I
nte
rna
t
io
na
l J
o
urna
l o
f
Rec
o
nfig
ura
ble a
nd
E
m
bedd
e
d Sy
s
t
em
s
(
I
J
RE
S)
Vo
l.
10
,
No
.
1
,
Ma
r
ch
2
0
2
1
,
p
p
.
47
~
55
I
SS
N
: 2
0
8
9
-
4
8
6
4
,
DOI
: 1
0
.
1
1
5
9
1
/ij
r
es
.
v
10
.i
1
.
p
p
47
-
55
47
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
r
es.ia
esco
r
e.
co
m
Desig
n and
devel
o
pment of a
pa
ra
llelized alg
o
rithm
for face
recog
nition in mo
bile cloud
env
iro
nment
K
.
N.
B
ha
t
t
,
Sa
nk
et
Dess
a
i,
V.
S.
Yer
ra
g
ud
i
De
p
a
rtem
e
n
t
o
f
Co
m
p
u
ter S
c
ien
c
e
a
n
d
E
n
g
i
n
e
e
rin
g
,
Ra
m
a
iah
Un
iv
e
rsity
o
f
A
p
p
l
ied
S
c
ien
c
e
s,
Be
n
g
a
lu
ru
,
In
d
ia
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
May
24
,
2
0
20
R
ev
is
ed
J
u
l
2
8
,
2
0
20
Acc
ep
ted
J
an
1
0
,
2
0
21
F
a
c
e
re
c
o
g
n
it
io
n
is
t
h
e
b
i
o
m
e
tri
c
a
p
p
li
c
a
ti
o
n
t
o
re
c
o
g
n
ise
th
e
id
e
n
ti
ty
.
F
a
c
e
re
c
o
g
n
it
i
o
n
a
p
p
li
c
a
ti
o
n
h
o
ld
s
a
se
t
o
f
ima
g
e
s
wh
ich
a
re
c
a
ll
e
d
d
a
tab
a
se
s
sto
re
d
b
y
t
h
e
u
se
r
a
t
c
lo
u
d
d
a
ta
b
a
se
.
Clo
u
d
c
o
m
p
u
ti
n
g
e
n
v
iro
n
m
e
n
t,
d
a
tab
a
se
c
a
n
b
e
sto
re
d
i
n
th
e
c
lo
u
d
e
n
v
ir
o
n
m
e
n
t
t
o
a
c
h
iev
e
h
u
g
e
d
a
ta
st
o
ra
g
e
a
re
a
.
Th
e
p
r
o
b
lem
wit
h
t
h
e
se
d
a
ta
st
o
ra
g
e
s
a
re
th
a
t
b
e
c
a
u
se
o
f
t
h
a
t
h
u
g
e
siz
e
p
ro
c
e
ss
in
g
o
n
t
h
is
st
o
ra
g
e
ta
k
e
s
t
o
o
m
u
c
h
o
f
c
o
m
p
il
i
n
g
ti
m
e
.
T
h
is p
a
p
e
r
a
ims
to
d
e
v
e
lo
p
fa
c
e
re
c
o
g
n
it
io
n
in
m
o
b
il
e
c
lo
u
d
e
n
v
i
ro
n
m
e
n
t
b
y
e
x
p
l
o
it
in
g
d
a
ta
o
r
tas
k
p
a
ra
ll
e
li
sm
in
e
x
isti
n
g
f
a
c
e
re
c
o
g
n
it
io
n
a
l
g
o
ri
th
m
s.
T
o
d
e
sig
n
a
n
d
d
e
v
e
lo
p
p
a
ra
ll
e
l
P
CA
b
a
se
d
fa
c
e
re
c
o
g
n
it
i
o
n
a
lg
o
rit
h
m
.
Th
e
p
a
ra
ll
e
l
P
CA
fa
c
e
re
c
o
g
n
it
io
n
a
lg
o
r
it
h
m
h
a
s
b
e
e
n
d
e
p
l
o
y
e
d
in
t
h
e
c
lo
u
d
se
rv
e
r
fo
r
p
e
rfo
rm
in
g
P
CA
b
y
re
q
u
e
st
o
f
u
s
e
r.
It
m
a
tch
e
s
th
e
ima
g
e
o
n
t
h
e
c
l
o
u
d
se
rv
e
r
a
n
d
g
iv
e
s
re
sp
o
n
se
b
a
c
k
to
th
e
u
se
r
in
th
e
fe
we
r
a
m
o
u
n
ts
o
f
ti
m
e
a
n
d
wit
h
re
d
u
c
e
d
late
n
c
y
.
Th
e
d
e
v
e
l
o
p
e
d
P
a
ra
ll
e
l
P
CA
fa
c
e
re
c
o
g
n
it
i
o
n
a
lg
o
rit
h
m
h
a
s
m
in
imiz
e
d
t
h
e
o
v
e
ra
ll
re
s
p
o
n
se
ti
m
e
fo
r
t
h
e
fa
c
e
re
c
o
g
n
i
ti
o
n
a
l
g
o
r
it
h
m
.
Th
e
p
e
rfo
rm
a
n
c
e
o
f
th
e
d
e
v
e
lo
p
e
d
sy
ste
m
is
tes
ted
a
n
d
a
n
a
ly
se
d
o
n
re
a
l
fa
c
e
ima
g
e
s.
To
a
n
a
ly
se
th
e
d
e
v
e
l
o
p
e
d
sy
ste
m
,
a
c
e
n
tralize
d
a
n
d
d
istr
ib
u
ted
b
a
se
d
se
rv
e
r
m
e
th
o
d
s
a
re
d
e
v
e
lo
p
e
d
a
n
d
c
o
m
p
a
ris
o
n
is
b
e
i
n
g
c
a
rried
o
u
t.
Th
e
c
o
n
c
lu
si
o
n
d
ra
wn
th
a
t
th
e
d
istri
b
u
ted
se
rv
e
r
imp
r
o
v
e
s
th
e
e
fficie
n
c
y
a
s
we
ll
a
s
th
e
c
o
m
p
u
ti
n
g
p
o
we
r
a
s
c
o
m
p
a
re
d
with
c
e
n
tralize
d
se
rv
e
r
sy
ste
m
.
Th
e
c
o
m
p
a
riso
n
o
f
c
e
n
tralize
d
a
n
d
d
istri
b
u
ted
b
a
se
d
se
rv
e
rs
is
c
a
rried
o
u
t
b
y
o
b
se
rv
i
n
g
th
e
ti
m
e
tak
e
n
w
h
il
e
v
a
ry
in
g
th
e
n
u
m
b
e
r
o
f
ima
g
e
s
in
t
h
e
train
i
n
g
d
a
tas
e
t.
K
ey
w
o
r
d
s
:
C
lo
u
d
co
m
p
u
tin
g
Face
r
ec
o
g
n
itio
n
Mo
b
ile
clo
u
d
e
n
v
ir
o
n
m
en
t
Pallelized
PC
A
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
:
San
k
et
Dess
ai
Dep
ar
tem
en
t o
f
C
o
m
p
u
ter
Sci
en
ce
an
d
E
n
g
in
ee
r
in
g
R
am
aiah
Un
iv
er
s
ity
o
f
Ap
p
lied
Scien
ce
s
#470
-
P,
Peen
y
a
I
n
d
u
s
tr
ial
Ar
e
a,
Peen
y
a
Seco
n
d
Stag
e,
Peen
y
a
B
en
g
alu
r
u
,
Kar
n
atak
a
5
6
0
0
5
4
,
I
n
d
ia
E
m
ail: san
k
etd
ess
ai0
8
0
8
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
C
l
o
u
d
c
o
m
p
u
t
i
n
g
i
s
a
n
a
p
p
r
o
a
c
h
o
f
u
s
i
n
g
r
e
m
o
t
e
s
e
r
v
e
r
s
i
n
o
r
d
e
r
t
o
p
r
o
v
i
d
e
h
u
g
e
d
a
t
a
s
t
o
r
a
g
e
,
c
o
m
p
u
t
a
t
i
o
n
a
n
d
o
t
h
e
r
p
r
o
c
e
s
s
i
n
g
s
e
r
v
i
c
e
.
I
n
c
l
o
u
d
c
o
m
p
u
t
i
n
g
s
e
r
v
e
r
,
i
t
i
s
p
o
s
s
i
b
l
e
t
o
s
a
v
e
l
a
r
g
e
n
u
m
b
e
r
o
f
d
a
t
a
a
n
d
w
h
e
n
u
s
e
r
r
e
q
u
e
s
t
f
o
r
a
p
a
r
t
i
c
u
l
a
r
d
a
t
a
s
e
r
v
i
c
e
t
h
e
n
i
t
w
i
l
l
g
i
v
e
t
h
e
r
e
s
p
o
n
s
e
b
a
c
k
t
o
t
h
e
u
s
e
r
i
n
t
h
e
f
r
a
c
t
i
o
n
o
f
t
h
e
t
i
m
e
.
T
h
i
s
s
e
r
v
i
c
e
o
f
c
l
o
u
d
c
o
m
p
u
t
i
n
g
i
s
r
e
f
e
r
r
e
d
a
s
t
h
e
o
n
-
d
e
m
a
n
d
e
n
v
i
r
o
n
m
e
n
t
o
f
t
h
e
c
l
o
u
d
.
O
n
e
o
f
t
h
e
b
i
g
c
h
a
l
l
e
n
g
e
s
i
n
u
s
i
n
g
c
l
o
u
d
s
e
r
v
e
r
s
i
s
t
h
e
l
a
t
e
n
c
y
t
i
m
e
t
a
k
e
n
i
n
o
r
d
e
r
t
o
g
i
v
e
r
e
s
p
o
n
s
e
b
a
c
k
t
o
t
h
e
u
s
e
r
f
r
o
m
t
h
e
s
e
r
v
e
r
.
