I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
p
ute
r
E
ng
in
ee
ring
(
I
J
E
CE
)
Vo
l.
6
,
No
.
6
,
Decem
b
er
201
6
,
p
p
.
282
8
~
2
8
3
5
I
SS
N:
2088
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v
6i
6
.
1
2
7
2
5
2828
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
jo
u
r
n
a
l.c
o
m/o
n
lin
e/in
d
ex
.
p
h
p
/I
JE
C
E
Face De
tectio
n
i
n
a
M
ix
ed
-
Subje
ct
Do
cu
m
ent
L
ho
us
s
a
ine B
o
uh
o
u
1
,
Ra
chi
d E
l A
y
a
chi
2
,
M
o
ha
m
e
d B
a
s
la
m
3
,
M
o
ha
m
ed
O
u
k
e
s
s
o
u
4
1,
2,
3
Co
m
p
u
ter
S
c
ien
c
e
s De
p
a
rte
m
e
n
t,
F
a
c
u
lt
y
o
f
S
c
ien
c
e
a
n
d
T
e
c
h
n
o
lo
g
y
,
S
u
lt
a
n
M
o
u
lay
S
li
m
a
n
e
U
n
iv
e
rsity
,
M
o
ro
c
c
o
4
M
a
th
e
m
a
ti
c
s D
e
p
a
rte
m
e
n
t,
F
a
c
u
lt
y
o
f
S
c
ien
c
e
a
n
d
T
e
c
h
n
o
lo
g
y
,
S
u
lt
a
n
M
o
u
lay
S
li
m
a
n
e
Un
iv
e
rsit
y
,
M
o
ro
c
c
o
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
u
l
22
,
2
0
1
6
R
ev
i
s
ed
No
v
1
1
,
2
0
1
6
A
cc
ep
ted
No
v
25
,
2
0
1
6
Be
f
o
re
y
o
u
re
c
o
g
n
ize
a
n
y
o
n
e
,
it
i
s
e
ss
e
n
ti
a
l
to
i
d
e
n
ti
f
y
v
a
rio
u
s
c
h
a
ra
c
teristics
v
a
riatio
n
s f
ro
m
o
n
e
p
e
rso
n
to
a
n
o
th
e
r.
Am
o
n
g
o
f
th
is
c
h
a
ra
c
teristics
,
w
e
h
a
v
e
th
o
se
re
latin
g
to
t
h
e
f
a
c
e
.
No
w
a
d
a
y
s th
e
d
e
tec
ti
o
n
o
f
sk
in
re
g
io
n
s in
a
n
im
a
g
e
h
a
s
b
e
c
o
m
e
a
n
i
m
p
o
rtan
t
re
se
a
r
c
h
to
p
ic
f
o
r
th
e
lo
c
a
ti
o
n
o
f
a
fa
c
e
in
th
e
im
a
g
e
.
In
th
is
re
se
a
r
c
h
stu
d
y
,
u
n
li
k
e
p
re
v
io
u
s
re
se
a
rc
h
stu
d
ies
re
late
d
to
th
is
t
o
p
ic
w
h
ich
h
a
v
e
f
o
c
u
se
d
o
n
im
a
g
e
s
in
p
u
ts
d
a
ta
f
a
c
e
s
,
w
e
a
re
m
o
re
in
tere
ste
d
to
th
e
f
ield
s
f
a
c
e
d
e
tec
ti
o
n
in
m
ix
e
d
-
su
b
jec
t
d
o
c
u
m
e
n
ts
(tex
t+
i
m
a
g
e
s).
T
h
e
fa
c
e
d
e
te
c
ti
o
n
sy
ste
m
d
e
v
e
lo
p
e
d
is
b
a
se
d
o
n
th
e
h
y
b
rid
m
e
th
o
d
t
o
d
isti
n
g
u
is
h
tw
o
c
a
teg
o
ries
o
f
o
b
jec
ts
f
ro
m
th
e
m
ix
e
d
d
o
c
u
m
e
n
t.
T
h
e
f
irst
c
a
te
g
o
r
y
is
a
ll
th
a
t
is
tex
t
o
r
i
m
a
g
e
s
c
o
n
tain
in
g
f
ig
u
re
s
h
a
v
in
g
n
o
sk
in
c
o
lo
r,
a
n
d
th
e
se
c
o
n
d
c
a
teg
o
ry
is
a
n
y
f
ig
u
re
w
it
h
th
e
s
a
m
e
c
o
lo
r
a
s
th
e
sk
in
.
In
th
e
se
c
o
n
d
p
h
a
se
th
e
d
e
tec
ti
o
n
sy
ste
m
is
b
a
se
d
o
n
T
e
m
p
late
M
a
tch
in
g
m
e
th
o
d
t
o
d
isti
n
g
u
ish
a
m
o
n
g
th
e
f
ig
u
re
s
o
f
th
e
se
c
o
n
d
c
a
teg
o
r
y
o
n
ly
th
o
se
th
a
t
c
o
n
tain
f
a
c
e
s
to
d
e
tec
t
th
e
m
.
T
o
v
a
li
d
a
te
th
is
stu
d
y
,
th
e
s
y
ste
m
d
e
v
e
lo
p
e
d
is
tes
ted
o
n
t
h
e
v
a
rio
u
s
d
o
c
u
m
e
n
ts
w
h
ich
in
c
lu
d
i
n
g
tex
t
a
n
d
im
a
g
e
.
K
ey
w
o
r
d
:
C
o
lo
r
o
f
s
k
i
n
Face
d
etec
tio
n
Mix
ed
-
s
u
b
j
ec
t d
o
cu
m
e
n
t
T
em
p
late
m
atc
h
i
n
g
Co
p
y
rig
h
t
©
2
0
1
6
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
:
R
ac
h
id
E
l
Ay
ac
h
i
,
C
o
m
p
u
ter
Scien
ce
s
Dep
ar
te
m
en
t,
Facu
lt
y
o
f
Sc
ien
ce
a
n
d
T
ec
h
n
o
lo
g
y
,
Su
lta
n
Mo
u
la
y
Sl
i
m
a
n
e
U
n
i
v
e
r
s
it
y
,
B
en
i M
ella
l,
Mo
r
o
cc
o
.
E
m
ail:
r
ac
h
id
.
ela
y
ac
h
i@
u
s
m
s
.
m
a
1.
I
NT
RO
D
UCT
I
O
N
Am
o
n
g
t
h
e
c
u
r
r
en
t
p
r
o
b
le
m
s
,
th
e
k
n
o
w
led
g
e
to
d
eter
m
in
e
th
e
id
e
n
tit
y
o
f
a
p
er
s
o
n
a
u
to
m
atica
ll
y
.
P
ass
w
o
r
d
s
an
d
I
D
ca
r
d
s
ca
n
n
o
t
p
r
o
v
id
e
v
ital
au
th
e
n
ticatio
n
f
u
n
c
tio
n
s
s
u
c
h
as
d
etec
tio
n
o
f
m
u
ltip
le
en
tr
ie
s
.
Fo
r
ex
a
m
p
le,
u
s
er
s
ca
n
ea
s
il
y
d
en
y
t
h
e
u
s
e
o
f
a
s
er
v
ice
b
y
clai
m
in
g
th
at
t
h
eir
p
as
s
w
o
r
d
h
a
s
b
ee
n
s
to
le
n
o
r
g
u
e
s
s
ed
.
I
n
d
iv
id
u
al
s
ca
n
also
h
id
e
th
eir
tr
u
e
id
e
n
tit
y
b
y
p
r
es
en
ti
n
g
f
als
if
ied
id
en
tit
y
d
o
cu
m
en
ts
d
u
p
licate
s
.
T
h
er
ef
o
r
e,
it
b
ec
o
m
e
s
in
cr
ea
s
in
g
l
y
e
v
id
en
t
t
h
at
t
h
ese
m
ec
h
an
i
s
m
s
ar
e
n
o
t
s
u
f
f
i
c
ien
t
to
d
eter
m
i
n
e
r
eliab
l
y
t
h
e
id
en
t
it
y
o
f
a
p
er
s
o
n
a
n
d
a
m
o
r
e
r
o
b
u
s
t
m
ec
h
a
n
is
m
f
o
r
id
en
ti
f
icatio
n
b
ased
o
n
s
o
m
et
h
i
n
g
th
at
y
o
u
ar
e,
ie
b
io
m
etr
ics,
is
m
o
r
e
t
h
a
n
n
ec
es
s
ar
y
.
B
io
m
e
tr
ic
f
ea
t
u
r
es
ar
e
an
alter
n
ati
v
e
to
f
o
r
m
er
i
d
en
tit
y
v
er
i
f
icatio
n
m
ea
n
s
.
T
h
e
ad
v
a
n
ta
g
e
o
f
t
h
ese
b
io
m
etr
ic
is
to
b
e
p
r
esen
t
i
n
all
p
eo
p
le
to
id
en
tify
.
Face
d
etec
tio
n
in
a
h
y
p
o
t
h
esi
s
w
it
h
o
u
t
i
m
a
g
e
i
s
a
v
er
y
d
i
f
f
icu
lt
p
r
o
b
le
m
b
ec
au
s
e
o
f
t
h
e
g
r
ea
t
v
ar
iab
ilit
y
o
f
t
h
e
s
h
ap
e
a
n
d
co
lo
r
to
d
etec
t (
Or
ien
tatio
n
a
n
d
an
y
s
iz
e,
v
ar
iab
le
lig
h
tin
g
,
d
i
f
f
er
e
n
t s
k
in
co
lo
r
)
.
2.
B
I
O
M
E
T
RICS
B
io
m
etr
ic
s
i
s
a
n
alter
n
ati
v
e
t
o
th
e
o
ld
m
et
h
o
d
s
o
f
id
en
t
if
ic
atio
n
.
