
lem
s
ca
u
s
ed
b
y
d
ir
ec
tio
n
al
o
r
n
o
n
-
u
n
if
o
r
m
il
lu
m
in
atio
n
.
I
n
th
e
s
am
e
y
ea
r
[
1
6
]
s
o
lv
ed
th
e
p
r
o
b
lem
o
f
d
et
ec
tio
n
o
f
f
ac
es
with
lar
g
e
d
ep
th
r
o
tatio
n
s
b
y
u
s
in
g
h
alf
-
f
ac
e
tem
p
lates.
Mo
r
e
r
ec
en
tly
,
[
1
7
]
in
cr
ea
s
ed
th
e
ac
cu
r
ac
y
o
f
th
r
ee
-
d
im
en
s
io
n
al
f
ac
e
r
ec
o
g
n
itio
n
d
ep
e
n
d
in
g
o
n
f
ac
e
s
y
m
m
etr
y
.
L
ater
,
[
1
8
]
s
h
o
wed
in
th
ei
r
ex
p
e
r
im
en
t
t
h
at
a
s
o
lu
tio
n
to
th
e
p
r
o
b
lem
o
f
lar
g
e
an
g
le
in
s
id
e
f
ac
e
im
ag
es is
u
s
in
g
th
e
h
alf
-
f
ac
e
tem
p
late.
E
last
ic
b
u
n
ch
Gr
ap
h
m
atch
in
g
alg
o
r
ith
m
was
u
s
ed
b
y
[
1
9
]
f
o
r
h
alf
-
f
ac
e
r
ec
o
g
n
i
tio
n
.
Sh
eh
za
d
et
a
l.
[
2
0
]
a
g
o
o
d
h
al
f
-
f
ac
e
b
ased
r
ec
o
g
n
itio
n
s
y
s
tem
was
p
r
o
p
o
s
ed
,
b
u
t
th
e
co
m
p
u
tatio
n
tim
e
was
to
o
lo
n
g
an
d
th
e
r
ed
u
cti
o
n
in
d
atab
ase
s
ize
was
n
o
t
s
tu
d
ied
.
L
ast
y
ea
r
,
[
2
1
]
p
u
b
lis
h
ed
th
o
s
e
r
esear
ch
er
s
f
r
o
m
th
e
Un
iv
er
s
ity
o
f
B
r
ad
f
o
r
d
claim
th
at
in
f
ac
e
r
ec
o
g
n
itio
n
tech
n
o
lo
g
y
a
h
alf
-
f
ac
e
is
en
o
u
g
h
.
Af
ter
s
tu
d
y
in
g
th
e
liter
atu
r
e,
it wa
s
co
n
clu
d
ed
th
at
th
e
n
ee
d
o
f
a
s
im
p
le,
r
ea
l
tim
e
(
f
ast)
f
ac
e
r
ec
o
g
n
itio
n
s
y
s
tem
is
s
till
a
s
c
o
p
e
o
f
r
esear
ch
.
Mo
r
e
o
v
er
,
th
e
ad
v
a
n
tag
e
o
f
d
ata
s
ize
r
ed
u
ctio
n
wh
ich
ef
f
ec
t sto
r
ag
e
r
eq
u
ir
e
m
en
ts
as we
ll a
s
t
h
e
s
p
ee
d
o
f
d
ata
tr
an
s
f
er
o
n
n
e
two
r
k
s
,
b
y
s
en
d
in
g
h
alf
-
f
ac
e
d
ata
f
r
o
m
ca
m
er
a.
2.
T
H
E
P
RO
P
O
SE
D
SYS
T
E
M
Mo
tiv
ated
b
y
th
e
ad
v
a
n
tag
es
o
f
r
ed
u
cin
g
th
e
d
atab
ase
s
ize,
w
h
ich
ef
f
ec
t
th
e
s
p
ee
d
o
f
co
m
p
u
tatio
n
,
as
well
as,
m
in
im
izin
g
s
to
r
ag
e
ar
ea
in
f
ac
e
r
ec
o
g
n
itio
n
,
th
e
id
ea
o
f
tak
in
g
h
alf
th
e
f
ac
e
is
ex
p
lo
r
ed
.
C
an
we
im
ag
in
e
th
at
if
we
s
to
r
e
f
ac
e
im
ag
es
o
f
a
co
m
p
lete
city
o
r
co
u
n
tr
y
,
we
wo
u
ld
n
ee
d
a
m
em
o
r
y
o
f
a
v
er
y
lar
g
e
s
ize
to
s
to
r
e
th
is
d
atab
ase.
Ho
wev
e
r
,
th
r
o
u
g
h
th
e
p
r
o
p
o
s
ed
s
y
s
tem
,
we
ca
n
r
e
d
u
ce
t
h
e
s
ize
o
f
th
e
d
atab
ase
to
h
al
f
b
y
s
to
r
in
g
a
h
alf
-
f
ac
e
f
o
r
ea
ch
p
e
r
s
o
n
'
s
im
ag
e
b
ec
au
s
e
a
p
er
s
o
n
h
as
two
n
ea
r
ly
id
en
tical
h
alv
es
wh
en
h
is
f
ac
e
is
d
iv
id
ed
v
er
tically
in
to
two
h
a
lv
es.
T
h
e
r
esu
lts
p
r
o
v
ed
th
r
o
u
g
h
th
e
Ma
tlab
p
r
o
g
r
am
th
at
we
ca
n
d
is
tin
g
u
is
h
p
eo
p
le
th
r
o
u
g
h
o
n
ly
h
alf
o
f
th
e
f
ac
e.
