I
n
t
ern
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
E
l
ect
ri
ca
l
a
n
d
C
o
m
p
u
t
er E
n
g
i
n
eeri
n
g
(
I
J
E
C
E
)
V
o
l.
8
,
No
.
5
,
O
c
t
obe
r
20
1
8
,
p
p.
28
12~
2
817
I
S
S
N
:
2088
-
8708
,
D
O
I
:
10.
11
591/
i
j
ece
.
v8
i
5
.
pp
281
2
-
2817
2812
Jou
r
n
al
h
om
e
p
age
:
h
ttp
:
//ia
e
s
c
o
r
e
.
c
o
m/
j
our
nal
s
/
i
nde
x
.
php/
I
J
E
C
E
Rea
l
-
t
i
me
M
ult
i
-
o
bj
ect
F
a
ce
Reco
g
nit
io
n Us
ing
Co
nt
ent
B
a
s
ed
I
m
a
g
e Re
t
riev
a
l (
CB
I
R)
M
uha
mm
a
d F
a
c
hr
ur
r
o
z
i
1
,
S
a
pa
r
udi
n
2
,
Er
w
i
n
3
,
M
a
r
di
a
na
4
,
C
la
r
a
F
in
B
a
d
illa
h
5
,
J
uni
a
E
r
l
i
na
6
,
A
uz
a
n
L
a
z
ua
r
di
7
1
,2
,5
,6
,7
I
n
f
o
r
m
at
i
cs
E
n
g
i
n
eer
i
n
g
D
ep
ar
t
m
en
t
,
F
acu
l
t
y
o
f
C
o
m
p
u
t
er
S
ci
en
ce,
U
n
i
v
er
s
i
t
as
S
r
i
w
i
j
a
y
a,
I
ndo
ne
s
i
a
3
C
om
put
e
r
E
ng
i
ne
e
r
i
ng
D
e
pa
r
t
m
e
nt
,
F
acu
l
t
y
o
f
C
o
m
p
u
t
er
S
ci
en
ce,
U
n
i
v
er
s
i
t
as
S
r
i
w
i
j
a
y
a,
I
ndone
s
i
a
4
L
a
w
D
ep
ar
t
m
en
t
,
F
a
c
u
lty
o
f
L
a
w
,
U
n
iv
e
r
s
ita
s
S
r
iw
ija
y
a
,
I
ndone
s
i
a
A
rt
i
cl
e I
n
f
o
AB
S
T
RAC
T
A
r
tic
le
h
is
to
r
y
:
R
ecei
v
ed
No
v
22
,
201
7
Re
v
i
se
d
Fe
b
12
,
2
01
8
A
ccep
t
ed
Se
p
1
0
,
2
01
8
F
ace r
eco
g
n
i
t
i
o
n
s
y
s
t
e
m
i
n
r
eal
t
i
m
e
i
s
d
i
v
i
d
ed
i
n
t
o
t
h
r
ee p
r
o
ces
s
es
,
n
a
m
el
y
f
e
at
u
r
e
ex
t
r
act
i
o
n
,
cl
u
s
t
er
i
n
g
,
d
et
ect
i
o
n
,
an
d
r
eco
g
n
i
t
i
o
n
.
E
ach
o
f
t
h
es
e s
t
ag
e
s
us
e
s
di
f
f
e
r
e
nt
m
e
t
hods
,
L
oc
a
l
B
i
na
r
y
P
a
t
t
e
r
n (
L
B
P
)
,
A
gg
l
om
e
r
a
t
i
ve
H
i
er
ar
ch
i
cal
C
l
u
s
t
er
i
n
g
(
A
H
C
)
an
d
E
u
cl
i
d
ean
D
i
s
t
an
ce
.
M
u
lti
-
f
ace i
m
a
g
e
s
ear
ch
u
s
i
n
g
C
o
n
t
en
t
B
as
ed
I
m
ag
e R
et
r
i
ev
al
(
C
B
IR)
m
e
t
h
o
d
.
CBI
R p
e
rf
o
r
m
s
i
m
a
g
e s
e
ar
ch
b
y
i
m
a
g
e f
eat
u
r
e i
t
s
el
f
.
B
as
ed
o
n
r
eal
t
i
m
e t
r
i
al
r
es
u
l
t
s
,
t
h
e
accu
r
ac
y
v
al
u
e o
b
t
ai
n
ed
i
s
6
1
.
6
4
%
.
Ke
y
wo
rd
:
AH
C
CB
I
R
E
u
cl
i
d
ean
d
is
ta
n
c
e
F
ace
r
e
c
o
g
n
itio
n
L
B
P
M
u
lti
-
o
b
j
ect
R
eal
-
t
i
me
C
opy
r
i
g
ht
©
201
8
I
n
s
tit
u
te
o
f
A
d
v
anc
e
d
E
ngi
ne
e
r
i
ng an
d Sc
i
e
nc
e
.
A
l
l
ri
g
h
t
s re
se
rv
e
d
.
Co
rre
sp
o
n
d
i
n
g
Au
t
h
o
r
:
M
uha
m
m
a
d
F
a
c
hr
ur
r
o
z
i
,
I
n
f
o
r
m
a
t
i
cs
E
n
g
i
n
eer
i
n
g
D
ep
a
r
t
m
en
t
,
F
acu
l
t
y
o
f
C
o
m
p
u
t
er
S
ci
en
ce,
U
n
i
v
e
r
s
ita
s
S
r
i
w
ij
a
y
a
,
I
n
d
r
al
ay
a,
I
nd
o
ne
s
i
a
.
E
m
a
il:
m
f
ach
r
z
@
u
n
s
r
i
.
ac.
i
d
1.
I
NT
RO
D
UCT
I
O
N
F
a
c
e
r
e
c
o
gni
t
i
o
n o
n m
ul
t
i
-
o
b
j
e
ct
i
m
ag
er
y
can
b
e i
d
en
t
i
f
i
e
d
a
n
d
r
e
co
g
n
i
ze
d
.
T
h
e f
ace i
s
o
n
e
o
f
h
u
m
a
n
b
i
o
l
o
gi
c
a
l
i
nf
o
r
m
a
t
i
o
n
i
n
a
dd
i
t
i
o
n
t
o
f
i
nge
r
p
r
i
nt
s
,
e
y
e
s
a
nd
vo
i
c
e
.