I
t
co
n
s
i
s
ts
o
f
id
e
n
ti
f
y
in
g
a
p
er
s
o
n
f
r
o
m
p
h
y
s
ical
o
r
b
eh
av
io
r
al
ch
ar
ac
ter
is
t
ics
[
1
]
.
Face
,
f
i
n
g
er
p
r
in
t,
ir
is
,
e
tc.
ar
e
e
x
a
m
p
les
o
f
p
h
y
s
ical
ch
ar
ac
ter
is
tic
s
.
Vo
ice,
w
r
i
tin
g
,
t
y
p
in
g
r
h
y
th
m
o
n
a
k
e
y
b
o
ar
d
,
etc.
ar
e
b
eh
av
io
r
al
ch
ar
ac
ter
is
tics
.
T
h
ese
f
ea
t
u
r
es,
as
t
h
e
y
ar
e
i
n
n
a
te
o
r
ac
q
u
ir
ed
d
ig
ital
f
i
n
g
er
p
r
in
t
a
s
s
i
g
n
atu
r
e,
ar
e
attac
h
ed
to
e
ac
h
i
n
d
iv
id
u
al
a
n
d
th
er
ef
o
r
e
d
o
es n
o
t s
u
f
f
er
f
r
o
m
th
e
w
ea
k
n
e
s
s
es o
f
th
e
m
et
h
o
d
s
b
ased
o
n
k
n
o
w
led
g
e
o
r
p
o
s
s
ess
io
n
[
2
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2088
-
8708
F
a
ce
Dete
ctio
n
in
a
Mixed
-
S
u
b
ject
Do
cu
men
t
(
L
h
o
u
s
s
a
in
e
B
o
u
h
o
u
)
2829
2
.
1
.
Arc
hite
ct
ure
O
f
A
B
io
m
et
ric
Sy
s
t
e
m
A
b
io
m
etr
ic
s
y
s
te
m
i
s
o
f
te
n
co
m
p
o
s
ed
o
f
t
h
r
ee
m
o
d
u
les
:
l
ea
r
n
in
g
,
r
ec
o
g
n
i
tio
n
a
n
d
ad
ap
tatio
n
[
3
-
4
]
.
Du
r
in
g
lear
n
in
g
,
t
h
e
s
y
s
te
m
w
i
ll
ac
q
u
ir
e
o
n
e
o
r
m
o
r
e
b
i
o
m
etr
ics
w
i
ll
b
e
u
s
ed
to
b
u
ild
a
m
o
d
el
o
f
t
h
e
in
d
i
v
id
u
al.
T
h
i
s
r
ef
er
en
ce
m
o
d
el
w
il
l
s
er
v
e
a
s
a
p
o
in
t
o
f
co
m
p
ar
i
s
o
n
d
u
r
in
g
r
ec
o
g
n
itio
n
.
T
h
e
m
o
d
el
w
i
ll b
e
r
ee
v
alu
ated
a
f
te
r
ea
ch
u
s
e
w
it
h
t
h
e
ad
ap
ter
m
o
d
u
le.
2
.
1
.
1
.
L
ea
rning
M
o
du
le
Du
r
in
g
th
e
p
h
ase
o
f
lear
n
i
n
g
,
b
io
m
e
tr
ic
ch
ar
ac
ter
i
s
tic
i
s
m
ea
s
u
r
ed
b
y
a
s
e
n
s
o
r
;
t
h
er
e
is
tal
k
o
f
ac
q
u
is
itio
n
o
r
ca
p
tu
r
e.
I
n
g
en
er
al,
th
is
ca
p
tu
r
e
i
s
n
o
t
d
ir
ec
tl
y
s
to
r
ed
an
d
tr
an
s
f
o
r
m
atio
n
s
ar
e
ap
p
lied
to
it.
I
n
d
ee
d
,
th
e
s
ig
n
a
l
co
n
tain
s
u
s
ele
s
s
i
n
f
o
r
m
a
tio
n
to
th
e
r
ec
o
g
n
itio
n
a
n
d
o
n
l
y
t
h
e
r
elev
an
t
p
ar
a
m
eter
s
ar
e
ex
tr
ac
ted
.
T
h
e
m
o
d
el
i
s
a
co
m
p
ac
t
r
ep
r
esen
tat
io
n
o
f
t
h
e
s
ig
n
a
l
w
h
ic
h
f
ac
ilit
ate
s
t
h
e
r
ec
o
g
n
itio
n
p
h
a
s
e,
b
u
t
also
to
r
ed
u
ce
th
e
a
m
o
u
n
t o
f
d
ata
to
s
to
r
e.
2
.
1
.
2
.
Rec
o
g
nitio
n
M
o
du
le
Du
r
in
g
t
h
e
p
h
ase
o
f
r
ec
o
g
n
it
i
o
n
,
th
e
b
io
m
e
tr
ic
ch
ar
ac
ter
is
ti
c
is
m
ea
s
u
r
ed
an
d
a
s
et
o
f
p
ar
a
m
eter
s
is
ex
tr
ac
ted
as
w
h
e
n
lear
n
i
n
g
.
T
h
e
s
e
n
s
o
r
u
s
ed
m
u
s
t
h
a
v
e
p
r
o
p
er
ties
as
clo
s
e
as
p
o
s
s
ib
le
t
h
e
s
en
s
o
r
u
s
ed
d
u
r
i
n
g
th
e
tr
ain
i
n
g
p
h
ase.
I
f
b
o
th
s
e
n
s
o
r
s
h
av
e
d
if
f
er
e
n
t
p
r
o
p
er
ties
to
o
,
it
w
il
l
g
e
n
er
all
y
ap
p
l
y
a
s
er
ies
o
f
ad
d
itio
n
al
p
r
e
-
tr
ea
t
m
e
n
ts
to
r
ed
u
ce
p
er
f
o
r
m
a
n
ce
d
eg
r
ad
atio
n
.
Fo
llo
w
i
n
g
t
h
e
r
ec
o
g
n
itio
n
w
i
ll
b
e
d
if
f
er
en
t
ac
co
r
d
in
g
to
th
e
p
r
o
ce
d
u
r
e
o
f
th
e
s
y
s
te
m
.
2
.
1
.
3
.
Ada
pta
t
io
n
M
o
du
le
Du
r
in
g
t
h
e
lear
n
i
n
g
p
h
a
s
e,
th
e
b
io
m
etr
ic
s
y
s
te
m
o
f
ten
ca
tc
h
a
f
e
w
i
n
s
tan
ce
s
o
f
th
e
s
a
m
e
at
tr
ib
u
te
s
o
as
to
li
m
it
t
h
e
in
co
n
v
e
n
ie
n
ce
f
o
r
th
e
u
s
er
.
I
t
is
d
if
f
ic
u
lt
en
o
u
g
h
to
b
u
ild
a
g
en
er
al
m
o
d
el
to
d
escr
ib
e
all
th
e
p
o
s
s
ib
le
v
ar
iatio
n
s
o
f
th
i
s
at
tr
ib
u
te.
I
n
ad
d
itio
n
,
t
h
e
c
h
ar
ac
ter
is
tics
o
f
th
i
s
b
io
m
etr
ics
an
d
it
s
ac
q
u
is
it
io
n
co
n
d
itio
n
s
m
a
y
v
ar
y
.
A
d
ap
tati
o
n
is
t
h
er
ef
o
r
e
n
ec
ess
ar
y
to
m
ain
tai
n
a
n
d
i
m
p
r
o
v
e
t
h
e
p
er
f
o
r
m
an
ce
o
f
a
s
y
s
te
m
u
s
e
a
f
ter
u
s
e.
T
h
e
ad
j
u
s
t
m
e
n
t
ca
n
b
e
d
o
n
e
i
n
s
u
p
er
v
is
ed
m
o
d
e
o
r
u
n
s
u
p
er
v
i
s
ed
b
u
t
th
e
s
ec
o
n
d
m
o
d
e
is
b
y
f
ar
th
e
m
o
s
t u
s
ef
u
l i
n
p
r
ac
tice
[
5
]
.
Am
o
n
g
t
h
e
b
io
m
etr
ic
tec
h
n
iq
u
es,
t
h
er
e
ar
e
f
ac
ial
r
ec
o
g
n
itio
n
.
I
n
ev
er
y
d
a
y
li
f
e
e
v
er
y
o
n
e
i
d
en
tifie
s
u
s
th
r
o
u
g
h
o
u
t t
h
e
d
a
y
d
i
f
f
er
en
t
f
ac
es.
So
w
h
e
n
w
e
en
co
u
n
ter
a
p
er
s
o
n
,
o
u
r
b
r
ain
w
ill
s
ea
r
ch
o
u
r
m
e
m
o
r
y
a
n
d
s
e
e
if
t
h
at
p
er
s
o
n
is
li
s
ted
o
r
n
o
t.
I
n
r
ec
o
g
n
itio
n
o
f
2
D
f
ac
e,
s
ev
er
al
m
et
h
o
d
s
h
a
v
e
b
ee
n
d
ev
elo
p
ed
.
Ho
w
e
v
er
,
th
e
y
h
av
e
a
n
u
m
b
er
o
f
li
m
ita
tio
n
s
r
elate
d
to
th
e
o
r
ien
tatio
n
o
f
t
h
e
f
ac
e
o
r
la
y
i
n
g
,
lig
h
ti
n
g
,
f
ac
ial
e
x
p
r
ess
io
n
,
o
cc
lu
s
io
n
s
,
etc.
I
n
r
ec
en
t
y
ea
r
s
,
w
e
tal
k
m
o
r
e
an
d
m
o
r
e
3
D
f
ac
e
r
ec
o
g
n
itio
n
tech
n
iq
u
e
s
as
a
n
alter
n
ati
v
e
s
o
lu
tio
n
to
s
o
lv
e
th
e
p
r
o
b
lem
s
m
en
tio
n
ed
ab
o
v
e.