T
h
e
p
r
o
p
o
s
ed
s
y
s
tem
c
an
b
e
ex
p
lain
ed
in
th
r
ee
s
tag
es.
2
.
1
.
P
re
-
pro
ce
s
s
ing
I
n
th
e
f
ir
s
t
s
tag
e,
s
u
p
p
o
s
e
th
e
im
ag
e
f
ac
e
I
(
x
,
y
)
b
e
a
two
-
d
im
en
s
io
n
al
in
ten
s
ity
v
alu
e
ar
r
ay
o
f
s
ize
A*
N.
T
h
e
o
r
ig
in
al
im
ag
e
wa
s
o
f
9
5
×1
1
2
p
i
x
els
s
ize.
W
h
ich
is
th
en
cr
o
p
p
ed
to
a
4
5
×
1
1
2
p
i
x
els
im
ag
e
in
d
atab
ase.
T
h
en
th
e
im
ag
e
is
co
n
v
er
ted
to
a
v
ec
to
r
o
f
d
im
en
s
io
n
5
,
0
4
0
.
Fig
u
r
e
1
s
h
o
ws
th
is
co
n
v
er
s
io
n
.
T
h
e
s
o
u
r
ce
im
ag
e
(
I
)
,
s
h
o
w
n
in
Fig
u
r
e
2
is
cu
t
to
th
e
s
ize
o
f
Mx
N
th
r
o
u
g
h
th
e
f
u
n
ctio
n
o
f
Ma
tlab
im
cr
o
p
(
im
ag
e
[
4
7
0
9
2
1
1
2
]
)
; w
h
ich
c
u
t th
e
i
m
ag
e
f
ac
e
to
h
alf
-
f
ac
e
.
T
h
e
o
b
tain
ed
r
esu
lt (
′
)
s
h
o
wn
in
Fig
u
r
e
2.
T
h
e
r
esu
lt
im
ag
e
′
o
f
cr
o
p
is
e
n
ter
ed
to
PC
A,
f
ea
tu
r
e
ex
tr
ac
t
io
n
as
well
as
d
im
en
s
io
n
ality
r
ed
u
ctio
n
is
ac
h
iev
ed
b
y
PC
A.
T
h
e
s
im
ilar
ity
b
etwe
en
th
e
im
ag
e
u
n
d
er
test
an
d
th
e
im
ag
es in
th
e
g
aller
y
(
o
n
e
b
y
o
n
e)
is
th
e
v
alu
e
o
f
th
e
ca
lcu
lated
E
u
clid
ea
n
d
is
tan
ce
b
etwe
en
th
eir
v
e
cto
r
s
in
th
e
f
ea
tu
r
e
s
p
ac
e.
T
h
e
m
in
im
u
m
th
e
v
alu
e
o
f
E
u
clid
ea
n
d
is
tan
ce
t
h
e
clo
s
est
is
th
e
two
im
ag
es
to
ea
ch
o
th
er
.
Fig
u
r
e
3
s
h
o
ws
th
e
f
lo
wch
ar
t
o
f
th
e
p
r
o
p
o
s
e
s
y
s
tem
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2502
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
22
,
No
.
3
,
J
u
n
e
2
0
2
1
:
1
4
0
4
-
1
4
1
0
1406
F
i
g
u
r
e
1
.
C
o
n
v
e
r
s
i
o
n
o
f
i
m
a
g
e
f
r
o
m
A
×
N
t
o
M
N
×
1
v
e
ct
o
r
Fig
u
r
e
2
.
C
r
o
p
th
e
in
p
u
t
im
ag
e
Fig
u
r
e
3
.
Pro
p
o
s
ed
s
y
s
tem
f
lo
wch
ar
t
2
.
2
.
P
rincipa
l c
o
m
po
nent
a
na
ly
s
is
Prin
cip
al
co
m
p
o
n
en
t
an
aly
s
is
(
PC
A)
is
a
p
o
wer
f
u
l
m
eth
o
d
,
it
is
u
s
ed
to
ex
tr
ac
t
th
e
f
ac
i
al
f
ea
tu
r
es
u
s
in
g
E
ig
en
f
ac
es.
T
h
e
p
u
r
p
o
s
e
o
f
th
e
PC
A
m
eth
o
d
is
p
r
im
a
r
ily
to
s
im
p
lify
t
h
e
v
ar
ia
b
les
o
b
s
er
v
ed
an
d
r
e
d
u
ce
th
eir
d
im
en
s
io
n
s
.
T
h
e
PC
A
also
h
as
two
m
ajo
r
ad
v
an
tag
es.
First,
th
e
way
o
f
ca
l
cu
latio
n
o
f
th
e
co
v
a
r
ain
ce
m
atr
ix
is
s
im
p
ler
.
Seco
n
d
,
we
d
o
n
'
t n
ee
d
a
lo
t
o
f
tim
es.
T
h
e
f
o
llo
win
g
ar
e
th
e
PC
A
s
tep
:
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
Ha
lf
-
fa
ce
b
a
s
ed
r
ec
o
g
n
itio
n
u
s
in
g
p
r
in
cip
a
l c
o
mp
o
n
en
t
a
n
a
l
ysis
(
A
h
med
M.