I
n
o
r
d
e
r
t
o
o
v
e
r
c
o
m
e
t
h
i
s
s
i
t
u
a
t
i
o
n
,
t
h
e
c
l
o
u
d
e
n
v
i
r
o
n
m
e
n
t
i
s
d
i
v
i
d
e
d
i
n
t
o
s
e
v
e
r
a
l
s
m
a
l
l
s
c
a
l
e
d
d
i
s
t
r
i
b
u
t
e
d
s
e
r
v
e
r
p
r
o
c
e
s
s
e
s
w
h
i
c
h
a
r
e
c
a
l
l
e
d
c
l
o
u
d
l
e
t
.
S
o
,
u
s
e
r
r
e
q
u
e
s
t
s
a
r
e
s
e
n
t
a
n
d
r
e
c
e
i
v
e
d
t
h
r
o
u
g
h
c
l
o
u
d
l
e
t
s
i
n
o
r
d
e
r
t
o
e
n
a
b
l
e
t
h
e
u
s
e
o
f
c
l
o
u
d
s
e
r
v
e
r
.
S
o
,
t
h
r
o
u
g
h
t
h
e
h
e
l
p
o
f
t
h
e
c
l
o
u
d
l
e
t
p
r
o
c
e
s
s
,
i
t
d
i
s
t
r
i
b
u
t
e
s
t
h
e
c
o
m
p
u
t
i
n
g
l
o
a
d
o
f
t
h
e
c
l
o
u
d
s
e
r
v
e
r
w
h
i
c
h
a
c
h
i
e
v
e
s
t
h
e
m
i
n
i
m
a
l
r
e
s
p
o
n
s
e
t
i
m
e
a
n
d
r
e
d
u
c
e
d
l
a
t
e
n
c
y
t
i
m
e
w
h
e
n
r
e
q
u
e
s
t
i
s
b
e
e
n
s
e
n
d
f
r
o
m
r
e
s
o
u
r
c
e
s
t
o
m
a
t
c
h
t
h
e
p
r
o
c
e
s
s
w
i
t
h
m
u
l
t
i
p
l
e
s
e
r
v
e
r
i
n
t
h
e
c
l
o
u
d
[
1
-
3
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
20
89
-
4
8
6
4
I
n
t J Reco
n
f
ig
u
r
a
b
le
&
E
m
b
ed
d
ed
Sy
s
t
,
Vo
l.
10
,
No
.
1
,
Ma
r
c
h
202
1
:
47
–
55
48
Mo
b
ile
clo
u
d
co
m
p
u
tin
g
is
s
i
m
p
lest
way
in
th
e
cl
o
u
d
s
er
v
e
r
.
T
h
is
r
ef
er
s
to
an
in
f
r
astru
ct
u
r
e
wh
e
r
e
b
o
th
th
e
d
ata
s
to
r
ag
e
a
n
d
p
r
o
c
ess
in
g
will
h
ap
p
en
o
u
ts
id
e
o
f
t
h
e
m
o
b
ile
d
ev
ice
th
at
is
th
e
cl
o
u
d
s
er
v
e
r
.
Mo
b
ile
clo
u
d
ap
p
licatio
n
s
m
o
v
e
th
e
co
m
p
u
tin
g
p
o
wer
an
d
d
ata
s
to
r
ag
e
awa
y
f
r
o
m
th
e
m
o
b
il
e
d
ev
ices
an
d
i
n
to
p
o
wer
f
u
l
an
d
ce
n
tr
alize
d
c
o
m
p
u
tin
g
p
latf
o
r
m
s
lo
ca
ted
in
clo
u
d
s
,
wh
ic
h
ar
e
ac
ce
s
s
ib
le
o
v
er
th
e
wir
eless
co
n
n
ec
tio
n
b
ased
o
n
th
e
n
ativ
e
clien
t.
I
t
ca
n
h
elp
to
o
v
er
co
m
e
th
e
m
o
b
ile
co
m
p
u
tin
g
ch
a
llen
g
es
lik
e
b
atter
y
life
,
s
to
r
ag
e
a
n
d
b
an
d
wid
th
.
T
h
is
also
im
p
r
o
v
es
th
e
r
eliab
ilit
y
,
s
ca
lab
ilit
y
,
d
ata
s
to
r
ag
e
an
d
p
r
o
ce
s
s
in
g
p
o
we
r
as m
o
b
ile.
Face
r
ec
o
g
n
itio
n
m
et
h
o
d
ca
n
b
e
u
s
ed
as
o
n
e
o
f
th
e
m
eth
o
d
in
o
r
d
er
t
o
p
r
o
v
id
e
s
ec
u
r
e
d
ata
ac
ce
s
s
f
r
o
m
th
e
clo
u
d
.
Face
r
ec
o
g
n
iti
o
n
is
a
p
r
o
ce
s
s
th
at
a
u
to
m
atica
lly
id
en
tify
an
in
d
iv
id
u
al
f
r
o
m
ca
p
tu
r
ed
im
ag
e
o
r
v
id
eo
,
f
o
r
d
if
f
er
e
n
t
ap
p
licatio
n
lik
e
air
p
o
r
t,
Pas
s
p
o
r
t
o
f
f
ice,
an
d
Go
v
er
n
m
en
t
o
f
f
ice.
I
n
th
e
f
ac
e
r
ec
o
g
n
itio
n
p
r
o
ce
s
s
,
it
s
to
r
es
th
e
d
ata
in
t
h
e
d
atab
ase
i
n
o
r
d
er
to
v
er
if
y
th
e
h
u
m
an
f
o
r
p
r
o
v
id
i
n
g
t
h
e
au
th
en
ticatio
n
.
T
h
e
d
atab
ase
ca
n
be
s
to
r
ed
eith
e
r
i
n
s
id
e
th
e
l
o
ca
l
s
er
v
er
o
r
o
n
cl
o
u
d
s
er
v
er
[
4
]
.
Fo
r
ex
am
p
le,
i
f
clo
u
d
d
atab
ase
is
u
s
ed
in
o
r
d
e
r
to
p
r
o
v
id
e
th
e
a
u
th
en
ticatio
n
,
th
e
n
with
th
e
h
elp
o
f
th
e
clo
u
d
d
atab
ase
it
m
atch
es
th
e
d
ata
an
d
g
iv
e
b
ac
k
r
esu
lt
in
ac
cu
r
ate
tim
e.
I
n
f
ac
e
r
ec
o
g
n
itio
n
m
eth
o
d
d
if
f
er
e
n
t
f
ac
e
r
ec
o
g
n
itio
n
al
g
o
r
ith
m
an
al
y
s
e
th
e
im
ag
e,
an
d
th
e
n
ex
tr
ac
t
th
at
i
m
ag
e
in
f
o
r
m
atio
n
s
u
ch
as
s
h
a
p
e,
s
ize
an
d
p
o
s
itio
n
o
f
th
e
f
a
cial
d
atab
ase
f
r
o
m
th
e
im
ag
e
to
lo
ca
te
m
atch
in
g
im
ag
e
[
5
]
.
T
o
ex
t
r
ac
t
th
e
im
ag
e,
it
p
er
f
o
r
m
s
th
e
d
if
f
er
e
n
t
p
ar
allel
s
tag
es
o
f
f
ac
e
r
ec
o
g
n
itio
n
alg
o
r
ith
m
in
th
e
cl
o
u
d
.
T
h
r
o
u
g
h
t
h
is
it g
iv
es b
etter
an
d
e
f
f
icien
t o
u
t
p
u
t in
m
in
i
m
al
tim
e
[
3
,
6
,
7
]
.
T
h
is
p
ap
er
is
ab
o
u
t
p
e
r
f
o
r
m
in
g
th
e
f
ac
e
r
ec
o
g
n
itio
n
u
s
in
g
cl
o
u
d
in
g
s
to
r
ag
e.
I
n
Face
r
ec
o
g
n
itio
n
o
v
e
r
th
e
m
o
b
ile
clo
u
d
co
m
p
u
tin
g
,
h
u
g
e
d
ata
th
at
ca
n
b
e
p
r
o
v
id
ed
as
d
atab
ase.
T
h
r
o
u
g
h
th
e
m
o
b
ile
clo
u
d
co
m
p
u
tin
g
it
in
teg
r
ate
th
e
clo
u
d
co
m
p
u
tin
g
m
o
b
ile
en
v
i
r
o
n
m
en
t
in
th
e
m
o
b
ile
p
latf
o
r
m
.
W
ith
m
o
b
ile
clo
u
d
co
m
p
u
tin
g
it
g
i
v
e
th
e
b
etter
o
u
tp
u
t
in
lim
ited
tim
e
p
er
io
d
.
Mo
v
er
o
v
er
m
u
ltip
le
u
s
er
ca
n
d
o
p
r
o
ce
s
s
o
n
clo
u
d
th
r
o
u
g
h
th
e
m
o
b
ile
clo
u
d
co
m
p
u
tin
g
s
im
u
ltan
eo
u
s
ly
.
T
h
e
g
r
o
wth
o
f
th
e
clo
u
d
c
o
m
p
u
tin
g
is
in
cr
ea
s
in
g
in
d
ay
to
d
ay
p
r
o
ce
s
s
o
f
t
h
e
o
r
g
an
izatio
n
.
No
t
o
n
ly
th
e
o
r
g
a
n
izatio
n
b
u
t
r
e
g
u
lar
in
ter
n
et
u
s
er
is
also
u
s
in
g
th
e
clo
u
d
c
o
m
p
u
tin
g
s
er
v
ices
lik
e
Go
o
g
le
D
o
es,
d
r
o
p
b
o
x
an
d
with
th
e
h
elp
o
f
th
ese
f
ea
tu
r
es
th
e
u
s
er
ca
n
ac
ce
s
s
th
eir
f
iles
f
r
o
m
wh
er
ev
er
th
ey
wan
t.