I
n
d
ee
d
,
th
e
w
ea
lt
h
o
f
in
f
o
r
m
a
tio
n
p
r
o
v
id
ed
b
y
3
D
m
ea
s
u
r
e
m
e
n
ts
allo
w
s
to
r
ec
o
n
s
tr
u
ct
th
e
th
r
ee
-
d
i
m
e
n
s
io
n
al
s
h
ap
e
o
f
t
h
e
f
ac
e.
T
h
is
t
y
p
e
o
f
f
ac
ial
r
ep
r
esen
tat
io
n
i
s
i
n
v
ar
ian
t
to
ch
a
n
g
es
in
ill
u
m
in
at
io
n
an
d
p
o
s
e
[
6
]
.
Facial
r
ec
o
g
n
i
tio
n
i
s
a
co
m
m
o
n
a
n
d
p
o
p
u
lar
tech
n
iq
u
e.
I
t
is
th
e
m
o
s
t
ac
ce
p
tab
le
b
ec
au
s
e
it
co
r
r
esp
o
n
d
s
to
w
h
a
t h
u
m
an
s
u
s
e
in
v
i
s
u
al
in
ter
ac
tio
n
; a
n
d
co
m
p
ar
i
n
g
it to
o
th
er
m
et
h
o
d
s
,
f
ac
ial
r
ec
o
g
n
itio
n
i
s
m
o
r
e
ad
v
a
n
ta
g
eo
u
s
,
o
n
t
h
e
o
n
e
h
a
n
d
it
is
a
n
o
n
-
i
n
tr
u
s
i
v
e
m
e
th
o
d
,
th
at
is
to
s
a
y
it
d
o
es
n
o
t
r
eq
u
ir
e
th
e
co
o
p
er
atio
n
o
f
th
e
s
u
b
j
ec
t
(
o
b
s
er
v
in
g
in
d
i
v
id
u
a
ls
r
e
m
o
te
)
,
an
d
o
n
th
e
o
th
er
h
an
d
t
h
e
s
en
s
o
r
s
u
s
ed
ar
e
in
ex
p
e
n
s
i
v
e
(
s
in
g
le
ca
m
er
a)
u
n
li
k
e
t
h
e
f
in
g
er
p
r
in
t
a
n
d
ir
is
w
h
er
e
t
h
e
s
u
b
j
ec
t
w
ill
b
e
v
er
y
clo
s
e
to
th
e
s
en
s
o
r
an
d
w
il
l
co
o
p
er
ate
in
th
e
d
ev
elo
p
m
e
n
t
o
f
p
ict
u
r
e
w
it
h
o
u
t
f
o
r
g
ettin
g
th
e
co
s
t
o
f
t
h
e
eq
u
i
p
m
e
n
t
n
ec
ess
ar
y
f
o
r
th
e
ac
q
u
i
s
itio
n
(
e
x
p
en
s
iv
e
s
p
e
cial
eq
u
ip
m
e
n
t)
.
2
.
2
.
F
a
ce
Det
ec
t
io
n:
St
a
t
e
O
f
T
he
Art
An
au
to
m
atic
f
ac
e
r
ec
o
g
n
it
io
n
s
y
s
te
m
co
n
s
is
t
s
o
f
t
h
r
e
e
s
u
b
s
y
s
te
m
s
:
f
ac
e
d
etec
tio
n
,
f
ea
t
u
r
e
ex
tr
ac
tio
n
a
n
d
f
ac
e
r
ec
o
g
n
itio
n
.
Face
d
etec
tio
n
i
s
th
e
p
r
o
b
lem
o
f
lo
ca
ti
n
g
f
ac
e
s
in
a
i
n
p
u
t
i
m
a
g
e.
T
h
e
in
ter
est
o
f
f
ac
ial
lo
ca
lizatio
n
g
o
es
b
e
y
o
n
d
th
e
i
m
p
le
m
en
ta
tio
n
o
f
t
h
e
p
r
esen
t
w
o
r
k
.
I
ts
u
t
ilit
y
i
s
m
a
n
i
f
ested
in
v
ar
io
u
s
f
ield
s
r
an
g
i
n
g
f
r
o
m
v
id
eo
s
u
r
v
eilla
n
ce
to
in
ter
ac
ti
v
e
g
a
m
e.
T
h
e
f
ir
s
t
d
i
f
f
icu
l
ties
en
co
u
n
t
er
ed
b
y
h
ar
n
es
s
in
g
m
et
h
o
d
s
to
d
etec
t
f
ac
e
s
ar
e
t
h
e
v
ar
iatio
n
s
i
n
p
o
s
e
(
p
r
o
f
ile
v
ie
w
,
f
r
o
n
t)
,
ex
p
r
es
s
io
n
,
f
ac
e
r
o
tatio
n
,
a
g
e
a
n
d
illu
m
i
n
atio
n
.
T
h
is
t
y
p
e
o
f
d
if
f
ic
u
lt
y
ca
n
b
e
o
v
er
co
m
e
b
y
p
r
etr
ea
t
m
e
n
t
s
ta
n
d
ar
d
s
an
d
illu
m
i
n
atio
n
co
m
p
e
n
s
at
io
n
.
I
n
a
co
n
tr
o
lled
en
v
ir
o
n
m
e
n
t,
s
ettin
g
s
s
u
c
h
as
th
e
b
ac
k
g
r
o
u
n
d
,
th
e
d
ir
ec
tio
n
an
d
in
ten
s
it
y
o
f
th
e
lig
h
t
s
o
u
r
ce
,
th
e
an
g
le
o
f
th
e
s
h
o
t,
a
r
e
p
ar
am
eter
s
t
h
at
ca
n
b
e
co
n
t
r
o
lled
.
Yan
g
et
al.
,
[
7
]
p
r
o
p
o
s
ed
a
cla
s
s
i
f
icatio
n
o
f
m
e
th
o
d
s
o
f
f
ac
ia
l lo
ca
tio
n
:
a.
"
Kn
o
w
led
g
e
-
b
ased
m
et
h
o
d
s
"
.
Kn
o
w
led
g
e
o
f
t
h
e
v
ar
io
u
s
e
le
m
e
n
ts
th
at
co
n
s
t
itu
te
a
f
ac
e
an
d
r
elatio
n
s
b
et
w
ee
n
th
e
m
ar
e
o
f
th
ese
m
e
th
o
d
s
b
asin
g
.
T
h
u
s
,
th
e
r
elati
v
e
p
o
s
itio
n
s
o
f
th
e
v
ar
io
u
s
k
e
y
ele
m
e
n
ts
s
u
c
h
as
th
e
m
o
u
t
h
,
n
o
s
e
an
d
e
y
es
ar
e
th
en
u
s
ed
to
m
ea
s
u
r
es
t
h
e
cl
ass
i
f
icatio
n
'
f
ac
e
'
a
n
d
'n
o
n
-
f
ac
e'
i
n
C
h
ia
n
g
e
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
J
E
C
E
Vo
l.
6
,
No
.
6
,
Decem
b
er
201
6
:
2
8
2
8
–
2
8
3
5
2830
al.
,
[
8
]
.
T
h
e
p
r
o
b
lem
w
it
h
th
i
s
m
et
h
o
d
is
th
a
t
it
is
d
i
f
f
ic
u
lt
to
d
ef
in
e
u
n
iq
u
el
y
a
f
ac
e.
I
f
t
h
e
d
ef
i
n
itio
n
is
t
o
o
d
etailed
,
s
o
m
e
f
ac
e
s
w
ill b
e
r
ates
w
h
ile
i
f
t
h
e
d
escr
ip
tio
n
is
to
o
g
en
er
al
th
e
f
al
s
e
p
o
s
itiv
e
r
ate
s
o
ar
.
b.
"
Featu
r
e
in
v
ar
ia
n
t
ap
p
r
o
ac
h
es
"
.
Fo
r
d
etec
tio
n
,
th
ese
ap
p
r
o
a
ch
es
u
s
e
t
h
e
ele
m
en
t
s
i
n
v
ar
ia
n
t
to
ch
a
n
g
e
s
i
n
illu
m
i
n
atio
n
,
o
r
ien
ta
tio
n
o
r
ex
p
r
ess
io
n
,
s
u
ch
as t
h
e
te
x
t
u
r
e
o
r
th
e
s
ig
n
at
u
r
e
co
lo
r
o
f
th
e
s
k
i
n
.
c.
"T
em
p
late
m
atc
h
i
n
g
m
et
h
o
d
s
"
.
T
h
e
ch
ar
ac
ter
is
tic
p
atter
n
s
o
f
an
en
tire
f
ac
e
o
r
f
ac
e
o
f
s
u
b
p
ar
t
(
m
o
u
th
,
e
y
e,
n
o
s
e)
ar
e
cr
ea
ted
.
T
h
e
l
o
ca
tio
n
is
th
en
m
ad
e
b
ased
o
n
th
e
co
r
r
elatio
n
o
f
th
ese
m
o
d
el
s
w
ith
t
h
e
ca
n
d
id
ates.
d.
"
A
p
p
ea
r
an
ce
-
b
ased
m
eth
o
d
s
"
.
T
h
ese
m
eth
o
d
s
u
s
e
t
h
e
s
a
m
e
p
r
in
cip
le
as t
h
e
o
n
e
p
r
ese
n
ted
in
t
h
e
p
r
ev
io
u
s
p
ar
ag
r
ap
h
b
u
t
ar
e
b
ased
o
n
s
p
ec
if
ic
m
o
d
el
s
.
T
h
ese
m
et
h
o
d
s
h
av
e
t
h
e
ad
v
an
tag
e
o
f
r
u
n
n
i
n
g
v
er
y
q
u
ick
l
y
b
u
t
r
eq
u
ir
e
a
lo
n
g
tr
ain
i
n
g
ti
m
e.
T
h
e
m
et
h
o
d
s
in
t
h
is
cla
s
s
h
av
e
s
h
o
w
n
g
o
o
d
r
esu
lts
a
g
ain
s
t
3
o
th
er
m
et
h
o
d
s
[
9
]
.
