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.
7
,
No
.
4
,
A
u
g
u
s
t
201
7
,
p
p
.
1
9
2
3
~
1
9
3
3
I
SS
N:
2088
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v7
i
4
.
p
p
1
9
2
3
-
1933
1923
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
EV
-
SIFT
-
A
n Ex
tended
Scale Inv
a
ria
nt
Face
Recog
nition
for
Plas
tic
Su
rg
ery F
a
ce Re
co
g
nition
Arc
ha
na
H
.
Sa
ble
1
,
Sa
nja
y
N.
T
a
lba
r
2
,
H
a
richa
ra
n A
m
a
rsin
g
Dhirba
s
i
3
1
S
c
h
o
o
l
o
f
Co
m
p
u
tatio
n
a
l
S
c
ien
c
e
s,
S
.
R.
T
.
M
.
Un
iv
e
rsity
,
Na
n
d
e
d
-
4
3
1
6
0
6
,
In
d
ia
2
De
p
a
rtme
n
t
o
f
El
e
c
tro
n
ics
a
n
d
T
e
lec
o
m
m
u
n
ica
ti
o
n
E
n
g
g
.
,
S
.
G
.
G
.
S
.
I.
E
&
T
.
,
Na
n
d
e
d
(M
.
S
)
,
In
d
ia
3
N.E
.
S
,
S
c
ien
c
e
Co
ll
e
g
e
,
S
.
R.
T
.
M
.
Un
iv
e
rsity
Na
n
d
e
d
(M
.
S
)
,
In
d
i
a
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Au
g
1
1
,
2
0
1
6
R
ev
i
s
ed
Ma
y
1
5
,
2
0
1
7
A
cc
ep
ted
Ma
y
3
0
,
2
0
1
7
T
h
is
p
a
p
e
r
p
re
se
n
ts
a
n
e
w
tec
h
n
iq
u
e
c
a
ll
e
d
En
tr
o
p
y
b
a
se
d
S
IF
T
(
EV
-
S
I
F
T
)
f
o
r
a
c
c
u
ra
te
f
a
c
e
re
c
o
g
n
it
io
n
a
f
ter
th
e
p
las
ti
c
su
rg
e
ry
.
T
h
e
c
o
rre
sp
o
n
d
in
g
f
e
a
tu
re
e
x
trac
ts
th
e
k
e
y
p
o
in
ts
a
n
d
v
o
l
u
m
e
o
f
th
e
sc
a
le
-
sp
a
c
e
str
u
c
tu
re
f
o
r
w
h
ich
th
e
in
f
o
rm
a
ti
o
n
ra
te
is
d
e
t
e
rm
in
e
d
.
T
h
is
p
ro
v
id
e
s
lea
st
e
ff
e
c
t
o
n
u
n
c
e
rtain
v
a
riatio
n
s in
th
e
f
a
c
e
si
n
c
e
th
e
e
n
tro
p
y
is
th
e
h
ig
h
e
r
o
rd
e
r
sta
ti
stica
l
f
e
a
tu
re
.
T
h
e
c
o
rre
sp
o
n
d
i
n
g
EV
-
S
IF
T
fe
a
tu
re
s
a
re
a
p
p
li
e
d
to
t
h
e
S
u
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
f
o
r
c
l
a
ss
i
f
ica
ti
o
n
.
T
h
e
n
o
r
m
a
l
S
IF
T
f
e
a
tu
re
e
x
tr
a
c
ts
th
e
k
e
y
p
o
i
n
ts
b
a
se
d
o
n
t
h
e
c
o
n
tras
t
o
f
th
e
i
m
a
g
e
a
n
d
th
e
V
-
S
IF
T
f
e
a
tu
re
e
x
trac
ts
th
e
k
e
y
p
o
in
ts
b
a
se
d
o
n
th
e
v
o
lu
m
e
o
f
th
e
str
u
c
tu
re
.
Ho
w
e
v
e
r,
th
e
EV
-
S
IF
T
m
e
th
o
d
p
ro
v
id
e
s
b
o
t
h
th
e
c
o
n
tra
st
a
n
d
v
o
l
u
m
e
in
f
o
r
m
a
ti
o
n
.