A
lka
b
a
b
ji
)
1407
−
C
o
n
s
tr
u
ct
th
e
f
ac
e
im
ag
e
N
×
N
o
f
im
ag
e
I
−
R
ea
d
all
f
ac
e
im
ag
es
−
Ad
ju
s
t th
e
im
ag
e
d
im
en
s
io
n
s
to
v
ec
to
r
s
ize
1
×
N
2
an
d
r
ep
r
es
en
t e
ac
h
im
ag
e
I
i
as v
ec
to
r
r
i
.
−
Dete
r
m
in
e
th
e
m
ea
n
m
atr
ix
=
1
∑
=
1
(
1
)
−
Su
b
s
tr
ac
t f
r
o
m
ea
c
h
v
ec
to
r
m
a
tr
ix
th
e
m
ea
n
m
atr
ix
=
−
(
2
)
−
Dete
r
m
in
e
th
e
co
v
a
r
ian
ce
m
at
r
ix
=
(
3
)
−
C
alcu
late
eig
en
v
alu
e
an
d
eig
e
n
v
ec
to
r
=
(
4
)
E
ig
en
v
alu
e
(
λ)
= D
et
(
C
–
λ
i
)
E
ig
en
v
ec
to
r
(u
i
)
=
(
C
–
λ
i
)
u
i
−
C
alcu
late
eig
en
f
ac
e
(
µ)
=
∑
=
1
(
5
)
W
h
er
e
is
th
e
m
ea
n
m
atr
ix
,
th
e
s
u
b
tr
ac
tio
n
m
atr
ix
,
th
e
co
v
a
r
ian
ce
m
atr
ix
,
λ
E
ig
en
v
al
u
e,
u
i
E
ig
en
v
ec
t
o
r
an
d
f
in
ally
µ
is
th
e
eig
en
f
ac
e.
As a
r
esu
lt,
lar
g
e
d
ec
r
ea
s
e
in
c
alcu
latio
n
is
o
b
tain
ed
,
f
r
o
m
th
e
o
r
d
er
o
f
n
u
m
b
e
r
o
f
p
ix
els in
th
e
im
ag
es
2
to
th
e
n
ew
o
r
d
er
,
wh
ich
is
th
e
n
u
m
b
er
o
f
im
a
g
es in
th
e
tr
ain
in
g
s
et
[
2
2
]
,
[
2
3
]
.
2
.
3
.
E
uclid
ia
n
d
is
t
a
nce
W
e
u
s
e
th
e
E
u
clid
ian
Dis
tan
c
e
to
m
ea
s
u
r
e
t
h
e
s
im
ilar
ity
b
e
twee
n
th
e
in
p
u
t
im
ag
e
a
n
d
t
h
e
d
ata
b
ase
im
ag
es.
I
t is wid
ely
u
s
ed
b
ec
a
u
s
e
o
f
its
s
im
p
licity
.
E
u
clid
ea
n
Dis
tan
ce
is
d
escr
ib
ed
b
y
t
h
e
(
7
)
as sh
o
wn
[
2
4
]
:
(
,
)
=
√
∑
(
−
)
2
=
1
(
7)
W
h
er
e
r
e
p
r
esen
t
th
e
f
ea
tu
r
es
n
u
m
b
er
,
r
ep
r
esen
t
t
h
e
f
ea
t
u
r
e
c
o
ef
f
icien
t
v
alu
es
o
f
test
im
ag
e,
is
th
e
f
ea
tu
r
e
co
ef
f
icien
t
v
alu
es
o
f
d
atab
ase
im
ag
es
an
d
(
,
)
is
th
e
E
u
clid
ea
n
Dis
tan
ce
b
etwe
en
th
e
tes
t
im
a
g
e
v
ec
to
r
an
d
th
e
d
atab
ase
im
ag
es v
ec
to
r
s
.
3.
I
M
P
L
E
M
E
NT
A
T
I
O
N
A
ND
T
E
ST
I
NG
T
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
im
p
le
m
en
ted
u
s
in
g
MA
T
L
AB
2
0
1
3
a
o
n
C
o
r
e
i7
-
8
5
5
0
U
C
PU
2
.
0
0
GHz
an
d
8
GB
o
f
R
AM
r
u
n
n
in
g
W
in
d
o
ws
1
0
(
6
4
b
it)
o
p
e
r
atin
g
s
y
s
tem
.
T
h
e
p
r
o
p
o
s
ed
s
y
s
tem
is
m
ad
e
u
p
o
f
th
r
ee
s
tep
s
:
p
r
e
-
p
r
o
ce
s
s
in
g
,
f
ea
tu
r
es
ex
tr
a
ctio
n
an
d
class
if
icatio
n
.
PC
A
is
u
s
ed
in
ex
tr
ac
tin
g
th
e
f
ea
tu
r
e
o
f
th
e
im
ag
e
an
d
E
u
clid
ian
d
is
tan
ce
is
u
s
ed
in
class
if
y
in
g
th
e
ex
t
r
ac
ted
f
ea
tu
r
es.
T
h
e
f
ir
s
t
two
p
r
in
cip
al
co
m
p
o
n
e
n
t
is
s
h
o
w
i
n
Fig
u
r
e
4
.
T
h
e
ex
p
er
im
e
n
t
is
ap
p
lied
i
n
ca
lcu
latin
g
th
e
s
y
s
tem
p
er
f
o
r
m
an
ce
o
n
OR
L
d
atab
ase.
T
h
is
d
atab
ase
is
m
ad
e
u
p
o
f
1
0
d
i
f
f
er
en
t
im
a
g
es
f
o
r
4
0
p
er
s
o
n
s
.