C
lo
u
d
co
m
p
u
tin
g
d
eliv
er
s
a
wid
e
r
an
g
e
o
f
r
eso
u
r
ce
s
lik
e
co
m
p
u
tati
o
n
al
p
o
wer
,
co
m
p
u
tatio
n
al
p
la
tf
o
r
m
,
s
to
r
ag
e
an
d
ap
p
licatio
n
to
u
s
er
in
in
ter
n
et
[
8
]
.
C
lo
u
d
co
m
p
u
tin
g
is
an
em
er
g
in
g
tech
n
o
lo
g
y
it
s
en
d
s
r
eq
u
est
f
o
r
s
er
v
ices
an
d
b
ased
o
n
th
e
r
eq
u
est
it
p
r
o
v
id
e
th
e
r
esu
lt
in
an
o
n
-
d
em
an
d
en
v
ir
o
n
m
e
n
t.
I
n
th
e
clo
u
d
co
m
p
u
tin
g
en
v
ir
o
n
m
en
t
u
s
er
ca
n
ac
ce
s
s
s
er
v
ices
b
ased
o
n
u
s
er
n
ee
d
s
with
o
u
t
k
n
o
win
g
h
o
w
th
e
s
er
v
ices
ar
e
d
eliv
er
ed
an
d
wh
er
e
t
h
e
s
er
v
ice
is
h
o
s
te
d
[
2
,
9
]
.
So
y
ata
T
.
et
al
.
[
1
0
]
d
escr
ib
e
ab
o
u
t
th
at
h
o
w
m
o
b
ile
co
m
p
u
tin
g
an
d
clo
u
d
co
m
p
u
tin
g
ar
e
en
ab
le
to
co
n
v
er
g
e
th
e
n
ew
m
u
ltime
d
ia
ap
p
licatio
n
f
o
r
th
e
clo
u
d
en
v
ir
o
n
m
e
n
t.
I
t
also
d
escr
ib
ed
t
h
at
h
o
w
p
r
o
ce
s
s
is
wo
r
k
in
g
to
th
e
en
d
-
to
-
e
n
d
n
e
two
r
k
b
an
d
wid
t
h
laten
cy
f
o
r
th
e
r
eso
u
r
ce
-
in
ten
s
iv
e
an
d
in
ter
ac
tio
n
-
in
ten
s
iv
e
ap
p
licatio
n
.
I
n
th
e
p
ap
er
it
d
e
s
cr
ib
ed
th
at
in
th
e
clo
u
d
en
v
i
r
o
n
m
en
t
h
o
w
two
lev
el
h
ier
ar
ch
ical
s
tr
u
ctu
r
e
ar
e
p
r
esen
tin
g
th
e
to
d
a
y
s
g
en
er
atio
n
clo
u
d
co
m
p
u
tin
g
in
f
r
a
s
tr
u
ctu
r
e.
I
t
also
d
escr
ib
es
th
at
m
o
b
ile
cl
o
u
d
co
m
p
u
tin
g
g
av
e
p
latf
o
r
m
to
v
ar
io
u
s
ap
p
licatio
n
t
o
p
er
f
o
r
m
v
ar
io
u
s
co
m
p
u
tatio
n
al
task
o
n
th
e
m
o
b
ile
d
e
v
ice.
Ha
K.
et
al
.
[
1
1
]
d
escr
ib
e
d
t
h
at
in
t
h
e
clo
u
d
en
v
ir
o
n
m
e
n
t
h
o
w
two
le
v
el
h
ie
r
ar
ch
ical
s
tr
u
ctu
r
e
a
r
e
p
r
esen
tin
g
th
e
to
d
a
y
s
g
en
e
r
atio
n
clo
u
d
c
o
m
p
u
tin
g
in
f
r
astru
ctu
r
e.
I
t
also
s
p
cif
y
t
h
at
m
o
b
i
le
clo
u
d
co
m
p
u
tin
g
g
av
e
p
latf
o
r
m
t
o
v
a
r
io
u
s
ap
p
licatio
n
to
p
e
r
f
o
r
m
v
a
r
io
u
s
c
o
m
p
u
tatio
n
al
task
o
n
th
e
m
o
b
ile
d
ev
ice.
On
th
e
o
th
er
h
an
d
,
Milo
s
Sto
jm
en
o
v
ic
[
1
2
]
d
is
cu
s
s
ab
o
u
t
h
o
w
b
i
o
m
etr
ic
ap
p
licatio
n
s
ar
e
r
u
n
n
in
g
o
n
th
e
m
o
b
ile
d
ev
ice
an
d
ap
p
licatio
n
ar
e
p
r
o
ce
s
s
in
g
f
aster
th
r
o
u
g
h
th
e
m
o
b
ile
clo
u
d
co
m
p
u
tin
g
.
C
lo
u
d
b
io
m
etr
ic
im
p
r
o
v
ed
th
e
r
ea
l
tim
e
b
io
m
etr
ic
d
ata
with
th
e
h
elp
o
f
t
h
e
clo
u
d
co
m
p
u
tin
g
en
v
i
r
o
n
m
e
n
t.
T
h
e
m
ain
ad
v
a
n
tag
e
is
th
at
it’s r
ed
u
ce
d
th
e
c
o
s
t o
f
th
e
b
io
m
etr
ic
d
ev
ices w
h
ich
ar
e
co
n
n
ec
ted
to
th
e
n
etwo
r
k
.
Yan
g
J
.
et
al
.
[
1
3
]
d
escr
ib
ed
t
h
e
two
ar
c
h
itectu
r
e
b
ased
o
n
t
h
e
I
C
A
alg
o
r
ith
m
f
o
r
r
ec
o
g
n
ize
th
e
f
ac
e.
B
o
th
ar
ch
itectu
r
e
-
I
an
d
ar
c
h
i
tectu
r
e
–
I
I
ar
e
in
v
o
lv
ed
with
th
e
PC
A
b
ased
alg
o
r
ith
m
an
d
b
o
th
ar
ch
itectu
r
es
ar
e
g
en
e
r
atin
g
th
e
m
o
s
tly
s
im
ilar
r
esu
lt.
T
h
e
a
u
th
o
r
d
escr
ib
ed
th
e
I
C
A
alg
o
r
ith
m
is
s
tr
o
n
g
ly
d
e
p
en
d
o
n
th
e
PC
A
b
as
ed
p
r
o
ce
s
s
.
I
C
A
b
as
ed
alg
o
r
ith
m
h
as
litt
le
ef
f
ec
t
o
n
p
er
f
o
r
m
a
n
ce
o
f
th
e
f
ac
e
r
ec
o
g
n
itio
n
.
I
n
t
h
e
p
r
ev
io
u
s
p
ap
er
it
d
escr
ib
ed
a
b
o
u
t
th
e
I
C
A
an
d
PC
A
f
ac
e
r
ec
o
g
n
itio
n
al
g
o
r
ith
m
b
ased
o
n
t
h
e
ar
ch
itectu
r
e
an
d
h
o
w
I
C
A
is
b
ased
o
n
th
e
PC
A
alg
o
r
ith
m
.
Yan
g
,
J
.,
et
al
.
[
1
4
]
d
escr
ib
ed
th
at
d
u
e
to
h
ig
h
d
im
e
n
s
io
n
ality
o
f
im
ag
e
s
p
ac
e
m
a
n
y
L
DA
b
ase
d
ap
p
r
o
ac
h
ar
e
u
s
in
g
PC
A
b
ased
alg
o
r
ith
m
f
ir
s
t
to
m
i
n
im
ize
th
e
d
im
en
s
io
n
ality
o
f
th
e
im
ag
e.
Af
te
r
r
ed
u
cin
g
th
e
s
ize
o
f
im
ag
e,
it
ca
n
p
e
r
f
o
r
m
th
e
L
DA
alg
o
r
ith
m
o
n
im
ag
e.
T
h
e
p
a
p
er
p
r
o
p
o
s
ed
n
ew
alg
o
r
ith
m
f
o
r
th
e
f
ac
e
r
ec
o
g
n
itio
n
wh
ic
h
ca
lled
th
e
d
ir
ec
t
L
DA
alg
o
r
ith
m
an
d
it
is
th
e
co
m
b
in
atio
n
o
f
th
e
L
DA
+
PC
A
alg
o
r
ith
m
[
1
3
,
1
4
]
.
L
u
J
.
et
al
.
[
1
5
]
also
p
r
o
p
o
s
ed
th
e
DF
-
L
DA
b
ased
alg
o
r
ith
m
in
th
e
f
ac
e
r
ec
o
g
n
itio
n
b
u
t
th
is
alg
o
r
ith
m
h
as
s
o
m
e
p
r
o
b
lem
with
th
e
lar
g
e
s
ize
o
f
th
e
f
ac
e
r
ec
o
g
n
itio
n
an
d
it
d
ev
el
o
p
in
g
th
e
n
o
n
-
lin
ea
r
alg
o
r
ith
m
.
Geo
r
g
escu
D.
[
4
]
a
u
th
o
r
p
r
o
p
o
s
ed
to
v
er
i
f
y
th
e
f
ac
e
r
ec
o
g
n
itio
n
alg
o
r
ith
m
with
h
elp
o
f
E
ig
en
f
ac
e
tech
n
iq
u
es.
E
ig
e
n
f
ac
e
tec
h
n
i
q
u
es
ca
n
b
e
s
o
lv
in
g
with
h
el
p
o
f
PC
A
b
ased
f
ac
e
r
ec
o
g
n
it
io
n
alg
o
r
ith
m
.
T
h
is
E
ig
en
f
ac
e
tech
n
iq
u
e
is
s
im
p
le,
r
o
b
u
s
t
an
d
ea
s
y
f
o
r
th
e
im
p
lem
en
tatio
n
.