T
h
u
s
EV
-
S
I
F
T
p
ro
v
id
e
b
e
tt
e
r
p
e
rf
o
rm
a
n
c
e
w
h
e
n
c
o
m
p
a
re
d
w
it
h
P
CA
,
n
o
rm
a
l
S
IF
T
a
n
d
V
-
S
IF
T
b
a
se
d
f
e
a
tu
re
e
x
tra
c
ti
o
n
.
K
ey
w
o
r
d
:
EV
-
SIFT
f
ea
t
u
r
e
Face
r
ec
o
g
n
itio
n
P
last
ic
s
u
r
g
er
y
SVM
class
if
icatio
n
Co
p
y
rig
h
t
©
2
0
1
7
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
:
A
r
c
h
an
a
H.
Sab
le,
Sch
o
o
l o
f
C
o
m
p
u
tatio
n
al
Scie
n
ce
s
,
S.R
.
T
.
M.
Un
iv
er
s
it
y
,
Nan
d
ed
-
4
3
1
6
0
6
,
I
n
d
ia.
E
m
ail:
h
ello
ar
ch
u
2
7
@
g
m
ai
l.c
o
m
1.
I
NT
RO
D
UCT
I
O
N
Hu
m
an
f
ac
es
ar
e
m
u
l
ti
-
d
i
m
e
n
s
io
n
al
an
d
co
m
p
lex
v
i
s
u
al
s
ti
m
u
li,
w
h
ic
h
ca
r
r
y
ab
u
n
d
an
t
u
s
e
f
u
l
in
f
o
r
m
atio
n
r
eg
ar
d
in
g
t
h
e
u
n
iq
u
e
n
es
s
o
f
a
p
er
s
o
n
.
Fac
e
r
ec
o
g
n
izi
n
g
u
t
ilizi
n
g
th
e
m
f
o
r
s
ec
u
r
it
y
a
n
d
au
th
e
n
tica
tio
n
p
u
r
p
o
s
es
h
as
t
ak
en
a
n
e
w
t
u
r
n
i
n
c
u
r
r
en
t
er
a
o
f
i
m
a
g
e
an
al
y
s
is
a
n
d
co
m
p
u
ter
v
i
s
io
n
,
f
o
r
in
s
ta
n
ce
,
in
ap
p
licatio
n
s
t
h
at
i
n
clu
d
e
s
u
r
v
e
illa
n
ce
,
i
m
a
g
e
r
etr
iev
al,
h
u
m
a
n
-
co
m
p
u
ter
in
ter
a
ctio
n
an
d
b
io
m
etr
ic
au
th
e
n
tica
tio
n
.
No
r
m
all
y
,
f
a
ce
r
ec
o
g
n
itio
n
s
y
s
te
m
d
o
es
n
o
t
n
ee
d
th
e
s
e
n
s
e
o
f
to
u
ch
o
r
in
ter
ac
tio
n
f
r
o
m
h
u
m
a
n
to
co
n
d
u
ct
t
h
e
p
r
o
ce
s
s
o
f
r
ec
o
g
n
itio
n
.
I
t
is
o
n
e
o
f
t
h
e
ad
v
a
n
ta
g
es
o
f
f
ac
e
r
ec
o
g
n
i
tio
n
o
v
er
t
h
e
o
t
h
er
r
ec
o
g
n
itio
n
m
et
h
o
d
s
.
Face
r
ec
o
g
n
itio
n
ca
n
d
en
o
te
eit
h
er
in
v
er
if
ica
tio
n
p
h
a
s
e
[
1
]
o
r
th
e
id
en
tific
atio
n
p
h
a
s
e
[
2
]
.
I
n
th
e
v
er
if
icat
io
n
p
h
ase,
m
atc
h
in
g
b
et
w
ee
n
t
w
o
f
ac
e
s
is
r
eso
lv
ed
.
T
h
er
e
ar
e
lo
t
o
f
m
eth
o
d
s
av
ailab
le
to
attain
f
ac
e
r
ec
o
g
n
itio
n
[3
-
8
]
.
B
u
t
t
h
e
ac
cu
r
ac
y
o
f
r
ec
o
g
n
i
tio
n
i
s
n
o
t
al
w
a
y
s
h
ig
h
.
T
h
i
s
is
b
ec
a
u
s
e
o
f
th
e
ch
an
g
i
n
g
d
e
g
r
ee
s
o
f
ill
u
m
i
n
ati
o
n
,
f
ac
ial
e
x
p
r
ess
io
n
s
,
p
o
s
es,
ag
in
g
,
lo
w
r
eso
lu
tio
n
i
n
p
u
t
i
m
ag
e
o
r
f
ac
ial
m
ar
k
s
[
9
]
,
[
1
0
]
.