T
h
e
f
ac
e
d
etec
tio
n
m
et
h
o
d
s
ca
n
b
e
d
iv
id
ed
in
to
f
o
u
r
ca
te
g
o
r
ies
[
6
]
:
a.
A
p
p
r
o
ch
es
B
ased
o
n
k
n
o
w
led
g
e:
t
h
e
m
ai
n
p
ar
ts
to
w
h
at
th
i
s
m
et
h
o
d
is
i
n
ter
ested
in
th
eir
ch
ar
ac
ter
is
tic
s
as
f
ac
ial
f
ea
t
u
r
es
ar
e
th
e
n
o
s
e,
m
o
u
t
h
an
d
e
y
es.
T
h
ese
m
et
h
o
d
s
ar
e
d
esig
n
ed
p
r
i
m
ar
il
y
f
o
r
f
ac
e
lo
ca
lizatio
n
.
b.
A
p
p
r
o
ch
es
B
ased
"T
em
p
late
-
m
atch
in
g
"
:
T
h
ese
m
eth
o
d
s
to
ca
lcu
late
th
e
co
r
r
elatio
n
b
et
w
ee
n
t
h
e
ca
n
d
id
ate
i
m
a
g
e
an
d
t
h
e
te
m
p
late.
T
h
ey
u
s
e
a
n
alg
o
r
it
h
m
t
h
at
ca
lcu
lates
a
n
d
lu
m
i
n
an
ce
r
atio
s
b
et
w
ee
n
th
e
ar
ea
s
o
f
th
e
f
ac
e
a
n
d
r
etai
n
s
lead
er
s
h
ip
o
f
t
h
ese
r
ep
o
r
ts
(
eg
,
is
th
e
r
eg
io
n
1
li
g
h
ter
o
r
d
ar
k
er
th
a
n
t
h
e
r
eg
io
n
2
)
.
Fig
u
r
e
1
s
h
o
w
s
a
p
r
ed
ef
in
ed
te
m
p
late
co
r
r
esp
o
n
d
in
g
to
2
3
r
elatio
n
s
h
ip
s
.
Fig
u
r
e
1
.
Fac
e
m
o
d
el
co
n
s
is
ts
o
f
1
6
r
eg
io
n
s
(
r
ec
tan
g
les)
as
s
o
ciate
d
w
it
h
r
elatio
n
s
h
ip
s
2
3
(
a
r
r
o
w
s
)
[
1
0
]
d
.
A
p
p
r
o
ch
es
B
ased
o
n
ap
p
ea
r
an
ce
:
Me
th
o
d
s
t
y
p
icall
y
u
s
e
m
a
ch
in
e
lear
n
in
g
tec
h
n
iq
u
e
s
.
T
h
e
y
ar
e
u
s
ed
f
o
r
d
etec
tio
n
.
T
h
e
m
ai
n
id
ea
o
f
th
ese
m
et
h
o
d
s
is
to
co
n
s
id
er
th
at
t
h
e
p
r
o
b
lem
o
f
f
ac
e
d
etec
tio
n
is
a
class
i
f
icatio
n
p
r
o
b
le
m
(
f
ac
e,
n
o
n
-
f
ac
e)
.
On
e
o
f
th
e
m
o
s
t
k
n
o
w
n
ap
p
r
o
ac
h
es
f
ac
e
d
etec
ti
o
n
is
t
h
e
E
i
g
en
f
ac
e,
it
i
n
v
o
lv
e
s
p
r
o
j
ec
tin
g
t
h
e
i
m
a
g
e
i
n
s
p
ac
e
a
n
d
ca
lc
u
lat
e
th
e
E
u
clid
ea
n
d
is
ta
n
ce
b
et
wee
n
t
h
e
i
m
ag
e
an
d
its
p
r
o
j
ec
t
io
n
.
e.
A
p
p
r
o
ch
es
B
ased
o
n
i
n
v
ar
ia
n
t
f
ea
t
u
r
es:
T
h
ese
ap
p
r
o
ac
h
es
ar
e
u
s
ed
p
r
i
m
ar
il
y
f
o
r
f
ac
e
lo
ca
lizatio
n
.
Dev
elo
p
ed
alg
o
r
ith
m
s
ai
m
to
f
i
n
d
th
e
ex
i
s
ti
n
g
s
tr
u
ct
u
r
al
ch
ar
ac
ter
is
tics
ev
e
n
i
f
th
e
p
o
s
e,
th
e
v
ie
w
p
o
i
n
t,
o
r
th
e
li
g
h
t
in
g
co
n
d
itio
n
c
h
a
n
g
e.
T
h
en
t
h
e
y
u
s
e
t
h
ese
in
v
ar
i
ab
le
f
ea
t
u
r
es
to
lo
ca
te
f
ac
es.
W
e
ca
n
m
en
t
io
n
t
w
o
f
a
m
i
lies
o
f
m
et
h
o
d
s
b
elo
n
g
i
n
g
to
th
is
ap
p
r
o
ac
h
:
Me
th
o
d
s
b
ased
o
n
s
k
in
co
lo
r
.
I
n
f
ac
t,
th
e
y
r
ed
u
ce
th
e
s
ea
r
c
h
s
p
ac
e
o
f
th
e
f
ac
e
r
e
g
io
n
i
n
t
h
e
i
m
a
g
e.
I
n
ad
d
itio
n
,
th
e
co
lo
r
o
f
th
e
s
k
i
n
is
a
r
o
b
u
s
t
in
f
o
r
m
at
io
n
f
ac
e
r
o
tatio
n
s
,
s
ca
le
c
h
a
n
g
e
s
an
d
p
ar
tial
o
cc
lu
s
io
n
s
.
Sev
er
al
co
lo
r
s
p
ac
es
ca
n
b
e
u
s
ed
f
o
r
d
etec
tin
g
in
t
h
e
i
m
a
g
e,
th
e
p
i
x
el
s
w
h
ich
h
a
v
e
th
e
co
lo
r
o
f
t
h
e
s
k
i
n
.
T
h
e
d
etec
tio
n
ef
f
ic
ien
c
y
d
ep
en
d
s
m
ai
n
l
y
o
n
t
h
e
co
lo
r
s
p
ac
e
s
elec
ted
.
Am
o
n
g
t
h
e
c
o
l
o
r
s
p
ac
es u
s
ed
,
t
h
e
R
GB
co
lo
r
s
p
ac
e,
it is
to
r
ep
r
esen
t t
h
e
co
lo
r
s
p
ac
e
f
r
o
m
th
r
ee
m
o
n
o
c
h
r
o
m
atic
r
ad
iatio
n
o
f
co
lo
r
s
:
R
ed
-
Gr
ee
n
-
B
lu
e.
Fo
r
ex
am
p
le,
th
e
s
k
i
n
is
cla
s
s
i
f
ied
in
to
th
e
R
GB
co
lo
r
s
p
a
ce
b
y
u
s
in
g
t
h
e
f
o
llo
w
in
g
r
u
le
s
[
1
1
]
:
(
1
)
W
e
p
r
esen
t
h
er
e
a
co
m
p
ar
ati
v
e
s
t
u
d
y
o
n
s
k
in
co
lo
r
d
etec
tio
n
b
y
t
h
r
es
h
o
ld
in
g
ac
co
r
d
in
g
t
o
th
e
R
GB
co
lo
r
s
p
ac
es.
Step
t
h
r
es
h
o
ld
in
g
p
r
o
v
id
e
u
s
b
in
ar
y
i
m
a
g
es
wh
o
s
e
s
k
in
s
ec
tio
n
s
tak
e
t
h
e
h
i
g
h
est
g
r
a
y
s
ca
le
o
r
1
(
w
h
ite)
w
h
ile
o
th
er
s
ec
tio
n
s
(
n
o
t sk
i
n
)
tak
e
t
h
e
lo
w
e
s
t
g
r
a
y
le
v
el
is
0
(
b
lack
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2088
-
8708
F
a
ce
Dete
ctio
n
in
a
Mixed
-
S
u
b
ject
Do
cu
men
t
(
L
h
o
u
s
s
a
in
e
B
o
u
h
o
u
)
2831
2
.
3
.
P
r
o
ble
m
a
t
ic
I
n
g
en
er
al,
a
f
ac
e
r
ec
o
g
n
itio
n
s
y
s
te
m
co
n
s
is
t
o
f
t
h
r
ee
p
ar
ts
:
Dete
ctio
n
,
E
x
tr
ac
tio
n
a
n
d
R
ec
o
g
n
itio
n
.
I
n
th
i
s
p
ap
er
,
o
u
r
g
o
o
d
is
to
m
ak
e
a
co
n
tr
ib
u
t
io
n
i
n
th
e
f
ir
s
t
p
ar
t.
T
h
is
co
n
tr
ib
u
t
io
n
allo
w
s
th
e
d
ev
elo
p
m
e
n
t
o
f
th
e
f
ir
s
t
p
ar
t
o
f
th
e
s
y
s
te
m
f
o
r
d
etec
tin
g
f
ac
es
i
n
m
i
x
ed
d
o
cu
m
en
t.
Fo
r
in
f
o
r
m
at
io
n
,
a
m
i
x
ed
d
o
cu
m
e
n
t
co
n
tain
s
s
e
v
er
al
ite
m
s
,
in
c
lu
d
i
n
g
: te
x
t,
f
i
g
u
r
e
an
d
i
m
a
g
e.
3.
SCE
NA
RIO
3
.
1
.
Def
ini
ng
t
he
O
bje
ct
iv
e
E
nu
m
er
a
t
io
n
a
nd
i
t
s
Co
n
s
equ
ence
s
T
h
e
aim
o
f
t
h
e
ap
p
licatio
n
i
s
to
allo
w
th
e
u
s
er
w
h
ic
h
to
d
etec
t
th
e
f
ac
es
o
f
p
eo
p
le
in
a
m
i
x
ed
d
o
cu
m
en
t
co
n
tain
in
g
tex
t
an
d
im
a
g
e.