E
ac
h
im
ag
e
is
g
r
ay
with
a
r
eso
lu
tio
n
o
f
9
2
x
1
1
2
an
d
a
PGM
f
o
r
m
at.
Face
s
h
av
in
g
d
if
f
e
r
en
t
ex
p
r
ess
io
n
s
an
d
m
u
ltip
le
p
o
s
es
ar
e
f
o
u
n
d
in
th
is
d
atab
ase
[
2
5
]
.
T
h
e
d
ata
b
ase
was
u
s
ed
f
o
r
tr
ain
in
g
.
Oth
er
im
ag
es
th
at
ar
e
n
o
t
in
d
atab
ase
ar
e
en
r
o
lled
is
th
e
test
in
g
p
h
ase
to
ex
am
in
e
th
e
ef
f
ec
tiv
en
ess
o
f
th
e
s
y
s
tem
in
r
ec
o
g
n
izin
g
f
ac
es.
Fig
u
r
e
5
s
h
o
ws
th
e
r
esu
lt
f
o
r
a
p
er
s
o
n
f
o
u
n
d
in
th
e
d
ata
b
ase
an
d
h
o
w
is
n
o
t.
T
h
e
r
esu
lt
s
h
o
wed
th
at
th
e
ac
cu
r
ac
y
o
f
th
e
s
y
s
tem
r
ea
ch
e
d
u
p
to
9
6
%,
th
e
d
atab
ase
is
m
i
n
im
ized
b
y
4
6
%
an
d
th
e
co
m
p
u
tatio
n
tim
e
is
d
ec
r
ea
s
ed
f
r
o
m
1
2
0
to
7
0
m
s
ec
with
a
4
1
.
6
%
r
e
d
u
ctio
n
.
Du
e
to
th
is
r
ed
u
ctio
n
th
e
s
y
s
tem
ca
n
wo
r
k
m
o
r
e
e
f
f
ec
tiv
ely
in
r
ea
l
tim
e
s
y
s
tem
s
.
I
n
T
ab
les
1
a
n
d
2
a
co
m
p
ar
itio
n
is
p
r
esen
ted
with
s
o
m
e
p
r
ev
io
u
s
wo
r
k
f
o
r
th
e
co
m
p
u
tatio
n
tim
e
a
n
d
ac
c
u
r
ac
y
.
I
t
ca
n
b
e
s
ee
n
th
at
th
e
p
r
o
p
o
s
ed
s
y
s
tem
h
as
g
o
o
d
r
esu
lts
b
o
th
in
co
m
p
u
tatio
n
ti
m
e
an
d
ac
c
u
r
ac
y
as c
o
m
p
ar
ed
to
o
th
er
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2502
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
22
,
No
.
3
,
J
u
n
e
2
0
2
1
:
1
4
0
4
-
1
4
1
0
1408
Fig
u
r
e
4
.
T
h
e
f
ir
s
t two
p
r
i
n
cip
al
co
m
p
o
n
en
t
Fig
u
r
e
5
.
E
x
am
p
les o
f
th
e
s
y
s
tem
r
esu
lts
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
Ha
lf
-
fa
ce
b
a
s
ed
r
ec
o
g
n
itio
n
u
s
in
g
p
r
in
cip
a
l c
o
mp
o
n
en
t
a
n
a
l
ysis
(
A
h
med
M.
A
lka
b
a
b
ji
)
1409
T
ab
le
1
.
C
o
m
p
a
r
is
o
n
o
f
p
r
o
p
o
s
ed
m
eth
o
d
c
o
m
p
u
tatio
n
tim
e
with
ex
is
tin
g
m
eth
o
d
s
M
e
t
h
o
d
F
u
l
l
o
r
h
a
l
f
-
f
a
c
e
C
o
m
p
u
t
a
t
i
o
n
t
i
m
e
(
mse
c
)
S
h
a
r
ma
e
t
a
l
.
[
1
9
]
(
2
0
1
2
)
F
u
l
l
3
8
4
S
h
a
r
ma
e
t
a
l
.
[
1
9
]
(
2
0
1
2
)
H
a
l
f
99
Pr
o
p
o
sed
Ha
l
f
70
T
ab
le
2
.
C
o
m
p
a
r
is
o
n
o
f
p
r
o
p
o
s
ed
m
eth
o
d
ac
c
u
r
ac
y
with
e
x
is
tin
g
m
eth
o
d
s
M
e
t
h
o
d
F
u
l
l
o
r
H
a
l
f
-
f
a
c
e
R
e
c
o
g
n
i
t
i
o
R
a
t
e
(
%)
D
a
t
a
b
a
s
e
S
h
a
r
ma
e
t
a
l
.
[
1
9
]
(
2
0
1
2
)
F
u
l
l
83
3D
S
h
a
r
ma
e
t
a
l
.
[
1
9
]
(
2
0
1
2
)
H
a
l
f
9
5
.
3
3D
S
h
e
h
z
a
d
e
t
a
l
.
[
2
0
]
(
2
0
1
4
)
F
u
l
l
92
Essex
f
r
o
n
t
a
l
S
h
e
h
z
a
d
e
t
a
l
.
[
2
0
]
(
2
0
1
4
)
H
a
l
f
9
6
.
1
5
Essex
f
r
o
n
t
a
l
H
a
l
v
i
e
t
a
l
.
[
8
]
(
2
0
1
7
)
F
u
l
l
9
8
.
4
O
R
L
Pr
o
p
o
sed
Ha
l
f
96
OR
L
4.