Var
io
u
s
r
esear
c
h
er
s
h
a
v
e
i
m
p
le
m
e
n
ted
d
if
f
er
e
n
t
m
et
h
o
d
o
lo
g
ies
o
f
f
ac
e
r
ec
o
g
n
itio
n
to
tack
le
t
h
e
ef
f
ec
t
o
f
p
o
s
e
[
1
1
]
,
illu
m
in
at
i
o
n
[
1
2
]
,
lo
w
r
eso
l
u
tio
n
[
1
3
]
,
ag
in
g
[
1
4
]
o
r
a
co
m
b
in
a
tio
n
o
f
a
n
y
o
f
t
h
e
m
[
1
5
]
.
Ho
w
e
v
er
t
h
ese
u
n
ce
r
tain
t
ies
co
u
ld
b
e
f
air
l
y
o
v
er
co
m
e
a
n
d
in
th
e
p
last
ic
s
u
r
g
er
y
f
ac
e
s
,
t
h
e
r
ec
o
g
n
itio
n
g
e
t
s
ev
en
m
o
r
e
i
n
te
n
s
i
f
ied
w
i
th
th
e
id
en
ti
f
icatio
n
o
f
p
er
s
o
n
.
T
h
e
f
ail
u
r
e
o
f
r
ec
o
g
n
i
tio
n
in
p
la
s
tic
s
u
r
g
er
y
f
ac
es
i
s
d
u
e
to
th
e
lack
o
r
th
e
v
ar
iatio
n
in
f
ac
e
co
m
p
o
n
e
n
ts
,
s
k
i
n
te
x
tu
r
e,
g
lo
b
al
f
ac
e
ap
p
ea
r
an
ce
an
d
th
e
g
eo
m
etr
ic
r
elatio
n
s
h
ip
ex
i
s
ti
n
g
a
m
o
n
g
t
h
e
f
ac
ial
f
ea
t
u
r
es
lac
k
o
r
th
e
v
ar
iat
io
n
in
f
ac
e
co
m
p
o
n
en
ts
[
1
6
-
1
8
]
.
T
h
e
co
s
t
-
ef
f
icien
t
as
w
e
ll
as
th
e
ad
v
a
n
c
ed
m
ea
n
s
o
f
p
last
ic
s
u
r
g
er
y
h
a
s
attr
ac
ted
p
eo
p
le
all
ar
o
u
n
d
th
e
g
lo
b
e.
Ho
w
e
v
er
,
o
n
l
y
f
e
w
r
esear
ch
co
n
tr
ib
u
tio
n
s
o
r
m
et
h
o
d
o
lo
g
ies
h
a
v
e
b
ee
n
r
ep
o
r
ted
in
th
e
liter
atu
r
e
to
ad
d
r
ess
th
e
p
r
o
b
lem
o
f
r
ec
o
g
n
izi
n
g
t
h
e
p
last
ic
s
u
r
g
er
y
f
ac
e
s
.
Fe
w
o
f
t
h
e
m
i
n
cl
u
d
e,
th
e
r
ec
o
g
n
itio
n
u
s
i
n
g
lo
ca
l
r
eg
io
n
a
n
al
y
s
is
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.
7
,
No
.
4
,
A
u
g
u
s
t
2017
:
1
9
2
3
–
1
9
3
3
1924
[
1
9
]
,
Sh
ap
e
L
o
ca
l
B
in
ar
y
T
ex
tu
r
e
Featu
r
e
(
S
L
B
T
)
ca
s
ca
d
ed
w
it
h
p
er
io
cu
lar
f
ea
t
u
r
es
[
2
0
]
.
A
r
ev
ie
w
h
as
al
s
o
b
ee
n
d
o
n
e
in
[
2
1
]
to
illu
s
tr
ate
th
e
u
s
e
o
f
m
u
lti
m
o
d
al
f
ea
t
u
r
e
s
in
r
ec
o
g
n
izi
n
g
th
e
p
la
s
tic
s
u
r
g
er
y
b
ased
o
n
t
h
e
co
n
tr
ib
u
tio
n
s
.
T
h
e
co
n
tr
ib
u
tio
n
o
f
t
h
e
p
ap
er
m
a
in
l
y
co
n
s
id
er
s
th
r
ee
s
et
o
f
d
e
m
o
n
s
tr
atio
n
s
.