A
s
ce
n
ar
io
lik
e
t
h
e
u
s
e
o
f
th
i
s
s
y
s
te
m
i
n
th
r
e
e
p
h
ases
u
s
in
g
t
w
o
m
et
h
o
d
s
.
*
1
s
t
p
h
a
s
e
s
eg
m
e
n
tatio
n
s
k
in
ar
ea
s
.
B
ased
o
n
th
e
H
y
b
r
id
Sk
in
co
l
o
r
m
et
h
o
d
,
th
is
p
h
ase
i
n
v
o
lv
e
s
cu
tti
n
g
s
k
i
n
co
lo
r
r
ep
r
esen
tativ
e
p
ix
el
s
f
r
o
m
a
m
i
x
ed
in
p
u
t d
o
cu
m
e
n
t a
n
d
f
ilt
er
s
in
d
ep
en
d
en
t
s
k
i
n
co
lo
r
p
ix
els.
*
2
n
d
p
h
ase:
s
eg
m
e
n
tatio
n
r
e
g
io
n
s
co
n
tai
n
i
n
g
f
ac
e
B
ase
d
o
n
T
em
p
late
Ma
tc
h
i
n
g
m
et
h
o
d
th
i
s
p
h
a
s
e
f
o
c
u
s
e
s
o
n
r
e
m
o
v
i
n
g
s
k
in
co
lo
r
r
eg
io
n
s
n
o
t
p
r
ese
n
ti
n
g
a
f
ac
e.
*
3
r
d
p
h
ase
: Fac
e
Dete
ctio
n
I
n
t
h
is
p
h
ase
f
ac
e
d
etec
tio
n
i
s
estab
lis
h
ed
b
y
m
atc
h
i
n
g
th
e
te
m
p
late
"
T
em
p
late
Ma
tc
h
i
n
g
"
w
h
o
is
in
ter
ested
i
n
co
m
p
ar
i
n
g
t
h
e
i
n
ten
s
it
y
o
f
p
ix
els
b
et
w
ee
n
a
p
r
ed
ef
in
ed
tem
p
late
an
d
s
e
v
er
al
s
u
b
r
eg
io
n
s
o
f
th
e
i
m
a
g
e
to
b
e
an
al
y
ze
d
.
T
h
is
p
r
o
ce
s
s
is
i
n
p
r
ac
tice
to
p
er
f
o
r
m
m
u
ltip
le
s
ca
n
s
co
v
er
in
g
t
h
e
en
tire
ar
e
a
o
f
th
e
i
m
ag
e.
T
h
e
m
o
s
t
f
a
v
o
r
ab
le
to
th
e
p
r
esen
ce
o
f
f
ac
es
p
lace
s
w
i
ll
t
h
er
e
f
o
r
e
b
e
ea
s
il
y
id
e
n
ti
f
ied
b
y
m
i
n
i
m
u
m
d
i
s
ta
n
ce
b
et
w
ee
n
th
e
te
m
p
late
a
n
d
th
e
u
n
d
er
l
y
in
g
i
m
a
g
e.
Fig
u
r
e
2
s
h
o
w
s
t
h
e
d
etailed
p
r
o
ce
s
s
o
f
th
e
v
ar
io
u
s
p
h
ase
s
lis
ted
d
is
ap
p
o
in
ted
:
B
e
g
i
n
A
c
q
u
i
s
i
t
i
o
n
o
f
a
mi
x
e
d
d
o
c
u
me
nt
T
h
r
e
sh
o
l
d
i
n
g
R
G
B
s
p
a
c
e
S
e
g
me
n
t
a
t
i
o
n
o
f
sk
i
n
r
e
g
i
o
n
s
Ex
t
r
a
c
t
i
o
n
o
f
i
mag
e
a
r
e
a
s
S
e
g
me
n
t
a
t
i
o
n
f
a
c
e
i
mag
e
s
-
f
a
c
e
d
e
t
e
c
t
i
o
n
En
d
Fig
u
r
e
2
.
P
r
o
ce
s
s
R
ef
lects t
h
e
Gen
er
al
Ob
j
ec
tiv
e
o
f
th
e
De
v
e
lo
p
ed
Sy
s
te
m
.
3
.
2
.
Det
ec
t
i
o
n
o
f
Sk
in Co
lo
r
Sect
io
ns
Sev
er
al
co
lo
r
s
c
h
e
m
es
t
h
at
ca
n
b
e
ap
p
lied
to
t
h
e
s
k
in
d
etec
t
io
n
t
h
is
v
ar
iab
ili
t
y
d
ep
en
d
s
o
n
th
e
co
lo
r
s
p
ac
e
ad
o
p
ted
f
o
r
th
e
r
ep
r
esen
tatio
n
o
f
p
ix
el
s
.
T
h
e
m
o
s
t
co
m
m
o
n
l
y
u
s
ed
m
o
d
els
ar
e:
R
G
B
s
p
ac
e,
n
o
r
m
alize
d
R
GB
,
HSV,
YC
b
C
r
.
I
n
th
is
wo
r
k
w
e
h
a
v
e
c
h
o
s
en
R
GB
s
p
a
ce
th
at
r
ef
lects
th
e
p
h
y
s
io
lo
g
y
o
f
t
h
e
e
y
e.
I
n
d
ee
d
th
e
h
u
m
an
e
y
e
d
is
ti
n
g
u
i
s
h
co
lo
r
s
th
r
o
u
g
h
r
ec
ep
to
r
s
ca
lled
co
n
es.
T
h
er
e
ar
e
all
in
all
th
r
ee
t
y
p
es
o
f
co
n
e
s
,
w
h
ic
h
m
ea
n
s
t
h
at
a
n
y
co
lo
r
d
is
ce
r
n
ib
le
b
y
m
an
i
s
ch
ar
ac
ter
i
ze
d
b
y
a
p
o
in
t in
a
t
h
r
ee
-
d
i
m
e
n
s
io
n
al
s
p
ac
e.
Sp
ec
if
icall
y
,
t
h
e
v
al
u
es
m
ea
s
u
r
ed
b
y
th
e
t
h
r
ee
t
y
p
e
s
o
f
co
n
e
s
es
s
en
tiall
y
r
ep
r
esen
t
t
h
e
i
n
te
n
s
it
ies
o
f
r
ed
,
g
r
ee
n
an
d
b
lu
e,
R
GB
s
p
ac
e
b
ein
g
d
ef
in
ed
f
r
o
m
t
h
e
th
r
ee
p
r
im
ar
y
co
lo
r
s
.
His
co
d
in
g
w
a
s
d
ev
elo
p
ed
in
1
9
3
1
b
y
t
h
e
I
n
ter
n
at
io
n
al
C
o
m
m
is
s
io
n
o
n
I
llu
m
i
n
atio
n
(
C
I
E
)
[
1
2
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
J
E
C
E
Vo
l.
6
,
No
.
6
,
Decem
b
er
201
6
:
2
8
2
8
–
2
8
3
5
2832
T
h
e
s
k
in
is
s
h
o
w
n
i
n
th
e
m
aj
o
r
it
y
o
f
ca
s
es,
b
y
a
p
o
r
tio
n
o
f
an
ar
ea
o
f
p
ar
ticu
lar
co
lo
r
s
.
T
o
is
o
late
th
e
ar
ea
s
li
k
el
y
to
co
n
tain
f
ac
es,
u
s
in
g
th
e
b
o
r
d
er
s
o
f
t
h
ese
r
eg
io
n
s
as
th
r
es
h
o
ld
v
al
u
es
o
n
a
n
i
m
ag
e,
it
is
p
o
s
s
ib
le
to
ex
tr
ac
t
p
i
x
els
w
h
o
s
e
co
lo
r
ca
n
b
e
m
atc
h
ed
to
t
h
at
o
f
t
h
e
s
k
in
[
1
3
]
.
I
n
d
ee
d
,
f
o
r
d
etec
tin
g
s
k
i
n
r
eg
io
n
s
,
a
th
r
es
h
o
ld
S c
h
ar
ac
ter
izin
g
t
h
e
co
lo
r
o
f
th
e
s
k
i
n
b
ein
g
d
ef
i
n
ed
,
an
d
to
d
etec
t sk
in
co
lo
r
p
ix
el
s
is
t
h
e
r
u
le:
(
)
(
2
)
Fig
u
r
e
3
s
h
o
w
s
an
ex
a
m
p
le
s
e
t b
y
o
u
r
s
y
s
te
m
th
at
r
ef
lect
s
t
h
e
co
n
v
er
s
io
n
o
f
a
m
i
x
ed
co
lo
r
d
o
cu
m
en
t
(
tex
t+
i
m
ag
e
s
)
in
to
t
w
o
clas
s
e
s
o
f
p
ix
el
s
o
r
th
e
w
h
ite
p
ix
el
s
w
h
ic
h
r
ep
r
esen
t
t
h
e
co
lo
r
o
f
th
e
s
k
in
a
n
d
th
e
b
lack
p
ix
els r
ep
r
esen
tin
g
t
h
e
c
o
lo
r
n
o
t sk
in
.
Fig
u
r
e
3
.
Sa
m
p
le
B
in
ar
y
C
o
n
v
er
ts
Mix
ed
Do
cu
m
e
n
t D
ev
e
lo
p
ed
b
y
o
u
r
S
y
s
te
m
,
W
h
ite
P
ix
els=
Sk
i
n
C
o
lo
r
-
B
la
ck
P
ix
els=
No
t S
k
i
n
C
o
lo
r
.
3
.
3
.