CO
NCLU
SI
O
N
I
n
th
is
p
ap
er
,
p
r
i
n
cip
al
c
o
m
p
o
n
en
t
an
al
y
s
is
an
d
eig
en
f
ac
e
ap
p
r
o
ac
h
is
u
s
ed
to
im
p
lem
en
t
a
n
ac
cu
r
ate
an
d
r
o
b
u
s
t
s
y
s
tem
f
o
r
f
ac
e
r
e
c
o
g
n
itio
n
in
MA
T
L
AB
en
v
ir
o
n
m
en
t.
Star
tin
g
f
r
o
m
th
e
m
o
tiv
atio
n
to
r
ed
u
ce
t
h
e
d
atab
ase
s
ize
an
d
m
i
n
im
ize
th
e
co
m
p
u
tatio
n
tim
e,
A
n
u
m
b
e
r
o
f
h
alf
-
f
ac
e
im
ag
es
ar
e
p
r
o
c
ess
ed
in
th
e
s
y
s
tem
an
d
test
ed
to
r
ec
o
g
n
ize
th
e
p
e
r
s
o
n
.
T
h
e
r
esu
lt
p
r
o
v
ed
th
at
u
s
in
g
h
alf
-
f
ac
e
with
th
e
co
m
b
i
n
atio
n
o
f
p
r
in
cip
al
co
m
p
o
n
en
t
a
n
aly
s
is
an
d
eu
cli
d
ian
d
is
tan
ce
is
ef
f
ec
tiv
e
in
th
is
f
ac
e
r
ec
o
g
n
itio
n
s
y
s
tem
.
T
h
e
p
r
o
p
o
s
ed
s
y
s
tem
r
ed
u
ce
d
t
h
e
s
ize
o
f
d
ata
b
ase
an
d
th
e
s
p
ee
d
o
f
r
ec
o
g
n
itio
n
wi
th
h
ig
h
d
eg
r
ee
o
f
ac
c
u
r
ac
y
.
RE
F
E
R
E
NC
E
S
[1
]
S
.
Zh
a
n
g
,
Z.
L
i,
a
n
d
Y.
Li
u
,
“
Re
se
a
rc
h
o
n
fa
c
e
re
c
o
g
n
it
io
n
tec
h
n
o
l
o
g
y
b
a
se
d
o
n
P
CA
a
n
d
S
VM,
”
Pr
o
c
e
e
d
in
g
s
o
f
t
h
e
1
1
t
h
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
M
o
d
e
ll
in
g
,
I
d
e
n
ti
fi
c
a
ti
o
n
a
n
d
Co
n
tro
l
(
ICM
IC
2
0
1
9
)
,
p
p
.
7
5
-
8
5
,
2
0
2
0
,
d
o
i:
1
0
.
1
0
0
7
/
9
7
8
-
9
8
1
-
15
-
0
4
7
4
-
7_8
.
[2
]
S
.
S
h
a
rm
a
a
n
d
K.
S
a
c
h
d
e
v
a
,
“
F
a
c
e
Re
c
o
g
n
it
io
n
u
si
n
g
P
CA an
d
S
V
M
with
su
rf
tec
h
n
iq
u
e
,”
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
,
v
o
l.
1
2
9
,
n
o
.
4
,
p
p
.
4
1
-
4
6
,
2
0
1
5
.
[3
]
Z.
Wan
g
,
e
t
a
l.
,
“
M
a
sk
e
d
fa
c
e
re
c
o
g
n
it
io
n
d
a
tas
e
t
a
n
d
a
p
p
li
c
a
ti
o
n
,”
a
rXiv p
re
p
ri
n
t
a
rXiv
,
2
0
2
0
.
[4
]
A.
Ch
o
u
d
h
a
r
y
a
n
d
R.
Vi
g
,
“
F
a
c
e
re
c
o
g
n
it
i
o
n
u
sin
g
m
u
lt
ires
o
l
u
ti
o
n
wa
v
e
let
c
o
m
b
in
in
g
d
isc
re
te
c
o
sin
e
tran
sfo
rm
a
n
d
Walsh
tran
sf
o
rm
,”
Pro
c
e
e
d
i
n
g
s
o
f
th
e
2
0
1
7
I
n
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Bi
o
me
trics
En
g
in
e
e
rin
g
a
n
d
Ap
p
li
c
a
t
io
n
,
p
p
.
3
3
-
3
8
,
2
0
1
7
,
d
o
i:
1
0
.
1
1
4
5
/
3
0
7
7
8
2
9
.
3
0
7
7
8
3
5
.
[5
]
X.
F
o
n
tai
n
e
,
R.
Ac
h
a
n
ta
,
a
n
d
S
.
S
ü
ss
tru
n
k
,
“
F
a
c
e
re
c
o
g
n
it
i
o
n
i
n
re
a
l
-
wo
rld
ima
g
e
s,”
2
0
1
7
IEE
E
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Ac
o
u
stics
,
S
p
e
e
c
h
a
n
d
S
i
g
n
a
l
Pro
c
e
ss
in
g
(ICA
S
S
P)
,
2
0
1
7
,
p
p
.
1
4
8
2
-
1
4
8
6
,
d
o
i:
1
0
.
1
1
0
9
/ICA
S
S
P
.
2
0
1
7
.
7
9
5
2
4
0
3
.
[6
]
A.
S
i
n
g
h
,
S
.
K
.
S
in
g
h
,
a
n
d
S
.