T
h
e
f
ir
s
t
o
n
e
is
d
etec
tio
n
o
f
ex
ac
t
p
last
ic
s
u
r
g
er
y
f
ac
e
u
s
i
n
g
th
e
n
o
v
el
SIFT
f
ea
tu
r
e
b
ased
o
n
e
n
tr
o
p
y
.
T
h
e
s
ec
o
n
d
o
n
e
is
to
ac
cu
r
atel
y
r
ec
o
g
n
izin
g
th
e
tr
u
e
s
u
r
g
er
y
p
o
r
tio
n
s
w
h
ile
t
h
e
th
ir
d
o
n
e
is
to
an
a
l
y
ze
th
e
p
er
f
o
r
m
an
ce
o
f
f
ac
e
r
ec
o
g
n
itio
n
b
ef
o
r
e
an
d
af
ter
p
l
asti
c
s
u
r
g
er
y
.
2.
L
I
T
E
R
AT
U
RE
R
E
VI
E
W
2
.
1
.
Rela
t
ed
Wo
rks
Ma
r
ia
De
Ma
r
s
ico
et
a
l.
[
2
2
]
h
av
e
m
ad
e
a
n
ac
c
u
r
ate
r
ec
o
g
n
itio
n
o
f
f
ac
e,
w
h
ich
h
a
s
u
n
d
er
g
o
n
e
p
last
ic
s
u
r
g
er
y
,
t
h
r
o
u
g
h
t
h
e
ap
p
licatio
n
o
f
t
h
e
r
e
g
io
n
-
b
ased
ap
p
r
o
ac
h
es
o
n
a
m
u
l
ti
m
o
d
al
s
u
p
er
v
i
s
ed
co
llab
o
r
ativ
e
ar
ch
itect
u
r
e,
ter
m
ed
as,
Sp
lit
Face
A
r
ch
i
tec
t
u
r
e
(
SF
A
)
.
T
h
e
y
h
a
v
e
v
al
id
ated
th
e
s
u
p
er
io
r
it
y
o
f
th
eir
m
eth
o
d
o
lo
g
y
t
h
r
o
u
g
h
a
p
p
ly
i
n
g
t
h
e
S
u
p
er
v
is
ed
SF
A
to
tr
ad
itio
n
al
P
C
A
a
n
d
FD
A
,
to
L
B
P
in
t
h
e
Mu
lti
s
ca
le,
R
o
tat
io
n
I
n
v
ar
ia
n
t
v
er
s
io
n
w
it
h
U
n
i
f
o
r
m
P
atter
n
s
,
F
A
R
O
(
Face
R
ec
o
g
n
itio
n
ag
ain
s
t
Occ
lu
s
io
n
s
an
d
E
x
p
r
ess
io
n
Var
iatio
n
s
)
an
d
FAC
E
(
Face
An
al
y
s
is
f
o
r
C
o
m
m
er
cial
E
n
titi
e
s
)
.
Na
m
a
n
Ko
h
li
et
al.
[
2
3
]
h
a
v
e
p
u
t
f
o
r
th
M
u
ltip
le
P
r
o
j
ec
tiv
e
Dictio
n
ar
y
L
ea
r
n
i
n
g
f
r
a
m
e
w
o
r
k
(
MP
DL
)
th
at
n
e
v
er
d
esire
s
to
co
m
p
u
te
th
e
an
d
n
o
r
m
s
to
r
ec
o
g
n
ize
n
o
r
m
al
f
ac
e
s
,
ev
e
n
af
ter
th
e
y
h
av
e
b
ee
n
m
o
d
i
f
ied
d
u
e
to
p
last
ic
s
u
r
g
er
y
.
M
u
lt
ip
le
p
r
o
j
ec
tiv
e
d
ictio
n
ar
ies
as
w
ell
as
th
e
c
o
m
p
ac
t
b
in
ar
y
f
a
ce
d
escr
ip
to
r
s
h
av
e
b
ee
n
u
t
ilized
to
lear
n
t
h
e
lo
c
al
an
d
t
h
e
g
lo
b
al
p
last
ic
s
u
r
g
er
y
f
ac
e
r
ep
r
esen
tat
io
n
s
,
in
o
r
d
er
to
f
ac
ilit
ate
t
h
e
d
is
cr
i
m
i
n
atio
n
o
f
t
h
e
p
last
ic
s
u
r
g
er
y
f
ac
es
f
r
o
m
th
e
o
r
ig
i
n
a
l
o
n
es.