I
t
r
e
q
u
ir
es
th
e
lar
g
e
am
o
u
n
t
o
f
th
e
d
ata
in
f
ac
e
r
ec
o
g
n
itio
n
s
y
s
tem
f
o
r
f
ast r
esu
lt.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Reco
n
f
ig
u
r
a
b
le
&
E
m
b
ed
d
ed
Sy
s
t
I
SS
N:
2089
-
4
8
6
4
Desig
n
a
n
d
d
ev
elo
p
men
t
o
f a
p
a
r
a
lleliz
ed
a
lg
o
r
ith
m
fo
r
fa
c
e
r
ec
o
g
n
itio
n
in
…
(
K
.
N
.
B
h
a
t
t
)
49
W
h
ile
o
n
r
esear
ch
it
co
n
cl
u
d
ed
th
at
f
o
r
th
e
f
ac
e
r
e
co
g
n
itio
n
alg
o
r
ith
m
ea
ch
an
d
ev
e
r
y
m
eth
o
d
ar
e
u
s
in
g
th
e
PC
A
b
ase
alg
o
r
ith
m
in
o
n
e
o
r
an
o
t
h
er
way
[
1
6
-
1
8
]
.
On
th
e
b
io
m
etr
ic
m
o
b
ile
clo
u
d
co
m
p
u
tin
g
th
er
e
is
s
y
s
tem
atic
ap
p
r
o
ac
h
o
f
p
r
o
p
er
r
ec
o
g
n
itio
n
o
f
f
ac
e.
E
ac
h
a
n
d
ev
er
y
alg
o
r
ith
m
ic
m
eth
o
d
s
an
d
tech
n
i
q
u
es
h
as
th
eir
o
wn
ad
v
an
tag
e
a
n
d
lim
itatio
n
.
T
h
er
e
is
n
o
t
p
r
o
p
e
r
ap
p
r
o
ac
h
f
o
r
th
e
p
ar
allel
alg
o
r
ith
m
wh
ich
ca
n
wo
r
k
p
ar
allel
o
n
th
e
clo
u
d
s
er
v
er
f
o
r
f
ac
e
r
ec
o
g
n
itio
n
to
m
atc
h
r
esu
lt.
A
p
r
o
p
er
ap
p
r
o
ac
h
o
f
co
m
m
u
n
icatio
n
b
etwe
en
th
e
s
o
u
r
ce
an
d
cl
o
u
d
d
ata
is
n
o
t f
o
u
n
d
i
n
s
u
r
v
ey
ed
l
iter
atu
r
e
[
1
6
-
1
8
]
.
T
h
e
m
o
tiv
atio
n
o
f
th
is
p
ap
e
r
i
s
to
d
esig
n
an
d
d
e
v
elo
p
p
ar
all
elize
d
alg
o
r
ith
m
f
o
r
th
e
f
ac
e
r
ec
o
g
n
itio
n
in
th
e
m
o
b
ile
clo
u
d
co
m
p
u
ti
n
g
.
T
h
e
m
ain
aim
is
to
d
ev
elo
p
th
e
p
ar
allel
b
ased
alg
o
r
ith
m
f
o
r
th
e
f
ac
e
r
ec
o
g
n
itio
n
m
et
h
o
d
o
n
th
e
clo
u
d
s
er
v
er
.
So
,
with
th
e
h
e
lp
o
f
p
ar
allel
f
ac
e
r
ec
o
g
n
itio
n
alg
o
r
ith
m
it
ca
n
r
ed
u
ce
d
th
e
laten
c
y
o
f
th
e
s
y
s
tem
an
d
g
iv
e
th
e
b
etter
o
u
tp
u
t
f
o
r
th
e
la
r
g
e
s
ca
le
d
a
ta
b
ase
p
r
o
ce
s
s
.
I
n
t
h
e
cu
r
r
en
t
s
ce
n
a
r
io
f
ac
e
r
ec
o
g
n
it
io
n
ap
p
licatio
n
is
wo
r
k
in
g
o
n
d
if
f
er
e
n
t
alg
o
r
ith
m
i
n
d
i
f
f
er
e
n
t
way
b
u
t
th
e
r
e
is
n
o
ef
f
icien
t a
l
g
o
r
ith
m
wh
ich
c
an
wo
r
k
p
ar
allely
an
d
r
ed
u
ce
t
h
e
tim
e
co
m
p
le
x
ity
o
f
th
e
s
y
s
tem
.
2.
DE
S
I
G
N
O
F
P
ARAL
L
E
L
A
L
G
O
RIT
H
M
F
O
R
F
ACE R
E
CO
G
NIS
AT
I
O
N
I
N
M
O
B
I
L
E
C
L
O
UD
CO
M
P
UT
I
NG
I
n
th
is
s
ec
tio
n
,
t
h
e
d
esig
n
p
r
o
c
ess
is
m
ain
ly
d
iv
id
ed
in
to
two
.
−
Ad
d
in
g
tr
ain
i
n
g
d
ataset
to
th
e
s
er
v
er
−
Valid
atio
n
o
f
th
e
im
a
g
e
I
f
t
h
e
f
ac
e
r
ec
o
g
n
itio
n
ap
p
licatio
n
f
o
llo
ws
ad
d
in
g
t
h
e
im
a
g
e
s
tr
ain
in
g
d
ataset
to
th
e
s
er
v
e
r
th
en
th
e
ap
p
licatio
n
f
o
llo
ws
th
e
f
ash
io
n
as f
o
llo
ws.
2
.
1
.
Sequ
ence
dia
g
ra
m
f
o
r
t
he
s
y
s
t
em
a
pp
lica
t
io
n
Fig
u
r
e
1
s
h
o
w
th
e
s
eq
u
en
ce
d
iag
r
am
f
o
r
r
eg
is
ter
in
g
th
e
im
a
g
e.
T
h
e
s
eq
u
e
n
ce
d
iag
r
am
s
h
o
ws
th
r
ee
o
b
jects
o
f
th
e
s
y
s
tem
an
d
t
h
e
i
n
ter
ac
tio
n
am
o
n
g
th
e
m
.
Firstl
y
,
f
ac
e
r
ec
o
g
n
itio
n
ap
p
licatio
n
is
r
eg
is
ter
in
g
with
th
e
n
am
e
a
n
d
s
en
d
it
to
s
er
v
er
.
No
w,
th
e
s
er
v
er
ac
k
n
o
wled
g
es
th
e
f
ac
e
r
ec
o
g
n
itio
n
ap
p
licatio
n
in
d
icatin
g
th
at
th
e
n
am
e
h
as
b
ee
n
r
eg
is
ter
.
On
ce
th
e
r
eg
is
tr
atio
n
h
as
b
ee
n
d
o
n
e
s
u
cc
ess
f
u
lly
n
ex
t
s
tep
is
to
ca
p
tu
r
e
th
e
im
ag
es
o
f
u
s
er
in
f
iv
e
d
if
f
er
e
n
t
an
g
els
an
d
s
en
d
in
g
to
th
e
s
er
v
er
.
T
h
e
s
er
v
er
in
ter
n
s
en
d
s
th
e
s
ca
l
ed
im
ag
es
to
th
e
tr
ain
in
g
d
ataset
o
f
th
e
s
y
s
tem
[
8
]
.
On
ce
t
h
e
tr
ain
in
g
d
at
aset
m
ag
es
ar
e
s
av
ed
a
co
n
f
o
r
m
atio
n
m
ess
ag
e
is
b
ee
n
s
en
d
to
b
e
s
er
v
er
.
On
ce
t
h
e
co
n
f
o
r
m
atio
n
m
ess
ag
e
f
r
o
m
th
e
tr
ain
in
g
d
ataset
is
r
ec
eiv
ed
,
th
e
s
er
v
e
r
s
en
d
s
an
ac
k
n
o
wled
g
m
en
t
m
ess
ag
e
to
th
e
f
ac
e
r
ec
o
g
n
itio
n
ap
p
lic
atio
n
to
asti
n
g
th
at
im
ag
es
ar
e
s
av
ed
s
u
cc
ess
f
u
lly
o
n
th
e
clo
u
d
s
er
v
e
r
.
Fig
u
r
e
1
.
Seq
u
en
ce
d
ia
g
r
am
f
o
r
r
eg
is
ter
in
g
th
e
im
ag
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
20
89
-
4
8
6
4
I
n
t J Reco
n
f
ig
u
r
a
b
le
&
E
m
b
ed
d
ed
Sy
s
t
,
Vo
l.
10
,
No
.
1
,
Ma
r
c
h
202
1
:
47
–
55
50
2
.
2
.
Sequ
ence
dia
g
ra
m
f
o
r
t
he
s
y
s
t
em
s
er
v
er
Fig
u
r
e
2
s
h
o
ws
th
e
s
eq
u
e
n
ce
d
iag
r
a
m
f
o
r
v
alid
atin
g
th
e
i
m
ag
e.
T
h
e
s
eq
u
en
ce
d
iag
r
a
m
s
h
o
ws
th
e
in
ter
ac
tio
n
s
b
etwe
en
th
r
ee
o
b
jects
o
f
th
e
s
y
s
tem
.
Firstl
y
,
u
s
er
s
en
d
s
r
eq
u
est
to
th
e
s
er
v
er
th
r
o
u
g
h
th
e
f
ac
e
r
ec
o
g
n
itio
n
a
p
p
licatio
n
.
W
h
e
n
s
er
v
er
g
ets
th
e
r
eq
u
est
f
r
o
m
th
e
ap
p
licatio
n
it
g
iv
es
r
e
s
p
o
n
s
e
b
ac
k
to
th
e
ap
p
licatio
n
th
at
r
eq
u
est
is
ac
c
ep
ted
.
On
ce
a
p
o
s
itiv
e
r
esp
o
n
s
e
co
m
es
b
ac
k
s
t
o
th
e
a
p
p
licatio
n
,
th
e
ap
p
licatio
n
s
ac
tiv
ates
th
e
ca
m
er
a
in
o
r
d
er
t
o
ca
p
tu
r
e
t
h
e
im
ag
e.