T
he
pr
o
c
e
s
s
o
f
t
he
i
nt
r
o
d
u
c
t
i
o
n
o
f
m
ul
t
i
-
o
b
j
e
ct
f
a
ce i
m
ag
e i
n
r
e
al
-
t
i
m
e d
et
ec
t
i
o
n
p
r
o
c
es
s
be
gi
ns
w
i
t
h t
he
f
a
c
e
,
t
h
e
n t
r
a
c
ki
ng t
o
d
e
t
e
r
m
i
ne
t
he
l
o
c
a
t
i
o
n
o
f
t
h
e f
ace t
o
b
e
r
e
co
g
n
i
z
e
d
[1
]
.
F
ace
d
et
ec
t
i
o
n
i
s
d
o
n
e
b
y
s
ep
ar
at
i
n
g
t
h
e f
ace an
d
b
ack
g
r
o
u
n
d
an
d
d
et
er
m
i
n
i
n
g
t
h
e l
o
cat
i
o
n
,
s
i
z
e an
d
n
u
m
b
er
o
f
f
ace
o
b
j
ec
t
s
p
r
es
en
t
i
n
t
h
e
i
m
ag
e
[2
]
.
T
h
e
d
et
ec
t
e
d
f
ac
e
i
s
ex
t
r
a
ct
ed
u
s
i
n
g
L
o
cal
B
i
n
ar
y
P
at
t
er
n
(
L
B
P
)
.
V
e
ct
o
r
f
eat
u
r
e
e
x
tr
a
c
ti
o
n
r
e
s
u
lts
w
ith
tr
a
in
e
r
im
a
g
e
is
c
la
s
s
if
i
e
d
u
s
in
g
H
i
e
r
a
r
ch
i
cal
A
g
g
l
o
m
er
at
i
v
e C
l
u
s
t
er
i
n
g
(
A
H
C
)
,
t
h
en
c
o
m
p
a
r
ed
w
i
t
h
t
h
e f
eat
u
r
e v
e
ct
o
r
o
f
t
h
e t
es
t
i
m
ag
es
u
s
i
n
g
t
h
e E
u
cl
i
d
e
an
d
i
s
t
an
c
e.
T
h
i
s
p
r
o
ces
s
i
s
b
as
e
d
C
o
n
t
en
t
B
as
ed
I
m
a
g
e
R
et
r
i
e
v
al
(
CB
I
R
)
due
t
o
t
he
s
e
a
r
c
h f
or
v
ect
o
r
f
ea
t
u
r
es
i
n
t
h
e
d
at
ab
as
e
i
s
d
o
n
e
i
n
a
cc
o
r
d
an
ce
w
i
t
h
t
h
e
v
i
s
u
al
c
o
n
t
en
t
o
f
t
h
e i
m
ag
e
i
t
s
el
f
[
3]
.
C
o
n
t
e
n
t
B
as
ed
I
m
a
g
e
R
et
r
i
e
v
al
h
as
s
ev
er
al
p
r
o
c
es
s
es
t
o
g
e
t
g
o
o
d
r
e
s
ul
t
s
,
o
ne
o
f
t
he
pr
oc
e
s
s
i
s
b
y
f
eat
u
r
e ex
t
r
a
ct
i
o
n
t
o
g
e
t
t
h
e
v
al
u
e
o
f
e
ach
f
a
ci
al
f
ea
t
u
r
e
[4
]
.
T
h
e
L
o
c
al
B
i
n
ar
y
P
a
t
t
er
n
a
lg
o
r
ith
m
is
c
a
p
a
b
le
o
f
c
ha
r
a
c
t
e
r
i
z
i
ng a
nd
d
i
s
t
i
ngui
s
hi
ng
s
ur
f
a
c
e
t
e
xt
ur
e
s
.
L
o
cal
B
i
n
ar
y
P
at
t
er
n
ha
s
a
w
e
a
kne
s
s
i
n e
xt
r
a
c
t
i
ng t
h
e
te
x
tu
r
e
o
f
th
e
f
a
c
e
,
th
is
is
d
u
e
to
L
o
c
a
l B
in
a
r
y
P
a
tte
r
n
a
lg
o
r
ith
m
to
e
x
tr
a
c
t a
ll im
a
g
e
s
w
ith
o
u
t d
is
tin
g
u
is
h
in
g
t
he
f
a
c
e
a
nd ba
c
kgr
o
und
[5
]
.
H
ig
h
a
c
c
u
r
a
c
y
c
a
n
b
e
a
c
h
ie
v
e
d
if
th
e
a
lg
o
r
i
th
m
is
im
p
le
m
e
n
te
d
o
n
a
te
x
tu
r
e
w
ith
l
o
w
v
ar
i
an
c
e
[6
]
.
I
m
a
ge
c
l
u
s
t
e
r
i
ng us
i
ng H
i
e
r
a
r
c
hi
c
a
l
C
l
us
t
e
r
i
ng A
l
gor
i
t
hm
c
a
n i
m
p
r
o
ve
t
h
e
s
pe
e
d
a
nd
accu
r
a
cy
o
f
i
m
ag
e
m
at
ch
i
n
g
o
n
C
o
n
t
e
n
t
B
as
ed
I
m
ag
e
R
et
r
i
e
v
al
[7
]
,
[
8]
.
A
g
g
lo
m
e
r
a
tiv
e
H
ie
r
a
r
c
h
i
c
a
l C
lu
s
te
r
in
g
M
e
t
hod
s
us
e
d
i
n f
a
c
e
r
e
c
o
gni
t
i
o
n s
y
s
t
e
m
s
a
r
e
t
he
S
i
ngl
e
L
i
nka
ge
,
C
o
m
p
l
e
t
e
L
i
nka
ge
a
nd A
ve
r
a
ge
L
i
nka
ge
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t
J
E
l
e
c
&
C
o
m
p
E
n
g
I
S
S
N
:
2088
-
8708
R
ea
l
-
tim
e
M
u
lti
-
o
bj
e
c
t
F
ac
e
R
e
c
ogni
t
i
on U
s
i
ng C
ont
e
nt
B
as
e
d I
m
age
.
...
(
M
uham
m
ad F
ac
hr
ur
r
oz
i
)
2813
T
es
t
i
m
ag
e an
d
t
r
ai
n
i
n
g
i
m
ag
e i
s
ex
t
r
act
ed
u
s
i
n
g
e
i
g
en
f
ace
al
g
o
r
i
t
h
m
.