T
h
e
test
in
g
t
h
at
w
as
d
o
n
e
o
n
th
e
p
last
i
c
s
u
r
g
er
y
d
atab
ase
h
a
s
r
esu
l
ted
in
an
ac
c
u
r
ac
y
o
f
ab
o
u
t 9
7
.
9
6
%.
C
h
o
llet
te
C
C
h
u
d
e
-
O
lis
a
h
et
al.
[
2
4
]
h
av
e
o
v
er
co
m
e
t
h
e
d
eg
r
ad
atio
n
i
n
t
h
e
p
er
f
o
r
m
an
c
e
o
f
f
ac
e
r
ec
o
g
n
itio
n
,
t
h
e
y
h
a
v
e
f
o
u
n
d
th
at
th
e
ir
ap
p
r
o
ac
h
h
as
o
u
ts
h
i
n
ed
th
e
p
r
ev
io
u
s
l
y
a
v
ailab
le
p
last
ic
s
u
r
g
er
y
f
ac
e
r
ec
o
g
n
itio
n
ap
p
r
o
ac
h
es,
ir
r
es
p
ec
tiv
e
o
f
th
e
ch
a
n
g
es
in
ill
u
m
in
at
io
n
a
s
w
ell
as
e
x
p
r
ess
io
n
s
a
n
d
t
h
e
f
ac
ial
m
o
d
i
f
icat
io
n
s
r
es
u
lti
n
g
f
r
o
m
p
last
ic
s
u
r
g
er
y
.
Ha
m
id
Ou
a
n
a
n
[
2
5
]
h
a
v
e
i
n
tr
o
d
u
ce
d
Gab
o
r
HOG
f
ea
t
u
r
es
b
ased
f
ac
e
r
ec
o
g
n
itio
n
s
ch
e
m
e,
w
h
i
ch
u
s
e
s
HO
G
i
n
s
tead
o
f
DO
G
in
t
h
e
SI
FT
.
M.
I
.
Ou
lo
u
l
[
2
6
]
in
tr
o
d
u
ce
d
an
e
f
f
icien
t
f
ace
r
ec
o
g
n
itio
n
u
s
in
g
SIFT
d
escr
ip
to
r
in
R
G
B
D
i
m
a
g
es
w
h
ich
is
b
a
s
ed
o
n
R
GB
-
D
i
m
a
g
es
p
r
o
d
u
ce
d
b
y
Ki
n
ec
t,
th
is
t
y
p
e
o
f
ca
m
er
a
s
co
s
t
le
s
s
a
n
d
it
ca
n
b
e
u
s
ed
in
a
n
y
en
v
ir
o
n
m
en
t
an
d
u
n
d
er
a
n
y
cir
cu
m
s
ta
n
ce
s
.
Hi
m
a
n
s
h
u
S.
B
h
att
et
al.
[
2
7
]
h
av
e
i
n
tr
o
d
u
ce
d
a
m
u
lt
i
-
o
b
j
ec
tiv
e
ev
o
lu
tio
n
ar
y
g
r
a
n
u
lar
alg
o
r
ith
m
,
w
h
ic
h
s
u
p
p
o
r
ts
i
n
th
e
m
a
tch
i
n
g
o
f
i
m
ag
e
s
t
h
a
t
w
er
e
ta
k
en
p
r
io
r
an
d
later
to
p
last
ic
s
u
r
g
er
y
.
I
n
itiall
y
,
t
h
is
alg
o
r
it
h
m
d
o
es
t
h
e
g
en
er
atio
n
o
f
o
v
er
lap
p
in
g
f
ac
e
g
r
an
u
le
s
at
t
h
r
ee
-
lev
el
s
o
f
g
r
an
u
lar
it
y
.
P
last
ic
s
u
r
g
er
y
f
ac
e
r
ec
o
g
n
it
io
n
h
av
e
u
n
d
er
g
o
n
e
v
ar
io
u
s
d
ev
elo
p
m
e
n
ts
in
t
h
e
r
ec
en
t
p
ast.