Ti
wa
ri
,
“
C
o
m
p
a
riso
n
o
f
fa
c
e
re
c
o
g
n
it
i
o
n
a
lg
o
rit
h
m
s
o
n
d
u
m
m
y
fa
c
e
s
,”
T
h
e
In
ter
n
a
t
io
n
a
l
J
o
u
rn
a
l
o
f
M
u
l
ti
me
d
ia
&
Its
Ap
p
li
c
a
t
io
n
s
,
v
o
l.
4
,
p
p
.
1
2
1
-
1
3
5
,
2
0
1
2
,
d
o
i:
1
0
.
5
1
2
1
/i
jma
.
2
0
1
2
.
4
4
1
1
.
[7
]
P
riy
a
n
k
a
a
n
d
Y.
S
in
g
h
,
“
A
st
u
d
y
o
n
fa
c
ial
fe
a
tu
re
e
x
trac
ti
o
n
a
n
d
fa
c
ial
re
c
o
g
n
it
i
o
n
a
p
p
r
o
a
c
h
e
s,
”
In
ter
n
a
t
io
n
a
l
Jo
u
rn
a
l
o
f
Co
m
p
u
ter
S
c
ien
c
e
a
n
d
M
o
b
il
e
C
o
mp
u
t
in
g
,
v
o
l
.
4
,
n
o
.
5
,
p
p
.
1
6
6
-
1
7
4
,
2
0
1
4
.
[8
]
S
.
Ha
lv
i,
N
.
Ra
m
a
p
u
r,
K.
B
.
R
a
ja,
a
n
d
S
.
P
ra
sa
d
,
“
F
u
sio
n
b
a
s
e
d
fa
c
e
re
c
o
g
n
it
i
o
n
s
y
ste
m
u
sin
g
1
D
tra
n
sfo
rm
d
o
m
a
in
s
,”
Pr
o
c
e
d
ia
c
o
mp
u
ter
sc
ien
c
e
,
v
o
l
.
1
1
5
,
p
p
.
3
8
3
-
3
9
0
,
2
0
1
7
,
d
o
i
:
1
0
.
1
0
1
6
/j
.
p
r
o
c
s.2
0
1
7
.
0
9
.
0
9
5
.
[9
]
A.
Ala
z
z
a
wi,
O.
N.
Uc
a
n
,
a
n
d
O.
Ba
y
a
t,
“
P
e
rfo
rm
a
n
c
e
o
f
fa
c
e
re
c
o
g
n
it
i
o
n
sy
ste
m
u
si
n
g
g
ra
d
ien
t
lap
l
a
c
ian
o
p
e
ra
to
rs
a
n
d
n
e
w
fe
a
tu
re
s
e
x
trac
ti
o
n
m
e
th
o
d
b
a
se
d
o
n
l
in
e
a
r
re
g
re
ss
io
n
slo
p
e
,
”
M
a
t
h
e
ma
ti
c
a
l
Pro
b
lem
s
in
En
g
in
e
e
rin
g
,
2
0
1
8
,
d
o
i:
1
0
.
1
1
5
5
/2
0
1
8
/
1
9
2
9
8
3
6
.
[1
0
]
Y.
Ch
u
,
L.
Zh
a
o
,
a
n
d
T
.
Ah
m
a
d
,
“
M
u
lt
i
p
le
fe
a
tu
re
su
b
sp
a
c
e
s
a
n
a
ly
s
is
fo
r
si
n
g
le
sa
m
p
le
p
e
r
p
e
rso
n
fa
c
e
re
c
o
g
n
it
i
o
n
,”
T
h
e
Vi
su
a
l
C
o
mp
u
ter
,
v
o
l.
3
5
,
n
o
.
2
,
p
p
.
2
3
9
-
2
5
6
,
2
0
1
9
,
d
o
i:
1
0
.
1
0
0
7
/s0
0
3
7
1
-
0
1
7
-
1
4
6
8
-
4
.
[1
1
]
R.
S
.
Ku
te,
V.
Vy
a
s,
a
n
d
A.
An
u
se
,
“
Co
m
p
o
n
e
n
t
-
b
a
se
d
fa
c
e
re
c
o
g
n
it
i
o
n
u
n
d
e
r
tran
sfe
r
lea
rn
i
n
g
f
o
r
f
o
re
n
sic
a
p
p
li
c
a
ti
o
n
s
,”
In
f
o
rm
a
ti
o
n
S
c
ien
c
e
s
,
v
o
l
.
4
7
6
,
p
p
.
1
7
6
-
1
9
1
,
2
0
1
9
,
d
o
i:
1
0
.
1
0
1
6
/j
.
in
s.
2
0
1
8
.
1
0
.
0
1
4
.
[1
2
]
S
.
Hu
a
,
G
.
Ch
e
n
,
H.
Wei,
a
n
d
Q.
Ji
a
n
g
,
“
S
imilarit
y
m
e
a
su
re
fo
r
ima
g
e
re
siz
in
g
u
sin
g
S
IF
T
fe
a
tu
re
,”
EU
RA
S
IP
J
o
u
rn
a
l
o
n
Im
a
g
e
a
n
d
Vi
d
e
o
Pr
o
c
e
ss
in
g
,
2
0
1
2
,
d
o
i:
1
0
.
1
1
8
6
/
1
6
8
7
-
5
2
8
1
-
2
0
1
2
-
6
.
[1
3
]
E.
G
u
m
u
s,
N.
Kili
c
,
A.
S
e
rtb
a
s,
a
n
d
O.
N.