T
h
e
ca
p
t
u
r
ed
im
ag
e
is
th
en
s
en
t
it
to
th
e
s
er
v
er
f
o
r
f
u
r
t
h
er
v
er
if
icatio
n
.
On
ce
t
h
e
s
er
v
er
a
cq
u
ir
es
th
e
im
ag
e,
it
s
ea
r
ch
es
f
o
r
th
e
h
i
g
h
est
m
atch
in
g
im
a
g
e.
On
th
e
b
ac
k
e
n
d
,
th
e
im
ag
e
is
m
atch
ed
with
th
e
tr
ain
in
g
d
ataset.
On
ce
th
e
s
er
v
er
h
as
th
e
r
esu
lt,
it
s
en
d
s
th
e
r
esu
lt
b
ac
k
t
o
th
e
f
ac
e
r
ec
o
g
n
itio
n
ap
p
licatio
n
.
T
h
e
f
ac
e
r
ec
o
g
n
itio
n
ap
p
licatio
n
ca
p
tu
r
es
th
e
im
ag
es
th
ese
ca
p
tu
r
ed
im
ag
es
ar
e
s
o
co
n
v
e
r
ted
f
r
o
m
co
lo
u
r
ed
im
ag
e
t
o
th
e
g
r
ey
s
ca
led
im
ag
e
a
n
d
s
en
d
to
th
e
s
er
v
er
.
I
f
th
e
im
ag
e
is
g
e
n
er
ate
d
f
o
r
th
e
f
ir
s
t
tim
e,
th
en
it
f
etch
es
a
n
d
s
av
e
o
n
th
e
tr
ain
in
g
d
ataset
o
f
t
h
e
d
at
ab
a
s
e
an
d
g
iv
e
th
e
r
esp
o
n
s
e
b
ac
k
to
th
e
s
er
v
er
[
7
]
.
If
th
e
u
s
er
wan
ts
to
v
alid
ate
t
h
e
im
ag
e,
th
en
th
e
ca
p
tu
r
e
d
im
ag
e
is
co
n
v
e
r
ted
in
to
s
ca
led
g
r
ey
im
ag
e
a
n
d
is
b
ee
n
s
en
d
to
th
e
cl
o
u
d
s
er
v
er
.
On
t
h
e
clo
u
d
s
er
v
er
s
id
e,
it
f
etc
h
e
s
th
e
im
ag
e
an
d
ap
p
lies
PC
A
an
d
f
in
d
s
th
e
m
o
s
t
s
im
ilar
with
th
e
h
elp
o
f
t
h
e
E
u
clid
ian
m
eth
o
d
in
o
r
d
er
to
g
en
er
ate
th
e
r
esu
lt.
Fig
u
r
e
2
.
Seq
u
en
ce
d
ia
g
r
am
f
o
r
v
alid
atin
g
th
e
im
ag
e
2
.
3
.
P
a
ra
llel P
CA
a
ng
o
rit
h
m
f
o
r
t
he
s
y
s
t
em
T
h
is
s
ec
tio
n
d
escr
ib
es
h
o
w
P
C
A
alg
o
r
ith
m
wo
r
k
s
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
tem
f
o
r
p
ar
allelize
d
a
lg
o
r
ith
m
f
o
r
f
ace
r
ec
o
g
n
itio
n
in
m
o
b
ile
clo
u
d
co
m
p
u
tin
g
an
d
h
o
w
p
ar
allel
PC
A
alg
o
r
ith
m
is
wo
r
k
in
g
in
th
e
s
y
s
tem
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Reco
n
f
ig
u
r
a
b
le
&
E
m
b
ed
d
ed
Sy
s
t
I
SS
N:
2089
-
4
8
6
4
Desig
n
a
n
d
d
ev
elo
p
men
t
o
f a
p
a
r
a
lleliz
ed
a
lg
o
r
ith
m
fo
r
fa
c
e
r
ec
o
g
n
itio
n
in
…
(
K
.
N
.
B
h
a
t
t
)
51
T
h
e
Par
allel
PC
A
alg
o
r
ith
m
wo
r
k
in
g
i
s
as f
o
llo
ws as sh
o
wn
in
Fig
u
r
e
3
:
−
Step
1
: PC
A
alg
o
r
ith
m
is
im
p
lem
en
ted
o
n
th
e
s
er
v
er
.
−
Step
2
:
Firstl
y
,
it
tak
es
a
s
ca
led
im
ag
e
f
r
o
m
im
ag
e
ca
p
tu
r
i
n
g
d
ev
ice
a
n
d
m
ap
th
e
im
ag
e
in
t
h
e
f
o
r
m
o
f
m
atr
ix
−
Step
3
:
Af
ter
co
n
v
er
tin
g
to
im
ag
e
in
to
m
atr
ix
f
o
r
m
,
it
s
av
es
th
e
h
ash
m
ap
in
g
v
alu
e
an
d
lo
ad
s
im
ag
e
in
tex
t f
o
r
m
at.
−
Step
4
: T
h
ese
g
en
er
ate
d
f
o
r
m
a
ts
ar
e
th
e
tr
ain
in
g
d
ataset
−
Step
5
:
On
ce
th
e
tr
ain
in
g
d
ataset
is
b
ein
g
g
en
er
ate
d
,
E
ig
e
n
f
ac
e
o
f
ea
ch
im
ag
es
tr
ain
in
g
d
ataset
ar
e
o
b
tain
ed
.
−
Step
6
:
Af
ter
g
en
er
atin
g
th
e
E
ig
en
f
ac
es
f
o
r
th
e
t
r
ain
i
n
g
d
at
aset
im
ag
es,
it
b
u
ild
s
m
atr
ix
f
o
r
ea
ch
im
ag
e
as
well
as
th
e
tr
an
s
p
o
s
e
m
atr
i
x
f
o
r
m
a
r
e
g
en
er
ated
an
d
f
r
o
m
th
e
t
r
an
s
p
o
s
e
m
atr
i
x
v
ec
to
r
s
o
f
th
e
im
ag
es
ar
e
g
en
er
ate
d
.
−
Step
7
:
Af
ter
g
en
er
atin
g
th
e
v
ec
to
r
v
alu
e
o
f
th
e
im
ag
e,
it
f
in
d
s
th
e
m
o
s
t
s
im
ilar
im
ag
es
o
f
th
e
ca
p
tu
r
ed
im
ag
e
b
y
c
o
m
p
ar
i
n
g
it with
th
e
tr
ain
in
g
d
ataset
im
ag
es
−
Step
8
:
Af
ter
m
a
p
p
in
g
th
e
im
ag
e
with
E
cu
lin
d
is
tan
ce
with
th
e
v
ec
to
r
o
f
an
im
a
g
e
f
in
al
r
esu
lt
is
b
ee
n
g
en
er
ated
.
Fig
u
r
e
3
.
B
lo
ck
d
iag
r
am
f
o
r
p
ar
allel
PC
A
alg
o
r
ith
m
3.
I
M
P
L
E
M
E
NT
A
T
I
O
N
O
F
P
ARAL
L
E
L
AL
G
O
RI
T
H
M
F
O
R
F
ACE
RE
CO
G
N
I
SA
T
I
O
N
I
N
M
O
B
I
L
E
CL
O
UD
CO
M
P
U
T
I
NG
I
n
th
is
p
ap
er
,
p
ar
allel
f
ac
e
r
ec
o
g
n
itio
n
alg
o
r
ith
m
is
d
esig
n
ed
an
d
d
ev
elo
p
ed
with
h
elp
o
f
th
e
m
o
b
ile
clo
u
d
co
m
p
u
tin
g
as
s
h
o
wn
in
Fig
u
r
e
4
.
T
h
is
is
b
ein
g
ac
h
iev
ed
b
y
d
ev
el
o
p
in
g
a
n
an
d
r
o
id
a
p
p
licatio
n
f
o
r
f
ac
e
r
ec
o
g
n
itio
n
an
d
m
ak
in
g
a
clo
u
d
s
er
v
er
.
H
er
e
an
an
d
r
o
id
ap
p
licatio
n
is
d
ev
elo
p
ed
f
o
r
th
e
f
ac
e
r
ec
o
g
n
itio
n
p
r
o
ce
s
s
.
An
an
d
r
o
id
a
p
p
licatio
n
is
u
s
ed
to
r
e
g
is
ter
o
r
v
alid
ate
th
e
u
s
er
.
T
h
is
ap
p
licatio
n
ca
n
r
eg
is
ter
th
e
n
ew
u
s
er
b
y
p
r
o
v
id
in
g
u
s
er
’
s
n
am
e
an
d
f
iv
e
im
ag
es
tak
en
in
d
i
f
f
er
en
t
an
g
le
.
T
h
ese
tak
en
im
ag
es
alo
n
g
with
th
e
u
s
er
’
s
n
am
e
is
k
e
p
t
as
th
e
tr
ai
n
in
g
d
ataset
if
th
e
u
s
er
r
eq
u
est
f
o
r
v
alid
atin
g
t
h
e
im
ag
e
af
te
r
n
ew
r
e
g
is
tr
at
io
n
.
T
h
e
tr
ain
in
g
d
ataset
is
s
to
r
ed
in
th
e
clo
u
d
s
ev
er
.
I
f
u
s
er
r
e
q
u
est
f
o
r
v
alid
atio
n
th
r
o
u
g
h
f
ac
e
r
ec
o
g
n
itio
n
ap
p
licatio
n
,
th
e
tak
e
n
im
a
g
e
is
s
en
t
to
th
e
s
er
v
er
s
o
as
to
p
er
f
o
r
m
PC
A
alg
o
r
ith
m
o
n
it
t
o
g
e
n
er
ate
th
e
r
esu
lt.