T
h
e
r
es
e
ar
ch
co
n
tr
ib
u
tio
n
s
h
av
e
b
ee
n
r
ep
o
r
ted
in
th
e
lite
r
atu
r
e
eith
er
i
n
th
e
f
ea
t
u
r
e
ex
tr
ac
tio
n
p
h
aseo
r
in
t
h
e
cla
s
s
i
f
icatio
n
p
h
ase
o
r
i
n
b
o
th
th
e
p
h
ase
s
.
3.
F
E
AT
U
RE
E
XT
RAC
T
I
O
N
USI
N
G
E
V
-
SI
F
T
L
et
D
i
I
b
e
th
e
d
ata
b
ase
w
it
h
D
N
i
.......
2
,
1
an
d
j
F
b
e
th
e
f
ac
e
i
m
a
g
e
w
h
ic
h
s
h
o
u
ld
s
a
tis
f
y
th
e
co
n
d
itio
n
D
i
j
I
F
w
h
er
e
S
N
j
.....
2
,
1
.
T
h
e
p
r
e
-
p
r
o
ce
s
s
i
n
g
s
ta
g
e
s
tar
ts
w
it
h
i
m
a
g
e
r
esizi
n
g
.
L
et
t
h
e
s
ize
o
f
t
h
e
d
atab
ase
D
i
I
be
)
(
N
M
.
T
h
e
r
esize
m
o
d
el
o
f
i
m
a
g
e
is
r
ep
r
esen
ted
in
E
q
u
atio
n
(
1
)
w
h
er
e
M
S
an
d
M
S
r
ep
r
esen
ts
th
e
s
ca
led
n
u
m
b
er
o
f
r
o
w
s
a
n
d
co
lu
m
n
s
.
M
r
M
r
N
r
N
r
S
m
S
m
u
S
n
S
n
v
i
N
M
r
r
i
v
u
I
S
S
n
m
I
y
x
I
)
1
(
)
1
(
)
,
(
1
)
,
(
)
,
(
(
1
)
I
n
E
q
u
atio
n
(
1
)
,
]
,
1
[
M
u
an
d
]
,
1
[
N
v
,
1
0
r
r
M
m
an
d
1
0
r
r
N
n
,
th
e
ter
m
r
ep
r
esen
ts
th
e
r
o
u
n
d
-
o
f
f
f
u
n
c
tio
n
o
f
th
e
n
ea
r
est
i
n
te
g
er
,
)
(
r
r
N
M
is
th
e
s
ize
o
f
th
e
r
esized
i
m
a
g
e.
T
h
e
i
m
a
g
e
-
s
ca
le
r
ep
r
esen
tatio
n
o
f
t
w
o
d
i
m
e
n
s
io
n
al
i
m
ag
e
is
d
eter
m
i
n
ed
as
g
iv
e
n
i
n
E
q
u
atio
n
(
1
)
,
w
h
er
e
)
,
,
(
y
x
G
is
th
e
v
ar
iab
le
-
s
ca
le
Ga
u
s
s
i
an
f
u
n
ctio
n
,
)
,
(
y
x
is
th
e
s
p
atial
c
o
o
r
d
in
ate
an
d
is
th
e
s
ca
le
co
o
r
d
in
ate.
)
,
(
*
)
,
,
(
)
,
,
(
y
x
I
y
x
G
y
x
L
(
2
)
Her
e
ea
ch
s
a
m
p
le
p
o
in
t
o
f
t
h
e
i
m
a
g
e
co
m
p
ar
ed
w
it
h
2
6
n
e
ig
h
b
o
u
r
s
i
n
t
h
e
DOG
s
ca
le
s
p
ac
e
in
o
r
d
er
to
f
in
d
all
th
e
e
x
tr
e
m
e
p
o
i
n
ts
a
n
d
2
D
i
m
a
g
e
s
p
ac
e.
T
h
e
tar
g
et
p
o
in
t
is
n
ee
d
ed
to
co
m
p
ar
e
w
ith
8
n
eig
h
b
o
u
r
s
i
n
t
h
e
cu
r
r
en
t
i
m
a
g
e
an
d
al
s
o
w
i
th
1
8
n
ei
g
h
b
o
u
r
s
i
n
t
h
e
s
ca
le
ab
o
v
e
a
n
d
b
elo
w
.
T
h
e
tr
u
e
s
ca
le
in
v
ar
ia
n
ce
ca
n
b
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
EV
-
S
I
F
T
-
A
n
E
xten
d
ed
S
c
a
le
I
n
va
r
ia
n
t F
a
ce
R
ec
o
g
n
itio
n
fo
r
P
la
s
tic
S
u
r
g
ery
….