E
u
cl
i
d
ean
D
i
s
t
an
c
e p
l
ay
s
a r
o
l
e
to
f
in
d
th
e
m
in
im
u
m
d
is
ta
n
c
e
b
e
tw
e
e
n
te
s
t im
a
g
e
a
n
d
tr
a
in
i
m
ag
e,
an
d
o
b
t
a
i
n
e
d
an
ac
cu
r
a
cy
o
f
9
4
%
[9
]
.
ch
an
g
i
n
g
f
aci
a
l
ex
p
r
es
s
i
o
n
s
,
l
i
g
h
t
an
d
p
o
s
e
d
i
f
f
er
en
ces
i
s
o
n
e
o
f
t
h
e m
ai
n
f
act
o
r
s
t
h
a
t
g
r
eat
l
y
af
f
ect
t
h
e
p
r
o
ces
s
o
f
f
aci
al
r
e
co
g
n
i
t
i
o
n
[1
0
]
.
s
uc
h c
ha
nge
s
m
a
y
a
l
t
e
r
t
he
n
o
n
-
c
o
n
v
e
x
f
a
ce
s
h
ap
e,
s
el
f
-
o
c
c
l
u
s
i
o
n
a
n
d
no
nl
i
ne
a
r
f
o
r
m
c
ha
nge
s
w
hi
c
h
m
a
y
co
m
p
l
i
cat
e t
h
e c
l
as
s
i
f
i
cat
i
o
n
[1
1
]
.
I
n
th
is
m
u
lti
-
o
b
j
e
c
t f
a
c
e
r
e
c
o
g
n
itio
n
s
y
s
te
m
,
th
e
f
ace
d
et
ect
i
o
n
p
r
o
ces
s
i
s
d
o
n
e
b
ef
o
r
e t
h
e f
ac
e ex
t
r
a
ct
i
o
n
p
r
o
c
es
s
,
s
o
t
h
at
t
h
e f
eat
u
r
es
o
b
t
ai
n
ed
r
i
g
h
t
o
n
t
h
e f
ace
.
T
h
e
n
th
e
f
e
a
tu
r
e
is
i
d
e
n
tif
ie
d
b
y
f
in
d
in
g
th
e
d
is
ta
n
c
e
o
f
E
u
c
lid
e
a
n
d
is
ta
n
c
e
b
e
tw
e
e
n
th
e
te
s
t im
a
g
e
v
e
c
to
r
f
e
a
t
ur
e
a
nd t
he
t
r
a
i
ni
ng
i
m
a
ge
r
e
s
ul
t
i
ng f
r
o
m
t
he
f
e
a
t
ur
e
e
xt
r
a
c
t
i
o
n s
t
a
ge
us
i
ng
L
o
c
a
l B
in
a
r
y
P
a
t
te
r
n
(
L
B
P
)
.
2.
F
ACE
RE
CO
G
N
I
T
I
O
N
2.
1.
L
o
c
a
l
B
i
n
a
r
y
P
a
tte
r
n
(L
B
P
)
L
o
cal
B
i
n
a
r
y
P
at
t
e
r
n
(
L
B
P
)
r
ep
r
es
en
t
s
a p
i
x
el
w
h
i
c
h
f
o
r
m
ed
b
y
a 3
x
3
m
at
r
i
x
a
s
a c
o
m
p
ar
i
s
o
n
b
e
t
w
e
e
n t
he
c
e
n
t
e
r
p
i
xe
l
a
nd
i
t
s
s
ur
r
o
u
nd
i
n
g p
i
xe
l
w
hi
c
h
t
he
n c
o
nve
r
t
e
d
i
nt
o
b
i
na
r
y
n
u
m
b
e
r
s
.
T
he
co
m
p
ar
i
s
o
n
as
s
u
m
es
t
h
at
i
f
t
h
e s
u
r
r
o
u
n
d
ed
p
i
x
el
v
al
u
e i
s
g
r
eat
er
t
h
an
t
h
e cen
t
r
al
p
i
x
el
v
al
u
e
th
a
n
it
w
ill b
e
1
o
t
h
er
w
i
s
e 0
.
A
f
t
er
w
e g
et
8
b
i
n
ar
y
n
u
m
b
er
s
i
n
each
p
i
x
el
t
h
en
i
t
w
i
l
l
b
e r
ep
l
aced
w
i
t
h
t
h
e d
eci
m
al
f
o
r
m
t
o
g
et
th
e
r
e
s
u
lt.
T
h
e L
B
P
al
g
o
r
i
t
h
m
eq
u
at
i
o
n
can
b
e ex
p
r
es
s
ed
as
t
h
e f
o
l
l
o
w
i
n
g
f
o
r
m
u
l
a:
p
c
p
p
c
c
g
g
f
y
x
LBP
2
)
(
)
,
(
7
0
−
=
∑
=
(1
)
wh
e
r
e
:
g
p
:
cen
t
r
al
p
i
x
el
v
al
u
e
g
c
:
t
h
e p
i
x
el
v
al
u
e ar
o
u
n
d
t
h
e c
en
t
er
p
:
n
u
m
b
er
o
f
p
i
x
el
s
ar
o
u
n
d
t
h
e cen
t
er
A
n
d
th
e
f
u
n
c
tio
n
f
(
x
)
is
d
e
f
i
n
e
d
a
s
f
o
llo
w
s
:
{
}
0
,
1
0
,
0
)
(
≥
<
=
x
x
x
f
(2
)
2.
2.
H
i
era
rch
i
ca
l
C
l
u
s
t
eri
n
g
H
i
er
ar
ch
i
cal
cl
u
s
t
er
i
n
g
i
s
u
s
ed
t
o
g
r
o
u
p
s
i
m
i
l
ar
i
m
a
g
er
y
i
n
t
o
cl
u
s
t
er
s
t
o
i
n
cr
eas
e t
h
e s
p
eed
o
f
s
e
a
r
c
hi
ng
i
m
a
ge
[
2
]
.
I
n t
hi
s
a
l
go
r
i
t
h
m
,
t
he
r
e
is
a h
i
er
ar
ch
a
l
tr
e
e
w
hi
c
h pr
ov
i
de
s
a
v
i
e
w
of
s
e
v
e
r
a
l
l
e
v
e
l
s
of
da
t
a
a
bs
t
r
a
c
t
i
on
kn
o
w
n
a
s
de
ndog
r
a
m
[3
].