Uc
a
n
,
“
E
v
a
lu
a
ti
o
n
o
f
fa
c
e
re
c
o
g
n
it
i
o
n
tec
h
n
iq
u
e
s
u
sin
g
P
CA
,
wa
v
e
lets
a
n
d
S
VM
,”
Exp
e
rt S
y
ste
ms
wit
h
Ap
p
li
c
a
ti
o
n
s
,
v
o
l.
3
7
,
n
o
.
9
,
p
p
.
6
4
0
4
-
6
4
0
8
,
2
0
1
0
,
d
o
i:
1
0
.
1
0
1
6
/
j.
e
sw
a
.
2
0
1
0
.
0
2
.
0
7
9
.
[1
4
]
S
.
G
u
tt
a
,
M
.
Traj
k
o
v
ic,
a
n
d
V.
P
h
i
lo
m
in
,
“
F
a
c
e
-
re
c
o
g
n
it
io
n
u
si
n
g
h
a
lf
-
fa
c
e
ima
g
e
s
,
”
U.S
.
P
a
ten
t
US2
0
0
3
0
2
2
3
6
2
3
A
1
.
De
c
.
4
,
2
0
0
3
.
[1
5
]
S.
G
u
tt
a
a
n
d
H.
Wec
h
sle
r,
“
P
a
rti
a
l
fa
c
e
s
fo
r
fa
c
e
re
c
o
g
n
it
io
n
:
L
e
ft
v
s
rig
h
t
h
a
lf,
”
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Co
mp
u
ter
A
n
a
lys
is o
f
Im
a
g
e
s a
n
d
Pa
tt
e
rn
s
,
v
o
l
2
7
5
6
,
p
p
.
6
3
0
-
6
3
7
,
2
0
0
3
,
d
o
i:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
540
-
4
5
1
7
9
-
2
_
7
7
.
[1
6
]
Yi
Zh
u
a
n
d
F
.
Cu
t
u
,
“
F
a
c
e
De
tec
t
io
n
u
sin
g
Ha
lf
-
F
a
c
e
Tem
p
late
s
,”
J
o
u
rn
a
l
o
f
Vi
sio
n
,
v
o
l.
3
,
p
p
.
8
3
9
-
839
,
2
0
1
0
,
d
o
i:
1
0
.
1
1
6
7
/
3
.
9
.
8
3
9
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2502
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
22
,
No
.
3
,
J
u
n
e
2
0
2
1
:
1
4
0
4
-
1
4
1
0
1410
[1
7
]
J.
Ha
rg
u
e
ss
,
S
.
G
u
p
ta
a
n
d
J.
K.
A
g
g
a
rwa
l,
“
3
D
fa
c
e
re
c
o
g
n
it
io
n
wi
t
h
t
h
e
a
v
e
ra
g
e
-
h
a
lf
-
fa
c
e
,
”
2
0
0
8
1
9
t
h
I
n
ter
n
a
t
io
n
a
l
Co
n
fer
e
n
c
e
o
n
Pa
tt
e
rn
Rec
o
g
n
it
i
o
n
,
2
0
0
8
,
p
p
.
1
-
4
,
d
o
i:
1
0
.
1
1
0
9
/I
CP
R.
2
0
0
8
.
4
7
6
1
5
0
3
.
[1
8
]
W.
Ch
e
n
,
T.
S
u
n
,
X.
Ya
n
g
,
a
n
d
L.
Wan
g
,
“
F
a
c
e
d
e
tec
ti
o
n
b
a
s
e
d
o
n
h
a
lf
fa
c
e
-
tem
p
late
,
”
2
0
0
9
9
th
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
El
e
c
tro
n
ic M
e
a
s
u
re
me
n
t
&
In
stru
me
n
ts
,
2
0
0
9
,
p
p
.
4
-
54
-
4
-
5
8
,
d
o
i:
1
0
.
1
1
0
9
/ICE
M
I.
2
0
0
9
.
5
2
7
4
6
4
2
.
[1
9
]
V.
S
h
a
rm
a
a
n
d
R.
Va
sh
is
h
t,
“
A
v
e
ra
g
e
h
a
lf
fa
c
e
re
c
o
g
n
it
i
o
n
b
y
e
las
ti
c
b
u
n
c
h
g
ra
p
h
m
a
tch
in
g
b
a
se
d
o
n
d
istan
c
e
m
e
a
su
re
m
e
nt
,”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
f
o
r
S
c
ien
c
e
a
n
d
Eme
rg
in
g
T
e
c
h
n
o
l
o
g
ies
wit
h
L
a
tes
t
T
re
n
d
s
,
v
o
l.
3
,
n
o
.
1
,
p
p
.
24
-
3
5
,
2
0
1
2
.
[2
0
]
M
.
S
h
e
h
z
a
d
,
M
.
Aw
a
is,
M
.
Am
i
n
,
a
n
d
Y.
A.
S
h
a
h
,
“
F
a
c
e
Re
c
o
g
n
it
io
n
u
si
n
g
a
v
e
ra
g
e
h
a
lf
fa
c
e
tem
p
late
,”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Tec
h
n
o
l
o
g
y
,
v
o
l.
5
,
n
o
.
2
,
p
p
.
1
5
9
-
1
6
8
,
2
0
1
4
,
d
o
i:
1
0
.
1
4
7
1
6
/i
jt
e
c
h
.
v
5
i
2
.
4
0
8
.
[2
1
]
Un
iv
e
rsity
o
f
Bra
d
fo
r
d
,
Ha
lf
a
fa
c
e
e
n
o
u
g
h
fo
r
re
c
o
g
n
it
i
o
n
tec
h
n
o
l
o
g
y
,
S
c
ien
c
e
Da
il
y
,
M
a
y
2
0
1
9
.