Her
e,
t
h
e
s
er
v
er
f
etc
h
es
th
e
im
a
g
es
f
r
o
m
th
e
tr
ain
in
g
d
ataset
an
d
co
m
p
a
r
es with
th
e
im
ag
e
s
en
d
b
y
th
e
f
ac
e
r
ec
o
g
n
itio
n
ap
p
licatio
n
in
o
r
d
er
to
g
en
e
r
ate
th
e
o
u
tp
u
t.
T
h
e
co
m
p
ar
is
o
n
r
esu
lt
is
n
o
th
in
g
b
u
t
f
in
d
in
g
th
e
m
o
s
t
s
im
ilar
im
ag
e
in
th
e
tr
ain
in
g
d
ataset.
T
h
e
d
esig
n
ed
s
y
s
tem
is
b
ein
g
te
s
ted
in
ce
n
tr
alis
ed
an
d
d
is
tr
ib
u
ted
s
er
v
er
s
y
s
tem
in
o
r
d
e
r
to
f
in
d
th
e
tim
e
co
m
p
lex
ity
o
f
ea
ch
s
y
s
tem
an
d
d
o
a
co
m
p
ar
ativ
e
a
n
aly
s
is
o
n
b
o
th
th
e
s
y
s
tem
as
s
h
o
w
n
i
n
Fig
u
r
e
s
5
a
n
d
6
.
T
h
e
m
ain
m
o
tiv
e
o
f
d
e
v
elo
p
in
g
th
e
s
y
s
tem
in
b
o
th
ce
n
tr
alis
e
d
an
d
d
is
tr
ib
u
ted
s
er
v
er
s
y
s
tem
is
to
co
m
m
en
t o
n
th
e
p
er
f
o
r
m
an
ce
o
f
t
h
e
d
ev
el
o
p
ed
s
y
s
tem
.
E
v
alu
atio
n
o
f
th
e
s
y
s
tem
was
ca
r
r
ied
o
u
t
b
ased
o
n
th
e
tim
e
tak
en
to
p
er
f
o
r
m
th
e
task
if
th
e
n
u
m
b
er
o
f
i
m
ag
es in
tr
ain
in
g
d
atas
et
k
ee
p
g
ettin
g
in
cr
ea
s
ed
.
T
h
e
an
aly
s
is
o
f
th
e
s
y
s
tem
is
ca
r
r
ied
o
u
t
b
y
co
n
s
id
er
in
g
th
e
s
er
v
er
as
ce
n
tr
alize
d
an
d
d
is
tr
ib
u
ted
.
T
h
e
p
ar
a
m
eter
h
er
e
c
o
n
s
id
er
e
d
in
o
r
d
e
r
to
co
m
p
a
r
e
th
e
p
er
f
o
r
m
an
ce
o
f
b
o
t
h
th
e
s
y
s
tem
i
s
th
e
tim
e
tak
e
n
to
p
er
f
o
r
m
th
e
p
ar
ticu
lar
task
.
I
n
ce
n
tr
alize
d
s
y
s
tem
,
im
ag
es
ar
e
s
en
t
f
r
o
m
th
e
f
a
ce
r
ec
o
g
n
iti
o
n
ap
p
licatio
n
to
a
s
in
g
le
s
er
v
er
in
o
r
d
er
t
o
p
er
f
o
r
m
th
e
p
a
r
ticu
lar
task
s
elec
ted
b
y
th
e
u
s
er
.
I
n
d
is
tr
ib
u
ted
s
y
s
tem
,
im
ag
es
tak
en
b
y
f
ac
e
r
ec
o
g
n
itio
n
ap
p
licatio
n
ar
e
s
en
t
to
th
e
ce
n
tr
al
s
er
v
er
o
f
th
e
s
y
s
tem
.
No
w,
ce
n
tr
al
s
er
v
er
o
f
th
e
s
y
s
tem
s
en
d
s
th
e
r
ec
eiv
ed
i
n
f
o
r
m
atio
n
to
f
o
u
r
o
th
er
s
er
v
e
r
s
wh
ich
ar
e
co
n
n
ec
te
d
to
th
e
m
ain
ce
n
tr
e
s
er
v
er
.
T
h
e
m
ain
m
o
tiv
e
is
to
d
iv
id
e
th
e
task
am
o
n
g
th
e
s
ev
er
s
o
as
to
g
en
er
ate
r
esu
lt
in
les
s
tim
e
an
d
m
ak
e
th
e
s
y
s
tem
f
aster
.
T
h
at
is
,
it in
th
e
d
is
tr
ib
u
ted
s
er
v
e
r
,
th
e
p
r
o
ce
s
s
is
r
u
n
n
in
g
co
n
cu
r
r
en
tly
t
o
g
en
e
r
ate
th
e
r
esu
lt
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
20
89
-
4
8
6
4
I
n
t J Reco
n
f
ig
u
r
a
b
le
&
E
m
b
ed
d
ed
Sy
s
t
,
Vo
l.
10
,
No
.
1
,
Ma
r
c
h
202
1
:
47
–
55
52
Fig
u
r
e
4
.
Face
R
ec
o
g
n
is
atio
n
ap
p
licatio
n
in
m
o
b
ile
clo
u
d
en
v
ir
o
n
m
e
n
t
Fig
u
r
e
5
.
B
lo
ck
d
iag
r
am
o
f
ce
n
tr
alis
ed
s
er
v
er
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Reco
n
f
ig
u
r
a
b
le
&
E
m
b
ed
d
ed
Sy
s
t
I
SS
N:
2089
-
4
8
6
4
Desig
n
a
n
d
d
ev
elo
p
men
t
o
f a
p
a
r
a
lleliz
ed
a
lg
o
r
ith
m
fo
r
fa
c
e
r
ec
o
g
n
itio
n
in
…
(
K
.
N
.
B
h
a
t
t
)
53
Fig
u
r
e
6
.
B
lo
ck
d
iag
r
am
o
f
d
is
tr
ib
u
ted
s
er
v
er
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
NS
Fig
u
r
e
7
s
h
o
ws
th
e
g
r
a
p
h
i
ca
l
r
ep
r
esen
tatio
n
o
f
co
m
p
ar
ativ
e
an
aly
s
is
b
etwe
en
th
e
ce
n
tr
a
lized
an
d
d
is
tr
ib
u
ted
s
y
s
tem
s
er
v
er
.
Fig
u
r
e
5
s
h
o
ws
th
e
tim
e
tak
en
in
o
r
d
er
to
co
m
p
lete
th
e
p
r
o
ce
s
s
wh
ile
v
ar
y
in
g
th
e
n
u
m
b
er
o
f
im
a
g
es
in
t
h
e
tr
ai
n
in
g
d
ataset.
T
h
e
o
r
an
g
e
b
ar
is
d
en
o
ted
as
d
is
tr
ib
u
ted
s
y
s
tem
wh
er
ea
s
t
h
e
b
l
u
e
b
ar
is
d
en
o
ted
as
c
en
tr
alize
d
s
er
v
er
s
y
s
tem
.
T
h
e
x
ax
is
is
lab
elled
as
th
e
T
im
e
-
tak
en
in
s
ec
o
n
d
s
an
d
Y
-
a
x
is
is
co
n
s
id
er
ed
as
th
e
n
u
m
b
er
o
f
i
m
ag
es
tak
en
in
th
e
tr
ain
in
g
d
a
taset.
Ob
s
er
v
in
g
th
e
Fig
u
r
e
7
i
t
is
clea
r
th
at
th
e
if
th
e
s
y
s
tem
is
d
esig
n
ed
in
d
is
tr
ib
u
ted
way
,
it
is
tak
in
g
less
tim
e
to
co
m
p
lete
th
e
p
r
o
ce
s
s
a
s
co
m
p
ar
ed
to
th
e
tim
e
tak
en
b
y
th
e
ce
n
tr
alize
d
s
y
s
tem
.
Hen
ce
,
it
ca
n
b
e
c
o
n
cl
u
d
ed
th
at
th
e
d
is
tr
ib
u
te
d
s
er
v
e
r
s
y
s
tem
ac
ts
f
aster
f
o
r
b
etter
a
n
d
e
f
f
icien
t o
u
t
p
u
t
th
an
th
e
ce
n
tr
alize
d
s
y
s
tem
.
Fig
u
r
e
7
.
C
o
m
p
a
r
itiv
e
tim
e
co
m
p
lex
ity
an
aly
s
is
b
etwe
en
ce
n
tr
alize
d
an
d
d
is
tr
ib
u
te
d
s
y
s
tem
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
20
89
-
4
8
6
4
I
n
t J Reco
n
f
ig
u
r
a
b
le
&
E
m
b
ed
d
ed
Sy
s
t
,
Vo
l.
10
,
No
.
1
,
Ma
r
c
h
202
1
:
47
–
55
54
I
n
th
e
ce
n
tr
alize
d
s
er
v
er
,
r
e
q
u
est
is
d
ir
ec
tly
s
en
t
f
r
o
m
th
e
f
ac
e
r
ec
o
g
n
itio
n
ap
p
licatio
n
to
th
e
R
E
ST
s
er
v
er
in
o
r
d
er
to
p
er
f
o
r
m
t
h
e
r
elev
an
t
task
.
No
w,
as
th
e
n
u
m
b
er
o
f
tr
ain
in
g
d
ataset
im
ag
es
k
ee
p
s
in
cr
ea
s
in
g
th
e
r
esp
o
n
s
e
tim
e
o
f
t
h
e
s
er
v
e
r
k
ee
p
s
g
r
ad
u
ally
g
ain
s
m
o
r
e
w
h
ich
lead
s
to
d
if
f
icu
lt
y
in
r
esp
o
n
d
in
g
b
ac
k
t
o
th
e
f
ac
e
r
ec
o
g
n
itio
n
ap
p
licatio
n
.