In
H
i
e
ra
rc
h
i
c
a
l
C
l
us
t
e
r
i
n
g,
t
h
er
e ar
e t
w
o
w
a
y
s
t
o
cl
u
s
t
er
t
h
e d
at
a,
n
a
m
e
l
y
a
g
g
l
o
m
er
at
i
v
e a
n
d
d
i
v
i
s
i
v
e.
A
g
g
l
o
m
er
at
i
v
e cl
u
s
t
er
i
n
g
p
r
o
ces
s
b
as
ed
o
n
t
h
e a
m
o
u
n
t
o
f
d
at
a g
r
o
u
p
ed
i
n
t
o
s
o
m
e
h
i
er
ar
ch
y
,
i
t
w
i
l
l
b
e
cl
u
s
t
er
ed
i
n
t
o
a
h
i
er
ar
ch
i
ca
l
u
n
i
t
.
D
i
v
i
s
i
v
e p
er
f
o
r
m
s
t
h
e
r
ev
er
s
e p
r
o
ces
s
o
f
ag
g
l
o
m
er
at
i
v
e,
t
h
at
i
s
cl
u
s
t
er
o
n
e h
i
er
ar
ch
y
i
n
t
o
s
ev
er
al
h
i
er
ar
ch
i
es
.
A
g
g
l
o
m
er
at
i
v
e H
i
er
ar
ch
i
cal
c
l
u
s
t
er
i
n
g
i
s
a cl
u
s
t
er
i
n
g
al
g
o
r
i
t
h
m
b
as
ed
o
n
t
h
e p
r
o
x
i
m
i
t
y
d
i
s
t
an
c
e
b
et
w
ee
n
t
w
o
i
m
ag
e
s
i
n
t
o
a h
i
er
ar
ch
y
.
T
h
i
s
p
r
o
ces
s
r
ep
eat
s
i
t
s
el
f
u
n
t
i
l
i
t
g
et
s
s
o
m
e h
i
er
ar
ch
y
.
T
h
e h
i
er
ar
c
h
y
w
i
t
h
t
h
e cl
o
s
e
s
t
d
i
s
t
a
n
ce i
s
co
m
b
i
n
ed
i
n
t
o
o
n
e h
i
er
ar
ch
y
.
T
h
e p
r
o
x
i
m
i
t
y
t
o
t
h
e
n
e
w
h
i
er
ar
ch
y
t
h
en
r
ecal
cu
l
at
ed
a
n
d
t
h
e cl
o
s
e
s
t
h
i
er
ar
ch
y
i
s
m
er
g
ed
ag
ai
n
.
T
h
e p
r
o
ces
s
i
s
r
ep
eat
ed
u
n
t
i
l
al
l
t
h
e d
at
a (
o
b
j
ect
)
cl
u
s
t
er
ed
i
n
t
o
o
n
e h
i
er
ar
ch
y
.
C
al
cu
l
a
t
i
n
g
t
h
e s
p
aci
n
g
b
et
w
ee
n
t
w
o
i
m
ag
e
s
u
s
i
n
g
t
h
e
M
an
h
a
t
t
an
D
is
ta
n
c
e
f
o
r
m
u
la
te
d
in
e
q
u
a
tio
n
3
:
i
i
n
i
v
u
d
−
=
∑
=
1
(3
)
w
h
er
e:
d
:
t
h
e d
i
s
t
an
ce b
et
w
ee
n
t
h
e
i
ma
g
e
o
f
u a
nd
v.
n
:
n
u
m
b
er
o
f
v
ar
i
ab
l
es
.
i
u
:
t
he
va
l
ue
o
f
u o
n
i
v
ar
i
ab
l
e.
i
v
:
t
he
va
l
ue
o
f
v o
n
i
v
a
r
ia
b
le
T
h
e d
i
s
t
an
ce b
et
w
een
i
m
a
g
es
i
s
w
r
i
t
t
e
n
i
n
t
o
a
ma
t
r
i
x
cal
l
e
d
d
i
s
t
an
ce
m
at
r
i
x
.
I
n
o
r
d
er
t
o
d
et
er
m
i
n
e
t
h
e d
i
s
t
an
ce b
et
w
ee
n
t
h
e t
w
o
cl
u
s
t
er
s
,
A
g
g
l
o
m
er
at
i
v
e H
i
er
ar
ch
i
cal
cl
u
s
t
er
i
n
g
h
a
s
3
m
et
h
o
d
s
o
f
g
r
o
u
p
i
n
g
d
at
a,
n
am
el
y
:
a.
S
i
n
gl
e
L
i
nka
ge
S
i
n
g
l
e
L
i
n
k
a
g
e cl
as
s
i
f
i
es
d
at
a b
as
ed
o
n
t
h
e cl
o
s
es
t
d
i
s
t
a
n
ce (
Mi
n
)
b
et
w
ee
n
t
h
e
h
i
er
ar
ch
y
.
S
i
n
g
l
e
L
i
nka
ge
ca
n
b
e f
o
r
m
u
l
at
ed
i
n
eq
u
at
i
o
n
4
:
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
20
88
-
8708
I
nt
J
E
l
ec &
C
o
m
p
E
n
g
,
V
o
l.
8
, N
o
.
5
,
O
c
t
obe
r
20
18
:
2
8
12
–
28
17
2814
[
]
)
(
),
(
)
(
vw
d
uw
d
Mi
n
w
uv
d
=
(4
)
w
h
er
e:
u
:
t
h
e i
m
ag
e o
f
u
v
:
t
h
e i
m
ag
e o
f
v
w
:
t
h
e i
m
ag
e o
f
w
(
)
w
uv
d
:
t
h
e d
i
s
t
a
n
ce b
et
w
een
t
h
e h
i
er
ar
ch
y
u
v
a
n
d
w
)
(
uw
d
:
t
h
e d
i
s
t
a
n
ce b
et
w
een
t
h
e h
i
er
ar
ch
y
u
an
d
w
)
(
vw
d
:
t
h
e d
i
s
t
a
n
ce b
et
w
een
t
h
e h
i
er
ar
ch
y
v
an
d
w
b.
C
o
m
p
l
e
t
e
L
i
nka
ge
C
o
m
p
l
et
e
L
i
n
k
ag
e ca
t
eg
o
r
i
zes
d
at
a b
y
t
h
e
f
u
r
t
h
es
t
d
i
s
t
an
ce (
Max
)
o
r
t
h
e
m
a
x
i
m
u
m
d
i
s
t
an
c
e b
et
w
ee
n
h
i
er
ar
ch
i
e
s
.