F
ilte
ring
I
nd
ependent
S
kin
Co
lo
r
Reg
io
ns
Du
r
in
g
th
is
s
ta
g
e,
tr
ea
t
m
e
n
t
is
ap
p
lied
to
th
e
d
o
cu
m
e
n
t
r
esu
l
tin
g
f
r
o
m
t
h
e
p
r
ev
io
u
s
p
h
ase,
th
en
u
s
ed
a
m
et
h
o
d
o
f
s
eg
m
en
tatio
n
o
f
th
e
w
h
ite
ar
ea
s
w
h
ic
h
is
to
b
r
o
w
s
e
th
r
o
u
g
h
a
w
i
n
d
o
w
,
all
p
ix
els
o
f
th
i
s
d
o
cu
m
en
t.
A
lo
n
g
t
h
e
w
a
y
,
ea
c
h
w
in
d
o
w
is
ca
lc
u
lated
th
e
r
ep
o
r
t
(
P
N:
T
h
e
n
u
m
b
er
o
f
b
lack
p
ix
els
a
n
d
P
B
:
n
u
m
b
er
o
f
w
h
ite
p
i
x
els)
to
b
e
co
m
p
ar
ed
to
a
t
h
r
es
h
o
ld
s
to
d
ed
u
ct
if
t
h
e
co
n
te
n
t
o
f
th
e
w
i
n
d
o
w
is
o
r
is
n
o
t
a
p
ar
ty
a
n
i
m
ag
e
(
3
)
T
h
u
s
t
h
e
p
i
x
els
n
o
t
ch
ec
k
i
n
g
r
u
le
(
3
)
ar
e
f
o
r
ce
d
to
ta
k
e
t
h
e
b
lack
d
esp
ite
th
e
ir
i
n
itial
w
h
ite
c
o
lo
r
,
co
n
s
eq
u
e
n
tl
y
o
n
l
y
p
ix
e
l
s
,
wh
ite
co
n
tai
n
ed
i
n
a
s
k
in
co
lo
r
i
m
ag
e
ar
e
w
h
ite.
T
h
e
f
i
g
u
r
e
4
s
h
o
w
n
b
elo
w
illu
s
tr
ates t
h
e
r
esu
lt o
b
tain
ed
b
y
ap
p
lien
g
t
h
e
f
iltri
n
g
p
r
o
ce
s
s
i
n
g
o
n
F
i
g
u
r
e
4
.
Fig
u
r
e
4
.
E
x
a
m
p
le
o
f
F
ilter
ed
Do
cu
m
e
n
t
b
y
t
h
e
Met
h
o
d
Des
cr
ib
ed
in
th
is
p
ar
ag
r
ap
h
3
.
4
.
E
x
t
ra
ct
ing
I
m
a
g
es
S
kin
Co
lo
r
a
nd
F
a
ce
Det
ec
t
io
n Re
g
io
ns
by
T
e
m
pla
t
e
M
a
t
ching
Du
r
in
g
th
i
s
s
tep
,
a
s
ca
n
to
lo
ca
te
th
e
ar
ea
s
w
i
th
w
h
ite
p
i
x
e
ls
.
T
h
e
d
if
f
ic
u
lt
y
o
f
t
h
is
p
h
as
e
lies
w
it
h
th
e
f
ac
t
t
h
at
t
h
e
in
p
u
t
ac
q
u
ir
e
d
d
o
cu
m
e
n
t
ca
n
co
n
tain
m
u
lti
p
le
s
k
in
co
l
o
r
i
m
a
g
es
(
Fi
g
u
r
e
5
)
.
I
n
th
is
ca
s
e,
th
e
p
r
o
ce
s
s
co
n
s
is
ts
i
n
d
etec
ti
n
g
t
h
e
f
ir
s
t i
m
a
g
e
w
h
o
s
e
p
i
x
els ar
e
w
h
ite,
u
n
d
er
g
o
h
er
tr
ea
t
m
e
n
t
in
v
o
lv
i
n
g
te
m
p
late
m
atc
h
in
g
m
et
h
o
d
to
d
etec
t
wh
eth
er
it
co
n
tain
s
o
r
n
o
t
a
f
ac
e,
th
e
n
iter
ate
t
h
e
s
a
m
e
p
r
o
ce
s
s
i
n
g
to
d
etec
t
t
h
e
p
r
esen
ce
o
r
ab
s
en
ce
f
ac
e
i
n
th
e
o
th
er
r
eg
io
n
s
w
h
o
s
e
p
ix
el
s
a
r
e
w
h
ite.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2088
-
8708
F
a
ce
Dete
ctio
n
in
a
Mixed
-
S
u
b
ject
Do
cu
men
t
(
L
h
o
u
s
s
a
in
e
B
o
u
h
o
u
)
2833
Fig
u
r
e
5
.
E
x
a
m
p
le
o
f
L
o
c
atio
n
o
f
B
o
th
S
k
i
n
C
o
lo
r
I
m
ag
e
s
i
n
a
Mix
ed
Do
cu
m
e
n
t
T
h
e
g
o
o
d
o
f
T
em
p
late
Ma
tc
h
in
g
m
et
h
o
d
is
to
s
ca
n
ea
c
h
ar
ea
a
m
o
n
g
t
h
e
d
etec
ted
ar
ea
s
,
an
d
co
m
p
ar
i
n
g
th
e
in
te
n
s
it
y
o
f
p
ix
els
b
et
w
ee
n
a
m
o
d
el
te
m
p
late
(
te
m
p
late)
p
r
ed
ef
in
ed
(
in
o
u
r
ca
s
e
Fi
g
u
r
e
6
)
an
d
s
ev
er
al
s
u
b
-
r
e
g
io
n
s
o
f
t
h
e
ar
ea
to
b
e
an
al
y
ze
d
.
T
h
is
p
r
o
ce
s
s
is
in
p
r
ac
tice
to
p
e
r
f
o
r
m
m
u
l
t
ip
le
s
ca
n
s
co
v
er
i
n
g
th
e
en
tire
ar
ea
o
f
th
e
zo
n
e.
T
h
e
m
o
s
t
f
a
v
o
r
ab
le
to
th
e
p
r
esen
ce
o
f
f
ac
es
p
lace
s
w
ill
th
er
ef
o
r
e
b
e
ea
s
il
y
id
en
ti
f
ied
b
y
m
in
i
m
u
m
d
i
s
ta
n
ce
b
et
w
ee
n
t
h
e
te
m
p
late
an
d
t
h
e
u
n
d
er
l
y
i
n
g
ar
ea
.
Am
o
n
g
th
e
p
o
s
s
ib
le
d
is
tan
ce
s
,
th
er
e
ar
e
p
ar
ticu
lar
l
y
t
h
e
n
o
r
m
L
1
(
Ma
n
a
h
a
n
d
is
ta
n
ce
)
an
d
th
e
L
2
n
o
r
m
(
E
u
c
lid
ea
n
d
is
ta
n
ce
)
,
cr
o
s
s
-
co
r
r
elatio
n
,
etc.
No
r
m
e
L
1
:
∑
(
)
(
)
(
)
(
4
)
No
r
m
e
L
2
:
√
∑
(
)
(
)
(
)
(
5
)
w
h
er
e
(
)
an
d
(
)
ar
e
th
e
v
alu
e
o
f
th
e
p
ix
el
(
i,
j
)
in
th
e
t
w
o
i
m
a
g
es
to
co
m
p
ar
e.
I
t
th
er
ef
o
r
e
f
o
llo
w
s
t
h
at
t
h
e
m
o
r
e
t
h
ese
v
al
u
es
ar
e
clo
s
er
,
th
e
m
o
r
e
i
m
ag
e
s
ar
e
s
i
m
ilar
[
1
4
]
.
W
e
p
r
esen
t
th
e
f
o
llo
w
i
n
g
(
Fi
g
u
r
e
6
)
,
th
e
te
m
p
late
t
h
at
w
a
s
u
s
ed
to
d
etec
t,
f
o
r
o
u
r
s
y
s
t
e
m
,
if
an
i
m
a
g
e
co
n
tai
n
s
a
f
ac
e
o
r
n
o
t.
Fig
u
r
e
6
.
Mo
d
el
T
em
p
late
U
s
ed
b
y
o
u
r
Sy
s
te
m
to
Dete
ct
W
h
eth
er
o
r
n
o
t
I
m
a
g
e
C
o
n
tai
n
s
a
F
ace
(
a)
(
b
)
Fig
u
r
e
7
.
Face
Dete
ctio
n
M
et
h
o
d
b
y
T
em
p
late
M
atc
h
i
n
g
By
s
et
tin
g
a
th
r
es
h
o
ld
f
o
r
th
e
d
etec
ted
p
o
s
itio
n
o
f
th
e
s
eg
m
en
t
th
at
is
m
o
s
t
s
i
m
ilar
to
th
e
te
m
p
late,
w
e
ca
n
j
u
d
g
e
th
i
s
s
e
g
m
en
t
co
n
ta
in
s
a
f
ac
e
o
r
n
o
t.
Fig
u
r
e
7
is
an
ex
a
m
p
le
th
at
i
llu
s
tr
ates
t
h
is
p
r
o
ce
s
s
,
i
n
f
ac
t,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
J
E
C
E
Vo
l.
6
,
No
.
6
,
Decem
b
er
201
6
:
2
8
2
8
–
2
8
3
5
2834
p
ar
t
(
a)
o
f
th
is
f
i
g
u
r
e
r
ef
lects
th
e
p
o
s
it
io
n
o
f
th
e
te
m
p
late
o
n
th
e
f
ac
e
b
y
co
n
s
in
p
ar
t
(
b
)
,
th
e
p
o
s
itio
n
o
f
th
e
te
m
p
late
is
b
e
y
o
n
d
th
e
t
h
r
es
h
o
ld
co
n
s
eq
u
en
tl
y
t
h
e
i
m
a
g
e
co
n
tain
s
n
o
f
ac
e.
T
h
e
p
ix
els
o
f
th
e
i
m
a
g
e
th
at
co
n
tain
s
n
o
f
ac
e
w
i
ll
th
e
n
b
e
d
eliv
er
ed
to
th
e
lo
w
est
g
r
ay
le
v
el
i
s
0
(
b
lack
)
,
c
o
n
s
eq
u
en
t
l
y
o
n
l
y
t
h
e
p
ix
els
o
f
t
h
e
i
m
a
g
e
co
n
tai
n
in
g
a
f
ac
e
(
Fi
g
u
r
e
8
)
w
h
ich
will
b
e
in
g
r
a
y
s
ca
le
1
(
W
h
ite)
.