T
h
is
m
ay
ev
en
lea
d
to
f
ailu
r
e
o
f
th
e
s
y
s
tem
.
W
h
er
ea
s
in
th
e
d
is
tr
ib
u
te
d
s
er
v
er
,
f
ir
s
t
r
eq
u
est
is
b
ein
g
s
en
t
to
th
e
ce
n
tr
al
s
ev
er
f
r
o
m
th
e
f
ac
e
r
ec
o
g
n
itio
n
ap
p
licatio
n
.
No
w,
in
b
ac
k
en
d
,
th
e
ce
n
tr
al
s
er
v
er
s
en
d
s
th
e
s
am
e
r
eq
u
est
to
th
e
r
est
o
f
th
e
d
if
f
er
e
n
t
s
er
v
er
in
o
r
d
er
to
p
er
f
o
r
m
th
e
task
in
p
ar
allel
way
.
T
h
is
d
is
tr
ib
u
tio
n
f
ash
io
n
o
f
task
ca
n
a
ch
iev
e
in
cr
ea
s
e
in
p
er
f
o
r
m
an
ce
o
f
th
e
s
y
s
tem
.
B
ec
au
s
e
as
th
e
task
i
s
b
ee
n
d
iv
id
ed
am
o
u
n
t
th
e
s
er
v
er
s
,
th
e
tim
e
tak
en
to
p
r
o
ce
s
s
th
e
task
will b
e
less
.
T
h
u
s
,
d
is
tr
ib
u
ted
s
y
s
tem
lead
s
to
a
b
etter
an
d
e
f
f
icien
t u
s
e
o
f
th
e
s
y
s
tem
.
5.
CO
NCLU
SI
O
NS
T
h
e
d
e
v
lo
p
e
d
s
y
s
tem
is
g
iv
in
g
r
esu
lt
o
n
th
e
d
if
f
er
en
t
ty
p
e
o
f
th
e
f
ac
e
ex
p
r
ess
io
n
an
d
em
o
tio
n
o
f
th
e
f
ac
e.
T
h
e
p
ar
allel
PC
A
alg
o
r
it
h
m
is
p
r
o
v
d
in
g
th
e
m
o
s
t
s
im
ilar
f
ac
e
wh
ile
v
alid
atin
g
t
h
e
f
a
ce
im
ag
e
f
r
o
m
th
e
im
ag
e
d
ataset.
I
d
en
tify
in
g
th
e
f
ac
e
r
ec
o
g
n
itio
n
p
r
o
ce
s
s
is
ac
h
iev
ed
with
th
e
h
elp
o
f
th
e
p
ar
allel
PC
A
an
d
E
cu
lian
d
is
tan
ce
.
Par
allel
P
C
A
f
ac
e
r
ec
o
g
n
itio
n
ag
o
r
ith
m
is
ap
p
ly
in
g
o
n
th
e
clo
u
d
s
er
v
er
an
d
it
m
atch
th
e
im
ag
e
o
n
th
e
cl
o
u
d
s
er
v
er
,
a
f
ter
m
atch
in
g
im
ag
e
,
it
g
i
v
es
th
e
r
esp
o
n
s
e
b
ac
k
to
t
h
e
u
s
er
.
T
h
e
d
ev
el
o
p
ed
s
y
s
tem
is
d
esig
n
ed
in
b
o
th
ce
n
tr
alize
d
an
d
d
is
tr
ib
u
ted
s
er
v
er
s
.
I
t
ca
n
b
e
o
b
s
er
v
ed
th
at;
c
en
tr
alize
d
s
er
v
er
is
tak
in
g
m
o
r
e
tim
e
to
g
iv
e
r
esp
o
n
s
e
as
co
m
p
ar
ed
to
th
e
d
is
tr
ib
u
ted
s
er
v
er
if
n
u
m
b
er
o
f
im
a
g
es
in
th
e
tr
ain
in
g
d
ataset
is
h
ig
h
.
B
u
t
if
n
u
m
b
er
o
f
im
ag
es
in
th
e
tr
ain
in
g
d
ataset
is
lo
w,
th
en
ce
n
tr
alize
d
s
er
v
er
is
tak
in
g
less
tim
e
as c
o
m
p
ar
ed
t
o
th
e
d
is
tr
i
b
u
ted
s
er
v
er
.
I
t
s
h
o
ws
th
at
d
ev
elo
p
ed
s
y
s
t
em
is
ab
le
to
ac
h
iev
e
h
ig
h
s
tab
ilit
y
an
d
h
ig
h
er
p
er
f
o
r
m
a
n
ce
wh
ile
d
esig
n
ed
in
d
is
tr
ib
u
ted
s
er
v
e
r
s
y
s
tem
.
W
ith
th
e
h
elp
o
f
cl
o
u
d
b
ased
s
y
s
tem
it’s
p
o
s
s
ib
le
to
p
r
o
v
id
e
h
u
g
e
n
u
m
b
er
o
f
t
h
e
d
ata
an
d
g
iv
e
t
h
e
ac
cu
r
ate
r
esu
lt
in
less
am
o
u
n
t
o
f
t
h
e
tim
e.
T
h
e
d
ev
el
o
p
e
d
ap
p
licatio
n
ca
n
b
e
u
s
ed
b
y
n
n
u
m
b
e
r
o
f
u
s
er
s
s
im
u
ltan
eo
u
s
ly
in
o
r
d
er
in
ad
d
h
im
s
elf
/h
er
s
elf
to
th
e
t
r
ain
in
g
d
ataset
o
r
to
r
eq
u
est
f
o
r
v
alid
atin
g
th
e
im
ag
e.
T
h
e
p
r
esen
t
wo
r
k
ca
n
s
till
b
e
im
p
r
o
v
ed
b
y
co
m
b
in
in
g
PC
A
b
ased
f
ac
e
r
ec
o
g
n
itio
n
alg
o
r
ith
m
with
s
o
m
e
o
th
e
r
f
a
ce
r
ec
o
g
n
itio
n
al
g
o
r
ith
m
f
o
r
b
etter
ef
f
icien
cy
an
d
ac
cu
r
ate
r
esu
lt.
T
o
im
p
r
o
v
e
th
e
ef
f
icien
cy
o
f
t
h
e
d
ev
el
o
p
ed
s
y
s
tem
,
in
teg
r
atio
n
o
f
au
d
io
an
d
v
id
eo
o
f
u
s
er
ca
n
b
e
p
r
o
v
id
ed
in
o
r
d
er
to
id
en
tify
p
er
s
o
n
with
f
ac
e
an
d
v
o
ice,
it
lead
s
to
ac
h
iev
e
h
ig
h
r
ec
o
g
n
itio
n
r
ate.
T
o
im
p
r
o
v
e
t
h
e
clo
u
d
with
m
o
r
e
h
ier
ar
ch
y
clo
u
d
b
ased
s
er
v
er
to
im
p
r
o
v
e
th
e
r
esu
lt a
n
d
em
e
r
g
in
g
with
h
i
g
h
ac
cu
r
ac
y
in
les
s
tim
e.
RE
F
E
R
E
NC
E
S
[1
]
Ay
a
d
,
M
.
,
Tah
e
r,
M
.
a
n
d
S
a
lem
,
A.,
“
Re
a
l
-
ti
m
e
m
o
b
il
e
c
lo
u
d
c
o
m
p
u
ti
n
g
:
A
c
a
se
stu
d
y
i
n
fa
c
e
re
c
o
g
n
it
i
o
n
,
”
I
n
2
8
t
h
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
A
d
v
a
n
c
e
d
In
fo
rm
a
ti
o
n
Ne
tw
o
rk
in
g
a
n
d
Ap
p
li
c
a
ti
o
n
s W
o
rk
sh
o
p
s (W
AINA
)
IEE
E
,
2
0
1
4
,
p
p
.
7
3
-
7
8
.
[2
]
Din
h
,
H.T
.
,
Lee
,
C
.
,
Ni
y
a
to
,
D.
a
n
d
Wan
g
,
P
.
,
“
A
su
rv
e
y
o
f
m
o
b
il
e
c
lo
u
d
c
o
m
p
u
ti
n
g
:
a
rc
h
it
e
c
tu
re
,
a
p
p
li
c
a
ti
o
n
s,
a
n
d
a
p
p
ro
a
c
h
e
s
,
”
W
ire
les
s c
o
mm
u
n
ic
a
ti
o
n
s a
n
d
mo
b
il
e
c
o
mp
u
ti
n
g
,
v
o
l
.
13
,
n
o
.
1
8
,
p
p
.
1
5
8
7
-
1
6
1
1
,
2
0
1
3
.
[3
]
G
e
o
rg
e
sc
u
,
D.,
“
A
r
e
a
l
-
ti
m
e
fa
c
e
re
c
o
g
n
it
io
n
sy
ste
m
u
sin
g
e
i
g
e
n
fa
c
e
s
,
”
J
o
u
rn
a
l
o
f
M
o
b
i
le,
Emb
e
d
d
e
d
a
n
d
Distrib
u
ted
S
y
ste
ms
,
v
o
l.
3
,
n
o
.
4
,
p
p
.
1
9
3
-
204
,
2
0
1
1
.
[4
]
G
il
b
e
rt,
J.M
.
a
n
d
Ya
n
g
,
W.
,
“
A
re
a
l
-
ti
m
e
fa
c
e
r
e
c
o
g
n
it
io
n
sy
ste
m
u
sin
g
c
u
st
o
m
VLS
I
h
a
r
d
wa
re
,
”
In
Co
mp
u
ter
Arc
h
it
e
c
tu
re
s fo
r M
a
c
h
i
n
e
Per
c
e
p
ti
o
n
,
Pro
c
e
e
d
i
n
g
s
IE
EE
,
1
9
9
3
,
p
p
.
5
8
-
6
6
.
[5
]
Ha
,
K.,
P
il
lai,
P
.