C
o
m
p
l
et
e
L
i
n
k
a
g
e can
b
e f
o
r
m
u
l
at
ed
i
n
eq
u
at
i
o
n
(
5
)
:
[
]
)
(
),
(
)
(
vw
d
uw
d
Max
w
uv
d
=
(5
)
w
h
er
e:
u
:
t
h
e i
m
ag
e o
f
u
v
:
t
h
e i
m
ag
e o
f
v
w
:
t
h
e i
m
ag
e o
f
w
(
)
w
uv
d
:
t
h
e d
i
s
t
a
n
ce b
et
w
een
t
h
e h
i
er
ar
ch
y
u
v
a
n
d
w
)
(
uw
d
: th
e
d
i
s
t
a
n
ce b
et
w
een
t
h
e h
i
er
ar
ch
y
u
an
d
w
)
(
vw
d
:
t
h
e d
i
s
t
a
n
ce b
et
w
een
t
h
e h
i
er
ar
ch
y
v
an
d
w
c.
A
ve
r
a
ge
L
i
nka
ge
A
v
er
ag
e
L
i
n
k
a
g
e cl
as
s
i
f
i
e
s
d
at
a b
as
ed
o
n
t
h
e av
er
ag
e d
i
s
t
an
ce b
et
w
ee
n
t
h
e
h
i
er
ar
ch
y
.
A
v
er
a
g
e
L
i
nka
ge
c
a
n b
e
f
o
r
m
ul
a
t
e
d
i
n
e
q
ua
t
i
o
n
(6
):
2
)
(
)
(
)
(
vw
d
uw
d
w
uv
d
+
=
(6
)
w
h
er
e:
u
:
t
h
e i
m
ag
e o
f
u
v
:
t
h
e i
m
ag
e o
f
v
w
:
t
h
e i
m
ag
e o
f
w
(
)
w
uv
d
:
t
h
e d
i
s
t
a
n
ce b
et
w
een
t
h
e h
i
er
ar
ch
y
u
v
a
n
d
w
)
(
uw
d
: th
e
d
i
s
t
a
n
ce b
et
w
een
t
h
e h
i
er
ar
ch
y
u
an
d
w
)
(
vw
d
:
t
h
e d
i
s
t
a
n
ce b
et
w
een
t
h
e h
i
er
ar
ch
y
v
an
d
w
2.
3.
E
uc
l
i
di
a
n D
i
s
t
a
nc
e
E
u
cl
i
d
ean
d
i
s
t
a
n
ce i
s
an
al
g
o
r
i
t
h
m
t
o
cal
c
u
l
at
e
t
h
e
d
i
f
f
er
en
c
e o
r
m
i
n
i
m
u
m
d
i
s
t
an
ce
b
et
w
e
en
t
h
e
t
e
s
t
i
m
a
ge
a
nd
t
r
a
i
ni
n
g i
m
a
ge
.
A
n
O
b
j
ect
th
a
t
w
i
ll b
e
u
s
e
d
a
s
a
r
e
c
o
g
n
itio
n
o
b
j
e
c
t is
th
e
o
n
e
w
h
ic
h
h
a
s
th
e
m
i
n
i
m
u
m
d
i
s
t
a
n
ce.
T
o
cal
cu
l
at
e t
h
e
m
i
n
i
m
u
m
d
i
s
t
an
ce b
et
w
ee
n
t
h
e
v
ect
o
r
v
al
u
e o
f
t
h
e
t
es
t
i
m
a
g
e an
d
t
h
e
t
r
ai
n
i
n
g
i
m
ag
e
p
r
es
en
t
i
n
t
h
e
d
at
ab
as
e,
t
h
e E
u
cl
i
d
ea
n
d
i
s
t
a
n
ce al
g
o
r
i
t
h
m
eq
u
at
i
o
n
ca
n
b
e ex
p
r
es
s
ed
b
y
t
h
e
fo
r
m
u
l
a
:
(
)
B
A
b
a
B
A
J
p
e
e
e
−
=
−
=
∑
=
1
2
)
,
(
(7
)
wh
i
c
h
:
)
,
(
B
A
J
:
t
h
e d
i
s
t
an
ce b
et
w
ee
n
i
m
ag
er
y
A
an
d
B
A
:
t
h
e v
al
u
e o
f
t
h
e i
m
ag
e
v
ect
o
r
p
r
es
en
t
i
n
t
h
e
d
at
ab
as
e
B
:
va
l
ue
o
f
t
h
e t
e
s
t
i
m
ag
e
v
ect
o
r
P
:
n
u
m
b
er
o
f
v
ar
i
ab
l
es
e
a
:
va
l
ue
a
o
n t
he
e
-
v
a
r
ia
b
le
e
b
:
va
l
ue
b
o
n t
he
e
-
v
a
r
ia
b
le
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t
J
E
l
e
c
&
C
o
m
p
E
n
g
I
S
S
N
:
2088
-
8708
R
ea
l
-
tim
e
M
u
lti
-
o
bj
e
c
t
F
ac
e
R
e
c
ogni
t
i
on U
s
i
ng C
ont
e
nt
B
as
e
d I
m
age
.
...
(
M
uham
m
ad F
ac
hr
ur
r
oz
i
)
2815
3.
M
E
T
HOD
OL
OG
Y
F
i
g
ur
e
1
is
a
b
l
o
c
k
d
ia
g
r
a
m
o
f
a
m
u
lti
-
o
b
j
ect
f
ace r
eco
g
n
i
t
i
o
n
s
y
s
t
e
m
b
as
e
d
o
n i
m
a
ge
r
e
t
r
i
e
va
l
,
w
h
er
e t
h
e p
at
h
o
f
t
h
e
s
y
s
t
e
m
i
s
d
i
v
i
d
ed
i
n
t
o
2
s
t
a
g
e
s
n
a
m
el
y
t
h
e
p
h
as
e o
f
d
at
a cap
t
u
r
e p
h
as
e o
f
f
ace
r
e
c
o
gni
t
i
o
n.
F
i
g
u
r
e 1
.
G
en
er
al
s
y
s
te
m
d
ia
g
r
a
m
I
n
g
e
n
er
al
,
t
h
e s
t
ep
s
o
f
d
at
a r
et
r
i
ev
al
i
n
t
h
i
s
r
es
ear
ch
i
s
as
f
o
l
l
o
w
s
:
a.
C
o
lle
c
t
f
ace i
m
a
g
e.
b.