Fig
u
r
e
8
.
Seg
m
en
ta
tio
n
o
f
t
h
e
Im
ag
e
C
o
n
tai
n
i
n
g
a
F
ace
T
o
d
etec
t
f
ac
e,
f
ir
s
t,
w
e
n
ee
d
to
ex
tr
ac
t
th
e
b
in
ar
y
i
m
ag
e
o
f
th
e
d
o
cu
m
e
n
t.
W
e
th
e
n
tr
y
to
lo
ca
te
th
e
f
ac
e
ar
ea
,
b
ased
o
n
t
h
e
p
o
s
itio
n
o
f
t
h
e
te
m
p
late
id
en
ti
f
ied
i
n
th
e
p
r
ec
ed
in
g
p
ar
ag
r
ap
h
as
an
in
i
tial
b
e
n
ch
m
ar
k
.
T
h
e
r
esu
lt is
a
g
r
a
y
s
ca
le
i
m
a
g
e
th
at
r
ep
r
esen
ts
t
h
e
f
ac
e
t
o
d
etec
t in
q
u
e
s
tio
n
(
Fi
g
u
r
e
9
).
Fig
u
r
e
9
.
E
x
tr
ac
tio
n
o
f
t
h
e
F
ac
e
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
NS
W
e
co
n
d
u
cted
s
o
m
e
tes
ts
o
n
t
ex
t
d
o
cu
m
en
t
s
w
h
ich
w
er
e
in
s
er
ted
im
a
g
e
s
w
e
h
a
v
e
d
o
w
n
lo
ad
ed
f
r
o
m
th
e
Net.
T
ests
ar
e
p
er
f
o
r
m
ed
o
n
i
m
a
g
e
s
th
a
t
co
m
e
i
n
d
i
f
f
er
en
t
co
lo
r
s
an
d
p
r
ese
n
ti
n
g
f
i
g
u
r
es
th
at
m
a
y
o
r
m
a
y
n
o
t
b
e
p
eo
p
le'
s
f
ac
es.
W
e
n
o
te
in
t
h
e
m
aj
o
r
it
y
o
f
th
e
te
s
ts
(
8
5
%
o
f
test
s
)
p
o
s
itiv
e
r
e
s
u
l
ts
f
o
r
b
o
th
f
ac
e
i
m
ag
e
s
th
at
th
e
i
m
ag
e
s
n
o
t
f
ac
es.
Am
o
n
g
th
e
f
a
ls
e
a
lar
m
s
,
tes
tin
g
o
n
a
n
i
m
a
g
e
co
n
tain
in
g
a
f
ac
e
o
f
a
c
h
ild
o
r
a
b
ab
y
(
Fig
u
r
e
1
0
)
.
T
h
is
f
al
s
e
alar
m
i
s
d
u
e
to
th
e
d
ev
iat
io
n
o
f
t
h
e
s
i
m
ilar
it
y
b
et
w
ee
n
th
e
te
m
p
lat
e
an
d
ea
c
h
w
i
n
d
o
w
u
n
d
er
t
h
e
i
m
a
g
e
o
f
t
h
e
i
n
p
u
t
d
o
cu
m
en
t
w
h
en
s
ca
n
n
i
n
g
i
n
s
ea
r
ch
o
f
t
h
e
s
eg
m
e
n
t
o
f
th
e
i
m
ag
e
m
o
s
t
s
i
m
i
lar
to
th
e
T
em
p
late.
Fig
u
r
e
1
0
.
E
x
a
m
p
le
Fals
e
A
le
r
t U
p
o
n
Dete
ctio
n
o
f
t
h
e
F
ace
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2088
-
8708
F
a
ce
Dete
ctio
n
in
a
Mixed
-
S
u
b
ject
Do
cu
men
t
(
L
h
o
u
s
s
a
in
e
B
o
u
h
o
u
)
2835
5.
CO
NCLU
SI
O
N
T
h
e
d
etec
tio
n
s
y
s
te
m
th
a
t
w
e
h
av
e
d
e
v
elo
p
ed
an
d
in
tr
o
d
u
c
ed
in
th
is
ar
ticle
i
s
a
p
r
eli
m
i
n
ar
y
s
tag
e
an
d
o
f
g
r
ea
t
i
m
p
o
r
tan
ce
i
n
an
y
f
ac
e
r
ec
o
g
n
it
io
n
s
y
s
te
m
as
b
ef
o
r
e
la
u
n
c
h
i
n
g
th
e
s
y
s
te
m
to
r
ec
o
g
n
ize
in
d
iv
id
u
als,
v
er
if
y
i
n
g
th
e
p
r
es
en
ce
o
r
n
o
t
o
f
a
f
ac
e
i
n
th
e
s
ce
n
e
ca
n
b
e
v
er
y
b
e
n
e
f
icial.
T
h
is
s
tep
w
i
ll
allo
w
th
e
s
y
s
te
m
to
d
iv
er
t t
h
e
r
ec
o
g
n
itio
n
o
p
er
atio
n
i
f
th
e
s
ce
n
e
c
o
n
tain
s
n
o
f
ac
e.
Ou
r
ap
p
licatio
n
is
b
ased
,
f
ir
s
tl
y
,
o
n
th
e
h
y
b
r
id
m
e
th
o
d
th
at
p
er
f
o
r
m
s
s
e
g
m
en
ta
tio
n
o
f
t
w
o
class
es
o
f
zo
n
es,
t
h
e
f
ir
s
t
c
lass
e
s
f
o
r
ar
ea
s
,
p
r
esen
t
in
th
e
s
ce
n
e,
t
h
e
c
o
lo
r
b
ein
g
t
h
at
o
f
t
h
e
s
k
i
n
a
n
d
s
ec
o
n
d
ca
teg
o
r
ies
f
o
r
ar
ea
s
,
p
r
esen
t
in
th
e
s
ce
n
e,
th
e
co
lo
r
o
f
w
h
ich
d
i
f
f
er
s
f
r
o
m
th
at
o
f
t
h
e
s
k
in
,
th
e
co
lo
r
r
ep
r
esen
tatio
n
s
p
ac
e
s
elec
ted
is
th
e
R
GB
s
p
ac
e.
An
d
s
ec
o
n
d
l
y
,
o
u
r
ap
p
licatio
n
i
s
b
ased
o
n
T
em
p
late
Ma
tch
i
n
g
m
et
h
o
d
to
d
etec
t
th
e
p
r
esen
ce
o
r
n
o
t in
t
h
e
s
ta
g
e
o
f
a
f
ac
e.
T
h
e
r
esu
lts
ar
e
s
a
tis
f
ac
to
r
y
b
e
ca
u
s
e
t
h
e
y
allo
w
ed
u
s
to
in
cr
e
ase
th
e
p
er
f
o
r
m
an
ce
o
f
t
h
e
m
e
th
o
d
co
lo
r
d
etec
tio
n
an
d
t
h
e
m
atc
h
i
n
g
o
f
th
e
te
m
p
late
u
s
ed
s
ep
ar
at
el
y
.
T
h
ese
r
es
u
lt
s
ar
e
s
ig
n
i
f
i
ca
n
tl
y
e
f
f
ec
tiv
e
in
co
m
p
ar
is
o
n
w
ith
t
h
o
s
e
m
ad
e
d
u
r
in
g
t
h
e
f
ac
e
d
etec
tio
n
p
h
as
e
in
o
u
r
o
w
n
p
r
ev
io
u
s
r
esear
c
h
[
1
5
]
.
Ho
w
e
v
er
,
I
n
o
r
d
er
to
b
etter
g
et
ar
o
u
n
d
t
h
e
p
r
o
b
lem
o
f
f
ac
e
d
etec
tio
n
,
we
p
r
o
p
o
s
e,
as
a
s
h
o
r
t
ter
m
p
er
s
p
ec
tiv
e,
to
tr
y
,
i
n
f
ac
e
d
etec
tio
n
p
h
a
s
e,
o
th
er
co
lo
r
s
p
ac
es
s
u
c
h
as
R
GB
a
n
d
n
o
r
m
alize
d
s
p
ac
e
HSV
s
p
ac
e
an
d
co
m
p
ar
e
th
eir
r
esu
lt
s
w
it
h
t
h
e
r
es
u
lt
s
o
f
t
h
e
R
GB
s
p
ac
e
an
d
tr
y
,
i
n
f
ac
e
r
ec
o
g
n
itio
n
p
h
ase,
to
i
m
p
r
o
v
e
th
e
m
et
h
o
d
s
u
s
ed
in
[
1
5
]
ten
n
an
t a
cc
o
u
n
t n
e
w
m
eth
o
d
o
f
ex
tr
ac
tio
n
a
n
d
an
al
y
s
is
n
o
n
-
li
n
ea
r
ch
ar
ac
ter
i
s
tics
(
E
A
N
F)
[
1
6
]
.
RE
F
E
R
E
NC
E
S
[1
]
A
.
K.
Ja
in
,
A
.
Ro
ss
a
n
d
S
.
P
ra
b
h
a
k
a
r,
“
A
n
in
tro
d
u
c
ti
o
n
t
o
b
io
m
e
tri
c
re
c
o
g
n
it
io
n
”
.
IEE
E
T
ra
n
s
a
c
ti
o
n
s
o
n
Circ
u
i
ts
a
n
d
S
y
ste
ms
fo
r V
id
e
o
T
e
c
h
n
o
lo
g
y
,
v
o
l.
1
4
,
p
p
.
4
–
2
0
,
Ja
n
2
0
0
4
.