,
Lew
is,
G
.
,
S
im
a
n
ta,
S
.
,
Cli
n
c
h
,
S
.
,
Da
v
ies
,
N.
a
n
d
S
a
ty
a
n
a
ra
y
a
n
a
n
,
M
.
,
“
Th
e
im
p
a
c
t
o
f
m
o
b
il
e
m
u
lt
ime
d
ia
a
p
p
l
ica
ti
o
n
s
o
n
d
a
ta
c
e
n
ter
c
o
n
so
li
d
a
ti
o
n
,
”
I
n
IEE
E
I
n
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
Cl
o
u
d
E
n
g
in
e
e
rin
g
(IC2
E)
,
2
0
1
3
,
p
p
.
1
6
6
-
1
7
6
.
[6
]
Lu
,
J.,
P
lata
n
i
o
ti
s,
K.
N.
a
n
d
Ve
n
e
tsa
n
o
p
o
u
lo
s,
A.
N.
,
“
F
a
c
e
re
c
o
g
n
it
io
n
u
sin
g
LDA
-
b
a
se
d
a
lg
o
ri
th
m
s,”
in
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Ne
u
ra
l
Ne
two
rk
s
,
v
o
l.
14
,
n
o
.
1
,
p
p
.
1
9
5
-
2
0
0
,
2
0
0
3
.
[7
]
Lu
,
X.
,
“
Im
a
g
e
a
n
a
ly
sis fo
r
fa
c
e
re
c
o
g
n
it
i
o
n
,
”
p
e
rso
n
a
l
n
o
tes
,
M
ic
h
ig
a
n
S
tate
Un
iv
e
rsit
y
,
(Eas
t
Lan
si
n
g
,
M
I),
2
0
0
3
.
[8
]
Ora
v
e
c
,
M
.
,
P
a
v
lo
v
ičo
v
á
,
J.,
M
a
z
a
n
e
c
,
J.,
Om
e
li
n
a
,
Ľ.
,
F
é
d
e
r,
M
.
a
n
d
Ba
n
,
J.,
“
Eff
icie
n
c
y
o
f
re
c
o
g
n
it
i
o
n
m
e
th
o
d
s
fo
r
sin
g
le
sa
m
p
le
p
e
r
p
e
rso
n
b
a
se
d
fa
c
e
re
c
o
g
n
it
i
o
n
,
”
in
Rev
iews
,
Refin
e
me
n
ts
a
n
d
Ne
w
I
d
e
a
s
in
Fa
c
e
Rec
o
g
n
it
i
o
n
,
20
11
,
p
p
.
1
8
1
-
2
0
6
.
[9
]
P
o
we
rs,
N.,
Alli
n
g
,
A.
,
Os
o
li
n
sk
y
,
K.,
S
o
y
a
ta,
T
.
,
Z
h
u
,
M
.
,
Wa
n
g
,
H.,
Ba
,
H.,
He
in
z
e
lma
n
,
W.
,
S
h
i,
J.
a
n
d
Kw
o
n
,
M
.
,
“
Th
e
c
lo
u
d
let
a
c
c
e
lera
to
r:
b
ri
n
g
i
n
g
m
o
b
il
e
-
c
lo
u
d
fa
c
e
re
c
o
g
n
it
io
n
i
n
t
o
re
a
l
-
ti
m
e
,
”
IEE
E
Glo
b
e
c
o
m
W
o
rk
sh
o
p
s
(GC W
k
sh
p
s),
2
0
1
1
,
p
p
.
1
-
7.
[1
0
]
S
a
n
k
e
t
De
ss
a
i
a
n
d
K.
Ra
m
a
k
rish
n
a
,
“
Im
p
lem
e
n
tati
o
n
o
f
v
i
d
e
o
c
a
p
t
u
re
a
n
d
p
lay
b
a
c
k
in
m
o
b
il
e
sy
ste
m
s
,
”
In
ter
n
a
t
io
n
a
l
J
o
u
rn
a
l
o
f
Rec
o
n
fi
g
u
ra
b
le
a
n
d
Em
b
e
d
d
e
d
S
y
ste
ms
,
v
o
l
.
2
,
no
.
3
,
p
p
.
1
0
6
-
1
1
5
,
2
0
1
3
.
[1
1
]
S
h
a
sh
id
h
a
r
H.G
.
,
S
a
n
k
e
t
De
ss
a
i
a
n
d
S
h
i
lp
a
Ch
a
u
d
h
a
ri,
“
De
sig
n
o
f
se
c
u
re
tran
sm
issio
n
o
f
m
u
lt
ime
d
ia
d
a
ta
u
sin
g
S
RTP
o
n
l
in
u
x
p
latfo
rm
,
”
I
n
ter
n
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
Rec
o
n
f
ig
u
ra
b
l
e
a
n
d
Emb
e
d
d
e
d
S
y
ste
ms
,
v
o
l
4
,
no.
2
,
p
p
.
7
1
-
81
,
2
0
1
3
.
[1
2
]
S
o
y
a
ta,
T
.
,
M
u
ra
lee
d
h
a
ra
n
,
R.
,
F
u
n
a
i,
C.
,
Kw
o
n
,
M
.
a
n
d
He
in
z
e
lma
n
,
W,
“
Clo
u
d
-
v
isi
o
n
:
Re
a
l
-
ti
m
e
fa
c
e
re
c
o
g
n
it
io
n
u
sin
g
a
m
o
b
il
e
-
c
lo
u
d
let
-
c
l
o
u
d
a
c
c
e
lera
ti
o
n
a
rc
h
it
e
c
tu
re
,
”
I
n
IEE
E
sy
m
p
o
siu
m
o
n
c
o
m
p
u
ter
s
a
n
d
c
o
mm
u
n
ic
a
ti
o
n
s
(IS
CC),
2
0
1
2
,
p
p
.
0
0
0
0
5
9
-
0
0
0
0
6
6
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Reco
n
f
ig
u
r
a
b
le
&
E
m
b
ed
d
ed
Sy
s
t
I
SS
N:
2089
-
4
8
6
4
Desig
n
a
n
d
d
ev
elo
p
men
t
o
f a
p
a
r
a
lleliz
ed
a
lg
o
r
ith
m
fo
r
fa
c
e
r
ec
o
g
n
itio
n
in
…
(
K
.
N
.
B
h
a
t
t
)
55
[1
3
]
S
to
jme
n
o
v
ic,
M
.
,
“
M
o
b
il
e
c
l
o
u
d
c
o
m
p
u
ti
n
g
fo
r
b
i
o
m
e
tri
c
a
p
p
l
ica
ti
o
n
s,”
I
n
1
5
t
h
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Ne
two
rk
-
Ba
se
d
In
f
o
rm
a
ti
o
n
S
y
ste
ms
,
2
0
1
2
,
p
p
.
6
5
4
-
6
5
9
.
[1
4
]
S
to
jme
n
o
v
ic,
I
,
“
F
o
g
c
o
m
p
u
t
in
g
:
a
c
lo
u
d
to
th
e
g
r
o
u
n
d
s
u
p
p
o
rt
fo
r
sm
a
rt
th
in
g
s
a
n
d
m
a
c
h
in
e
-
to
-
m
a
c
h
in
e
n
e
two
rk
s,”
I
n
A
u
stra
l
a
sia
n
tele
c
o
mm
u
n
ica
ti
o
n
n
e
two
rk
s
a
n
d
a
p
p
l
ica
ti
o
n
s c
o
n
fer
e
n
c
e
(AT
NAC)
,
2
0
1
4
,
p
p
.
1
1
7
-
1
2
2
.
[1
5
]
Ya
n
,
Y.
a
n
d
Hu
a
n
g
,
L,
“
Larg
e
-
sc
a
le i
m
a
g
e
p
ro
c
e
ss
in
g
re
se
a
rc
h
c
lo
u
d
,
”
Cl
o
u
d
C
o
mp
u
ti
n
g
,
2
0
1
4
,
p
p
.
8
8
-
93.
[1
6
]
Ya
n
g
,
J.
,
Z
h
a
n
g
,
D.
a
n
d
Ya
n
g
,
J
.
Y,
“
Is
ICA
si
g
n
ifi
c
a
n
t
ly
b
e
tt
e
r
t
h
a
n
P
CA
fo
r
fa
c
e
re
c
o
g
n
i
ti
o
n
,
”
In
T
e
n
t
h
I
EE
E
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
C
o
mp
u
ter
Vi
si
o
n
(ICCV'0
5
)
,
vol
.
1
,
2
0
0
5
,
p
p
.
1
9
8
-
2
0
3
.
[1
7
]
Ya
n
g
,
J.,
Y
u
,
H.
a
n
d
Ku
n
z
,
W.
,
“
An
e
fficie
n
t
LDA
a
lg
o
rit
h
m
fo
r
fa
c
e
re
c
o
g
n
it
io
n
,
”
In
Pr
o
c
e
e
d
in
g
s
o
f
t
h
e
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
A
u
t
o
ma
ti
o
n
,
R
o
b
o
ti
c
s,
a
n
d
Co
m
p
u
ter
Vi
sio
n
(ICAR
CV
2
0
0
0
)
,
2
0
0
0
,
p
p
.
3
4
-
4
7
.
[1
8
]
Zh
a
o
,
W.
a
n
d
Ch
e
ll
a
p
p
a
,
R.
,
“
Im
a
g
e
-
b
a
se
d
fa
c
e
re
c
o
g
n
it
io
n
:
Iss
u
e
s
a
n
d
m
e
th
o
d
s
,
”
Op
ti
c
a
l
e
n
g
in
e
e
r
in
g
-
Ne
w
Y
o
rk
-
ma
rc
e
l
d
e
k
k
e
r in
c
o
rp
o
ra
te
d
,
v
o
l
.
78
,
p
p
.
3
7
5
-
4
0
2
,
2
0
0
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.