T
h
e f
eat
u
r
e e
x
t
r
act
i
o
n
p
r
o
ces
s
u
s
i
n
g
L
B
P
m
et
h
o
d
t
o
g
et
t
h
e c
h
ar
act
er
i
s
t
i
c o
f
f
ac
e i
m
a
g
e t
h
e
n
t
r
an
s
f
o
r
m
ed
i
n
t
o
v
ect
o
r
f
eat
u
r
e f
o
r
m
w
h
i
ch
w
i
l
l
b
e s
t
o
r
ed
i
n
t
h
e d
at
ab
as
e.
c.
T
he
n d
o
t
he
c
l
us
t
e
r
i
n
g p
r
o
c
e
s
s
us
i
n
g t
he
A
H
C
m
e
t
ho
d
o
n t
he
ve
c
t
o
r
o
f
t
he
f
a
ce i
m
ag
e i
n
t
h
e d
at
ab
as
e
.
C
l
u
s
t
er
i
n
g
r
es
u
l
t
s
t
h
at
h
av
e b
een
o
b
t
ai
n
ed
l
at
er
u
s
ed
as
a co
m
p
ar
at
o
r
v
al
u
e
of
c
a
lc
u
la
tin
g
th
e
d
is
ta
n
c
e
f
o
r
f
ace r
eco
g
n
i
t
i
o
n
.
I
n
t
h
e
f
ace r
eco
g
n
i
t
i
o
n
p
h
a
s
e t
h
e p
r
o
ces
s
o
f
t
h
e
s
y
s
t
e
m
i
s
as
f
o
l
l
o
w
s
:
a.
O
p
en
w
eb
ca
m
t
o
d
et
ect
f
aces
.
T
h
e
p
r
o
ces
s
o
f
f
ace d
et
ect
i
o
n
an
d
r
eco
g
n
i
t
i
o
n
i
s
d
o
n
e i
n
r
eal
-
t
i
me
.
b.
T
h
e d
et
ect
ed
f
ace t
h
en
cap
t
u
r
e
d
.
c.
T
h
e f
eat
u
r
e e
x
t
r
act
i
o
n
p
r
o
ces
s
i
s
d
o
n
e
u
s
i
n
g
L
o
cal
B
i
n
ar
y
P
at
t
er
n
m
et
h
o
d
t
o
g
et
t
h
e
f
eat
u
r
e
o
f
t
h
e
f
ace
i
m
a
g
e t
h
e
n
t
r
an
s
f
o
r
m
ed
i
n
t
o
t
h
e v
ect
o
r
f
eat
u
r
e f
o
r
m
.
T
h
en
,
d
o
t
h
e p
r
o
ces
s
o
f
r
eco
g
n
i
t
i
o
n
b
y
cal
c
u
l
at
i
n
g
t
h
e d
i
s
t
a
n
ce b
et
w
ee
n
t
h
e n
e
w
f
ace i
m
a
g
e f
eat
u
r
es
an
d
f
eat
u
r
es
o
f
t
h
e e
x
i
s
t
i
n
g
o
n
t
h
e d
at
ab
as
e b
y
u
s
i
n
g
E
u
cl
i
d
i
an
d
i
s
t
an
ce
w
h
i
c
h
t
h
e
n
m
at
ch
ed
w
i
t
h
t
h
e
c
l
us
t
e
r
i
n
g r
e
s
ul
t
s
.
4.
IM
P
LEM
EN
TA
TI
O
N
A
N
D
R
ES
U
LT
T
h
i
s
s
t
u
d
y
u
s
es
d
at
as
et
as
m
an
y
as
5
0
i
m
ag
e
s
f
r
o
m
5
p
eo
p
l
e,
each
t
a
k
en
1
0
i
m
ag
e
s
.
I
m
ag
e
s
ar
e
t
ak
en
f
r
o
m
d
i
f
f
er
en
t
s
i
d
es
,
b
u
t
w
i
t
h
t
h
e s
a
m
e b
ack
g
r
o
u
n
d
.
F
ace i
m
ag
e d
et
er
m
i
n
ed
s
t
i
l
l
v
i
s
i
b
l
e b
o
t
h
ey
e
s
,
n
o
s
e
a
nd
m
o
ut
h.
T
he
i
m
a
ge
us
e
d
a
s
t
r
a
i
ni
ng d
a
t
a
ha
s
d
i
m
e
ns
i
o
n
s
of
150x
150
pi
x
e
l
s
.
E
x
a
m
pl
e
s
of
f
a
c
e
i
m
a
g
e
da
t
a
can
b
e s
een
i
n
F
i
gur
e
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
20
88
-
8708
I
nt
J
E
l
ec &
C
o
m
p
E
n
g
,
V
o
l.
8
, N
o
.
5
,
O
c
t
obe
r
20
18
:
2
8
12
–
28
17
2816
F
i
gu
r
e
2.
E
x
a
m
pl
e
o
f
i
m
a
g
e f
a
ce d
at
a
F
ace
i
m
a
g
e
r
eco
g
n
i
t
i
o
n
d
o
n
e
b
y
co
m
p
ar
i
n
g
t
h
e
d
i
s
t
a
n
ce
f
r
o
m
t
h
e
t
es
t
v
ect
o
r
i
m
a
g
e
w
i
t
h
t
r
ai
n
i
n
g
i
m
a
g
e
f
eat
u
r
e v
a
l
u
e
u
s
i
n
g
E
u
cl
i
d
ean
.
T
h
e s
m
al
l
e
s
t
d
i
s
t
an
ce
o
f
v
ect
o
r
f
eat
u
r
e
b
et
w
een
t
h
e
t
es
t
i
m
a
g
e
an
d
t
h
e
t
r
a
i
ni
ng i
m
a
ge
w
i
l
l
d
e
t
e
r
m
i
ne
t
he
o
ut
c
o
m
e
o
f
t
he
r
e
c
o
gni
t
i
o
n.
F
i
g
ur
e
3
is
a
m
u
lt
i
-
o
b
j
ect
f
ace i
m
a
g
e
r
e
c
o
gni
t
i
o
n f
o
r
m
.
F
i
gu
r
e
3.
R
e
s
ul
t
of
f
ace i
m
a
g
e
r
eco
g
n
i
t
i
o
n
M
eas
u
r
e
m
en
t
accu
r
ac
y
r
at
e i
s
o
b
t
ai
n
ed
b
y
co
m
p
ar
i
n
g
t
h
e
v
al
u
e o
f
t
h
e
n
u
m
b
er
o
f
o
b
j
ect
s
t
o
b
e
i
d
en
t
i
f
i
ed
an
d
t
h
e n
u
m
b
er
o
f
o
b
j
ect
s
co
n
t
ai
n
ed
i
n
t
h
e t
es
t
i
m
ag
es
.