[On
l
in
e
].
Av
a
il
a
b
le:
ww
w.sc
ien
c
e
d
a
il
y
.
c
o
m
/rele
a
se
s/2
0
1
9
/0
5
/1
9
0
5
0
1
1
1
4
6
0
2
.
h
tm
[2
2
]
M.
T
u
rk
a
n
d
A.
P
e
n
tl
a
n
d
,
“
E
ig
e
n
fa
c
e
s
fo
r
r
e
c
o
g
n
it
i
o
n
,”
J
o
u
r
n
a
l
o
f
Co
g
n
it
ive
Ne
u
ro
sc
ien
c
e
,
v
o
l.
3
,
n
o
.
1
,
p
p
.
7
1
-
8
6
,
1
9
9
1
,
d
o
i:
1
0
.
1
1
6
2
/
jo
c
n
.
1
9
9
1
.
3
.
1
.
7
1
.
[2
3
]
J.
C.
S
u
th
e
rlan
d
a
n
d
A.
P
a
re
n
te
,
“
Co
m
b
u
stio
n
m
o
d
e
li
n
g
u
sin
g
p
ri
n
c
ip
a
l
c
o
m
p
o
n
e
n
t
a
n
a
ly
sis,
”
Pro
c
e
e
d
in
g
s
o
f
th
e
Co
mb
u
sti
o
n
In
stit
u
te
,
v
o
l.
3
2
,
n
o
.
1
,
p
p
.
1
5
6
3
-
1
5
7
0
,
2
0
0
9
,
d
o
i:
1
0
.
1
0
1
6
/j
.
p
ro
c
i.
2
0
0
8
.
0
6
.
1
4
7
.
[2
4
]
Li
we
i
Wan
g
,
Ya
n
Z
h
a
n
g
,
a
n
d
Ju
f
u
F
e
n
g
,
“
On
th
e
Eu
c
li
d
e
a
n
d
ista
n
c
e
o
f
ima
g
e
s,”
in
I
EE
E
T
r
a
n
s
a
c
ti
o
n
s
o
n
Pa
tt
e
rn
An
a
lys
is
a
n
d
M
a
c
h
i
n
e
In
tell
ig
e
n
c
e
,
v
o
l
.
2
7
,
n
o
.
8
,
p
p
.
1
3
3
4
-
1
3
3
9
,
Au
g
.
2
0
0
5
,
d
o
i:
1
0
.
1
1
0
9
/T
P
AMI
.
2
0
0
5
.
1
6
5
.
[2
5
]
F
a
c
e
re
c
o
g
n
it
i
o
n
h
o
m
e
p
a
g
e
,
Da
tab
a
se
s,
2
0
0
5
.
[On
li
n
e
].
Av
a
il
b
a
le:
h
tt
p
s:/
/www
.
fa
c
e
-
re
c
.
o
rg
/d
a
tab
a
se
s/
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Ahm
e
d
M
.
Alk
a
b
a
b
ji
,
re
c
e
iv
e
d
t
h
e
B.
S
c
.
,
M
.
S
c
.
a
n
d
P
h
D
d
e
g
re
e
i
n
El
e
c
tri
c
a
l
E
n
g
i
n
e
e
rin
g
fro
m
Un
iv
e
rsit
y
o
f
M
o
su
l,
Ira
q
in
1
9
9
4
.
1
9
9
6
a
n
d
2
0
0
7
.
Cu
rre
n
tl
y
,
a
n
a
ss
istan
t
p
ro
fe
ss
o
r
a
t
c
o
m
p
u
ter
e
n
g
in
e
e
rin
g
d
e
p
a
rtme
n
t
in
Un
i
v
e
rsity
o
f
M
o
su
l
,
Ira
q
.
R
e
se
a
rc
h
in
tere
st
in
sig
n
a
l
p
ro
c
e
ss
in
g
,
re
a
lt
ime
sy
ste
m
s a
n
d
b
io
m
e
tri
c
e
n
g
in
e
e
rin
g
.
S
a
r
a
R
a
e
d
Ab
d
,
re
c
e
iv
e
d
th
e
B
.
S
.
a
n
d
M
.
S
.
d
e
g
re
e
s
fro
m
t
h
e
De
p
a
rtme
n
t
o
f
C
o
m
p
u
ter
En
g
i
n
e
e
rin
g
a
t
th
e
M
o
s
u
l
Un
i
v
e
rs
it
y
i
n
2
0
1
3
a
n
d
2
0
1
8
re
sp
e
c
ti
v
e
ly
.
S
h
e
is
c
u
rre
n
tl
y
wo
r
k
in
g
to
wa
rd
th
e
P
h
D
d
e
g
re
e
in
C
o
m
p
u
ter
En
g
i
n
e
e
rin
g
a
t
t
h
e
sa
m
e
u
n
iv
e
rsity
.
S
h
e
is
in
tere
ste
d
in
d
o
i
n
g
re
se
a
rc
h
in
wire
les
s
se
n
s
o
r
n
e
tw
o
rk
s,
c
ro
ss
-
lay
e
r
d
e
sig
n
fo
r
WS
Ns
,
B
o
d
y
Are
a
Ne
two
rk
,
I
o
T,
a
n
d
b
io
m
e
tri
c
se
c
u
rit
y
.
Evaluation Warning : The document was created with Spire.PDF for Python.