[2
]
F
lo
re
n
t
P
ERRO
NN
IN,
Je
a
n
-
L
u
c
DU
G
E
LA
Y
,
“
In
tro
d
u
c
ti
o
n
à
la
Bio
m
é
tri
e
A
u
th
e
n
ti
f
ica
ti
o
n
d
e
s
In
d
iv
id
u
s
p
a
r
T
ra
it
e
m
e
n
t
A
u
d
io
-
V
id
é
o
”
,
Rev
u
e
T
ra
it
e
me
n
t
d
u
S
ig
n
a
l
,
V
o
lu
m
e
1
9
–
n
o
4
,
2
0
0
2
.
[3
]
S
.
L
iu
,
M
.
S
il
v
e
m
a
n
,
“
A
p
ra
ti
c
a
l
G
u
id
e
to
Bio
m
e
tri
c
S
e
c
u
rit
y
T
e
c
h
n
o
lo
g
y
”
,
IEE
E
Co
mp
u
ter
S
o
c
iety
,
IT
P
ro
-
S
e
c
u
rit
y
,
Ja
n
v
ier
-
F
é
v
rier 2
0
0
1
.
[4
]
A.
K.
Ja
in
,
L
.
Ho
n
g
,
S
.
P
a
n
k
a
n
ti
,
“
Bio
m
e
tri
c
s
:
P
ro
m
isin
g
F
ro
n
ti
e
rs
f
o
r
Em
e
r
g
in
g
Id
e
n
ti
f
ic
a
ti
o
n
M
a
rk
e
t
”
,
Co
mm
u
n
ica
ti
o
n
s o
f
th
e
ACM
,
p
p
.
9
1
-
9
8
,
F
e
b
r
u
a
ry
2
0
0
0
.
[5
]
C.
F
re
d
o
u
il
le,
J.
M
a
rieth
o
z
,
C.
Ja
b
o
u
let,
J.
He
n
n
e
b
e
rt,
J.
-
F
.
B
o
n
a
stre
,
C.
M
o
k
b
e
l,
F
.
Bim
b
o
t,
“
Be
h
a
v
io
r
o
f
a
Ba
y
e
sia
n
A
d
a
p
tatio
n
M
e
t
h
o
d
f
o
r
In
c
re
m
e
n
tal
En
ro
ll
m
e
n
t
in
S
p
e
a
k
e
r
V
e
rif
ica
ti
o
n
”
,
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Aco
u
stics
,
S
p
e
e
c
h
,
a
n
d
S
ig
n
a
l
Pro
c
e
ss
in
g
,
p
p
.
1
1
9
7
-
1
2
0
0
,
Ista
n
b
u
l
,
T
u
rq
u
ie,
5
-
9
Ju
n
2
0
0
0
.
[6
]
S.
G
u
e
rf
i
A
b
a
b
sa
,
“
A
u
th
e
n
ti
f
ica
ti
o
n
d
’i
n
d
iv
id
u
s
p
a
r
re
c
o
n
n
a
issa
n
c
e
d
e
c
a
ra
c
téristiq
u
e
s
b
io
m
é
tri
q
u
e
s
li
é
e
s
a
u
x
v
isa
g
e
s 2
D/3
D
”
,
Do
c
to
ra
l
t
h
e
sis,
UN
IV
ERS
IT
E
D’EV
RY
V
A
L
D'
ES
S
ON
NE,
o
c
t
o
b
e
r
2
0
0
8
.
[7
]
M
in
g
-
Hs
u
a
n
Ya
n
g
,
Da
v
id
J.
K
rieg
m
a
n
,
a
n
d
Na
re
n
d
ra
A
h
u
ja.
De
tec
ti
n
g
f
a
c
e
s
in
im
a
g
e
s:
A
su
rv
e
y
.
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
p
a
tt
e
rn
a
n
a
lys
is
a
n
d
ma
c
h
i
n
e
i
n
telli
g
e
n
c
e
,
2
4
(1
):
1
7
4
6
-
1
7
6
2
,
2
0
0
2
.
[8
]
Ch
e
n
g
-
Ch
in
C
h
ian
g
,
W
e
n
-
Ka
i
Ta
i,
M
a
u
-
T
su
e
n
Ya
n
g
,
Yi
-
T
in
g
Hu
a
n
g
,
a
n
d
Ch
i
-
Ja
u
n
g
Hu
a
n
g
.
A
n
o
v
e
l
m
e
th
o
d
f
o
r
d
e
tec
ti
n
g
li
p
s,
e
y
e
s an
d
f
a
c
e
s
in
r
e
a
l
ti
m
e
.
Rea
l
-
T
ime
Ima
g
in
g
,
9
(4
)
:
2
7
7
-
2
8
7
,
2
0
0
3
.
[9
]
W
e
n
lo
n
g
Zh
e
n
g
a
n
d
S
u
c
h
e
n
d
ra
M
.
Bh
a
n
d
a
rk
a
r.
F
a
c
e
d
e
tec
ti
o
n
a
n
d
t
ra
c
k
in
g
u
si
n
g
a
b
o
o
ste
d
a
d
a
p
ti
v
e
p
a
rti
c
le
fi
lt
e
r.
J
o
u
rn
a
l
o
f
Vi
su
a
l
C
o
mm
u
n
i
c
a
ti
o
n
a
n
d
Ima
g
e
Rep
re
se
n
ta
ti
o
n
,
2
0
(
1
):
9
-
2
7
,
2
0
0
9
.
[1
0
]
P
.
S
i
n
h
a
,
“
Ob
jec
t
Re
c
o
g
n
it
i
o
n
v
ia
I
m
a
g
e
In
v
a
rian
ts:
A
Ca
se
S
tu
d
y
”
,
In
v
e
stig
a
ti
v
e
Op
h
th
a
lmo
l
o
g
y
a
n
d
Vi
su
a
l
S
c
ien
c
e
,
v
o
l.
3
5
,
n
o
.
4
,
p
p
.
1
7
3
5
-
1
7
4
0
,
1
9
9
4
.
[1
1
]
P
.
P
e
e
r,
J
.
Ko
v
a
c
,
F
.
S
o
li
n
a
.
Hu
m
a
n
sk
in
c
o
l
o
u
r
c
lu
ste
rin
g
f
o
r
f
a
c
e
d
e
tec
ti
o
n
.
In
su
b
m
it
ted
to
EUROCON
2
0
0
3
–
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
C
o
mp
u
ter
a
s
a
T
o
o
l
,
2
0
0
3
.
[1
2
]
Ch
.
Be
n
c
h
e
riet,
A
.
H.
Bo
u
a
ll
e
g
&
H.
T
e
b
b
ik
h
&
B.
G
u
e
rz
ize
&
W
.
Be
lg
u
id
o
u
m
,
“
Dé
tec
ti
o
n
d
e
V
isa
g
e
s
p
a
r
M
é
th
o
d
e
H
y
b
rid
e
Co
u
leu
r
d
e
P
e
a
u
e
t
Te
m
p
late
M
a
tch
in
g
”
,
4
th
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
:
S
c
ien
c
e
s
o
f
El
e
c
tro
n
ic,
T
e
c
h
n
o
l
o
g
ies
o
f
I
n
fo
rm
a
ti
o
n
a
n
d
T
e
lec
o
mm
u
n
ica
ti
o
n
s
,
M
a
rc
h
2
5
-
2
9
,
2
0
0
7
–
T
UN
IS
IA
.
[1
3
]
G
a
r
c
ia
C.
,
T
z
iri
tas
G
.
,
"
F
a
c
e
De
tec
ti
o
n
Us
in
g
Qu
a
n
ti
z
e
d
S
k
i
n
Co
l
o
r
Re
g
io
n
s
M
e
rg
in
g
a
n
d
W
a
v
e
let
P
a
c
k
e
t
A
n
a
l
y
si
s"
,
IEE
E
T
ra
n
s
a
c
ti
o
n
s o
n
M
u
lt
ime
d
ia
,
1
(
3
),
S
e
p
tem
b
e
r
1
9
9
9
,
p
.
2
6
4
-
2
7
7
.
[1
4
]
G
.
He
u
sc
h
:
"
Dé
te
c
ti
o
n
a
u
t
o
m
a
ti
q
u
e
d
e
v
isa
g
e
s d
a
n
s u
n
e
sé
q
u
e
n
c
e
v
id
é
o
"
,
S
S
C
se
m
e
stre
5
-
1
8
F
é
v
ri
e
r
2
0
0
2
.
[1
5
]
L
.
Bo
u
h
o
u
,
R.
E
l
Ay
a
c
h
i,
M
.
F
a
k
ir,
M
.
Ou
k
e
ss
o
u
,
“
Re
c
o
g
n
it
i
o
n
o
f
a
F
a
c
e
in
a
M
ix
e
d
Do
c
u
m
e
n
t
”
,
T
EL
KOM
NIKA
In
d
o
n
e
sia
n
J
o
u
rn
a
l
o
f
El
e
c
trica
l
En
g
i
n
e
e
rin
g
,
V
o
l
1
5
No
2
,
2
0
1
5
p
a
g
e
s 3
0
1
-
3
1
2
.
[1
6
]
A
.
M
a
h
d
a
v
i
Ho
r
m
a
t,
K.
F
a
e
z
,
Z
.
S
h
o
k
o
o
h
i,
M
.
Zah
e
r
Ka
rim
i,
“
T
h
e
n
e
w
m
e
th
o
d
o
f
Ex
trac
ti
o
n
a
n
d
A
n
a
ly
sis
o
f
No
n
li
n
e
a
r
F
e
a
tu
re
s
f
o
r
f
a
c
e
re
c
o
g
n
it
io
n
”
,
I
n
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
El
e
c
trica
l
a
n
d
Co
mp
u
ter
En
g
in
e
e
rin
g
(
IJ
ECE
)
,
V
o
l
2
N
o
6
,
2
0
1
2
p
a
g
e
s 7
6
6
-
7
7
3
.
Evaluation Warning : The document was created with Spire.PDF for Python.