T
h
e l
ev
el
o
f
accu
r
ac
y
o
f
f
ace r
eco
g
n
i
t
i
o
n
i
n
r
eal
t
i
m
e i
s
s
m
al
l
er
t
h
an
f
aci
al
r
eco
g
n
i
t
i
o
n
i
n
n
o
n
r
eal
t
i
m
e.
O
n
t
h
e
i
nt
r
o
d
uc
t
i
o
n i
n o
n r
e
a
l
t
i
m
e
w
i
t
h t
he
s
a
m
e
da
t
a
a
n
d
m
e
t
h
ods
obt
a
i
n
e
d a
n
a
c
c
u
r
a
c
y
of
65.
32%
[
4]
.
T
h
e
r
es
ear
ch
“
A
F
ace
R
eco
g
n
i
t
i
o
n
S
y
s
t
e
m
B
as
ed
o
n
E
i
g
e
n
f
ace
s
Met
h
o
d
”,
t
h
e
ei
g
e
n
f
ac
e
w
i
t
h
t
h
e
s
m
al
l
es
t
E
u
cl
i
d
i
a
n
d
i
s
t
a
n
ce i
s
t
h
e o
n
e t
h
e p
er
s
o
n
r
es
e
m
b
l
es
t
h
e
m
o
s
t
.
S
i
m
u
l
a
t
i
o
n
s
ha
ve
b
e
e
n d
o
ne
us
i
n
g t
he
M
a
t
l
a
b pr
og
r
a
m
.
T
h
e
s
u
c
c
e
s
s
r
a
t
e
f
or
t
h
e
l
a
r
g
e
da
t
a
ba
s
e
us
e
d i
s
f
oun
d t
o be
94.
74%
[
7]
.
R
eco
g
n
i
zi
n
g
f
ace o
n
A
RM
p
r
o
c
e
ss
o
r
u
s
i
n
g
K
-
N
ea
r
es
t
N
ei
g
h
b
o
r
(
K
N
N
)
a
lg
o
r
ith
m
.
T
h
is
r
e
s
e
a
r
c
h
s
o
u
g
h
t b
e
s
t k
-
v
al
u
e t
o
cr
eat
e
p
r
o
p
er
f
ace r
eco
g
n
i
t
i
o
n
w
i
t
h
l
o
w
-
p
o
w
er
p
r
o
ces
s
o
r
.
T
h
e p
r
o
p
o
s
ed
al
g
o
r
i
t
h
m
w
a
s
t
e
s
t
ed
o
n
t
h
r
ee d
at
as
et
s
t
h
at
w
er
e O
l
i
v
et
t
i
R
e
s
ear
ch
L
ab
o
r
at
o
r
y
(
O
R
L
)
,
Y
al
e
f
ace an
d
M
U
C
T
.
O
p
en
C
V
w
a
s
c
h
o
s
en
as
m
a
i
n
c
or
e
i
m
a
ge
p
r
o
c
e
s
s
in
g
lib
r
a
r
y
,
d
u
e
to
its
h
ig
h
-
s
pe
e
d.
P
r
opos
e
d
a
l
g
or
i
t
hm
w
a
s
i
m
pl
e
m
e
n
t
e
d on
A
R
M
11 7
00M
H
z
.
10
-
f
ol
d
c
r
o
s
s
-
v
al
i
d
at
i
o
n
s
h
o
w
ed
t
h
at
K
N
N
f
ace r
eco
g
n
i
t
i
o
n
d
et
e
ct
ed
9
1
.
5
%
f
ace
w
i
t
h
k
=1
.
O
v
er
al
l
e
x
p
er
i
m
e
n
t
s
h
o
w
ed
t
h
at
p
r
o
p
o
s
ed
al
g
o
r
i
t
h
m
d
et
ect
ed
f
ace
on
2.
66 s
on
A
R
M
pr
oc
e
s
s
or
[
10]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t
J
E
l
e
c
&
C
o
m
p
E
n
g
I
S
S
N
:
2088
-
8708
R
ea
l
-
tim
e
M
u
lti
-
o
bj
e
c
t
F
ac
e
R
e
c
ogni
t
i
on U
s
i
ng C
ont
e
nt
B
as
e
d I
m
age
.
...
(
M
uham
m
ad F
ac
hr
ur
r
oz
i
)
2817
T
a
b
le
1
.
T
e
s
tin
g
o
f
R
eal
-
t
i
me
F
ace R
eco
g
n
i
t
i
o
n
N
o.
T
e
s
t im
a
g
e
R
ec
o
g
n
i
zed
o
b
j
ec
t
A
ccu
r
acy
(%
)
T
i
m
e ex
ec
u
t
i
o
n
(
s
e
c
o
n
d
)
1
.
2
o
bj
e
c
t
s
o
n
i
m
a
ge
-
1
1
5
0
0
,
0
75
0
2
.
2
o
bj
e
c
t
s
o
n
i
m
a
ge
-
2
1
5
0
0
,
0
74
0
3
.
2
o
bj
e
c
t
s
o
n
i
m
a
ge
-
3
2
1
0
0
0
,
0
73
0
4
.
2
o
bj
e
c
t
s
o
n
i
m
a
ge
-
4
1
5
0
0
,
0
72
0
5
.
2
o
bj
e
c
t
s
o
n
i
m
a
ge
-
5
2
1
0
0
0
,
0
71
0
6
.
3
o
bj
e
c
t
s
o
n
i
m
a
ge
-
1
1
3
3
.
3
0
,
0
73
0
7
.
3
o
bj
e
c
t
s
o
n
i
m
a
ge
-
2
2
6
6
.
6
0
,
0
73
0
8
.
3
o
bj
e
c
t
s
o
n
i
m
a
ge
-
3
2
6
6
.
6
0
,
0
72
0
9
.
3
o
bj
e
c
t
s
o
n
i
m
a
ge
-
4
2
6
6
.
6
0
,
0
72
0
1
0
.
3
o
bj
e
c
t
s
o
n
i
m
a
ge
-
5
1
3
3
.
3
0
,
0
71
0
5.
CO
NCL
U
S
I
O
N
M
u
lti
-
o
b
j
ect
f
ace r
eco
g
n
i
t
i
o
n
s
y
s
t
e
m
can
r
eco
g
n
i
ze s
i
n
g
l
e o
r
m
u
l
t
i
o
b
j
ect
i
n
r
eal
-
ti
m
e
w
it
h
a
n
accu
r
ac
y
o
f
6
1
.
6
4
%
.
T
h
e f
eat
u
r
e ex
t
r
act
i
o
n
p
r
o
ces
s
o
n
t
h
e
i
n
p
u
t
i
m
a
g
e p
l
a
y
s
a
n
i
m
p
o
r
t
an
t
r
o
l
e i
n
d
et
er
m
i
n
i
n
g
t
he
s
u
cce
s
s
r
at
e o
f
f
ace i
m
a
g
e r
eco
g
n
i
t
i
o
n
.
T
h
e r
eco
g
n
i
zab
l
e t
es
t
i
n
g
p
r
o
ces
s
h
as
t
h
e
s
a
m
e l
i
g
h
t
i
n
g
,
d
i
s
t
a
n
ce
a
nd
o
t
he
r
e
f
f
e
c
t
s
d
ur
i
n
g t
r
a
i
ni
ng.
R
EF
ER
EN
C
ES
[
1]
K
.
C
h
en
an
d
L
.
J
.
Z
h
ao
,
“R
o
b
u
s
t
R
eal
t
i
m
e F
ace R
eco
g
n
i
t
i
o
n
a
nd
T
r
a
c
ki
n
g S
ys
t
e
m
,
”
vo
l
/
i
ssu
e
:
9
(
2
)
,
p
p
. 8
2
-
8
8
,
200
9.
[
2]
C
. L
i
n
,
e
t a
l.
,
“
G
a
bor
F
i
l
t
e
r
s
a
nd
F
e
a
t
ur
e
F
us
i
on,
”
v
ol
/
i
ssu
e
:
11
(
10
)
,
p
p.
59
86
-
59
94,
2
013
.
[
3]
D
.
H
.
Z
.
a
nd
P
.
D
.
D
.
F
.
F
u
hu
i
L
ong
,
“
F
unda
m
e
nt
a
l
s
of
C
ont
e
nt
-
B
as
ed
I
m
a
g
e
R
et
r
i
ev
al
,
”
M
u
ltim
e
d
.
I
n
f.
Re
tr
.
M
ana
g.
T
e
c
h
nol
.
F
un
dam
.,
p
p
. 1
-
26,
20
03
.
[
4]
M.
Fa
c
hr
ur
r
oz
i
,
“
M
ul
t
i
-
O
b
j
ect
F
ace R
eco
g
n
i
t
i
o
n
U
s
i
n
g
C
o
n
t
en
t
B
as
ed
I
m
a
g
e
R
et
r
i
ev
al
(
C
B
I
R
)
,
”
vo
l
.
x
,
pp.
193
-
197
,
2
01
7.
[
5]
T
.
A
hone
n,
e
t a
l.
,
“F
ace
D
es
cr
i
p
t
i
o
n
w
i
t
h
L
o
cal
B
i
n
ar
y
P
at
t
er
n
s
:
A
p
p
l
i
cat
i
o
n
t
o
F
ace
R
eco
g
n
i
t
i
o
n
,
”
v
o
l
/
i
ssu
e
:
28
(1
2
)
,
p
p.
20
37
-
20
41,
2
00
6.
[
6]
V.
S
. V
.
S
. M
u
r
t
h
y
,
e
t a
l.
,
“C
o
n
t
en
t
B
as
ed
I
m
a
g
e R
et
r
i
e
v
al
u
s
i
n
g
H
i
er
ar
ch
i
cal
an
d
K
-
M
e
a
n
s
C
lu
s
te
r
in
g
T
ech
n
i
q
u
es
,
”
I
nt
.
J
.
E
ng.
Sc
i
.
T
e
c
hno
l
.
,
vol
/
i
ssu
e
:
2
(
3
)
,
pp
.
2
09
-
21
2,
2
01
0.
[
7]
M
. ü
g
. Ç
a
r
ı
k
ç
ı
a
n
d
F
. Ö
z
e
n
,
“A
F
ace R
e
co
g
n
i
t
i
o
n
S
y
s
t
e
m
B
as
ed
o
n
E
i
g
en
f
a
ces
M
et
h
o
d
,
”
P
r
oc
e
di
a T
e
c
hnol
.,
v
o
l
.
1
,
pp.
11
8
-
123
,
2
01
2.
[
8]
A
.
K
at
ar
e,
e
t a
l.
,
“C
o
n
t
en
t
B
as
ed
I
m
a
g
e R
et
r
i
ev
al
S
y
s
t
e
m
f
o
r
M
u
l
t
i
O
b
j
ect
I
m
ag
es
u
s
i
n
g
C
o
m
b
i
n
ed
F
eat
u
r
es
,
”
200
7.
[
9]
N
.
A
la
jla
n
,
e
t a
l.
,
“
M
u
lti
-
o
b
j
ect
i
m
ag
e
r
e
t
r
i
e
va
l
ba
s
e
d on s
ha
pe
a
nd t
o
po
l
og
y
,
”
Si
gnal
P
r
oc
e
s
s
.
I
m
age
C
om
m
un
.
,
vo
l
/
i
ssu
e
:
21
(
10
)
,
pp.
9
04
-
91
8,
20
06.
[
1
0]
E
.
S
e
t
i
a
w
a
n
a
nd A
.
M
ut
t
a
qi
n
,
“
I
m
pl
e
m
e
nt
a
t
i
on of
K
-
N
e
a
r
e
s
t
N
e
i
g
hbor
s
F
a
c
e
R
e
c
og
ni
t
i
on on L
ow
-
pow
e
r
P
r
o
ces
s
o
r
,
” v
o
l
/
i
ssu
e
:
13
(
3
)
,
2
01
5.
[
1
1]
M
.
E
. W
i
b
o
w
o
,
e
t a
l.
,
“
I
m
pr
ov
e
d F
a
c
e
R
e
c
og
ni
t
i
on
a
c
r
os
s
P
os
e
s
us
i
ng
F
us
i
o
n
of
P
r
o
ba
bi
l
i
s
t
i
c
L
a
t
e
nt
V
a
r
i
a
bl
e
M
ode
l
s
,
”
v
ol
/
i
ssu
e
:
15
(
4
)
,
p
p.
19
76
-
19
86,
20
17
.
Evaluation Warning : The document was created with Spire.PDF for Python.