(
A
r
ch
a
n
a
H.
S
a
b
le
)
1925
esti
m
ated
f
r
o
m
t
h
e
n
o
r
m
al
iz
atio
n
o
f
L
ap
lacia
n
,
G
2
2
w
i
th
t
h
e
f
ac
to
r
2
.
Af
ter
co
n
v
o
l
u
tio
n
with
th
e
n
o
r
m
alize
d
L
ap
lacia
n
f
u
n
ctio
n
,
th
e
au
to
m
atic
s
ca
le
s
elec
tio
n
ca
n
b
e
p
er
f
o
r
m
ed
as
f
o
llo
w
s
,
)
,
,
(
)
,
(
*
)
,
,
(
2
2
2
2
2
2
y
x
L
y
x
I
G
y
x
O
(
3
)
T
h
e
o
u
tp
u
t
w
h
ic
h
is
ca
lcu
late
d
f
r
o
m
t
h
e
co
n
v
o
lu
tio
n
o
f
i
m
a
g
e
w
it
h
)
,
,
(
2
y
x
O
in
E
q
u
atio
n
(
3
)
w
il
l
b
e
at
ex
tr
e
m
u
m
if
th
e
s
ca
le
o
f
t
h
e
i
m
ag
e
s
tr
u
ctu
r
e
is
clo
s
e
to
t
h
e
v
al
u
e
o
f
th
e
n
o
r
m
alize
d
L
ap
lacia
n
f
u
n
ctio
n
.
T
h
e
p
o
in
ts
th
at
ar
e
e
x
tr
e
m
a
i
n
b
o
th
s
p
atia
l
an
d
s
ca
le
s
p
ac
es
s
h
o
u
ld
b
e
s
elec
ted
in
ca
s
e
o
f
d
etec
ti
n
g
th
e
b
lo
b
s
tr
u
ctu
r
e
a
n
d
s
e
lecti
n
g
t
h
e
m
at
m
o
s
t
o
p
ti
m
u
m
s
ca
le
.
T
h
e
n
ex
t
p
o
in
t
is
to
d
eter
m
i
n
e
th
e
k
e
y
p
o
in
t
s
at
th
e
s
ca
le
s
p
ac
e
ex
tr
e
m
e
in
t
h
e
d
if
f
er
en
ce
o
f
Gau
s
s
ia
n
f
u
n
ctio
n
e
x
p
r
ess
ed
w
i
th
t
h
e
i
m
a
g
e.
I
t
f
o
llo
w
s
s
o
m
e
s
tep
s
s
u
ch
a
s
,
a.
A
llo
ca
te
t
h
e
Or
ie
n
tatio
n
a
n
d
Gr
ad
ien
t M
o
d
u
les to
ea
ch
Ke
y
P
o
in
t
T
h
e
p
ar
am
eter
s
o
f
t
h
e
k
e
y
p
o
in
t
ar
e
co
m
p
lete
l
y
d
ep
en
d
s
o
n
th
e
d
is
tr
ib
u
tio
n
p
r
o
p
er
t
y
o
f
th
e
g
r
ad
ien
t
o
r
ien
tatio
n
o
f
t
h
e
i
m
ag
e
w
h
ic
h
is
ar
o
u
n
d
t
h
e
k
e
y
p
o
in
ts
.
So
th
e
o
r
ien
ta
tio
n
a
n
d
g
r
ad
ie
n
t
m
o
d
u
les
o
f
t
h
e
k
e
y
p
o
in
t
ar
e
ca
lcu
lated
w
h
ic
h
r
eg
is
ter
s
in
v
ar
ia
n
ce
to
i
m
a
g
e
r
o
tatio
n
.
T
h
e
g
r
ad
ien
t
m
a
g
n
itu
d
e
an
d
o
r
ien
tatio
n
is
ca
lcu
lated
f
o
r
ea
ch
i
m
a
g
e
s
a
m
p
le.
b.
Vo
lu
m
e
b
ased
Feat
u
r
e
D
escr
i
p
to
r
V
-
SIFT
f
ea
t
u
r
e
is
s
a
m
e
as
S
I
FT
w
h
er
e
it
u
s
e
s
t
h
e
v
o
lu
m
e
o
f
th
e
s
tr
u
ctu
r
e
i
n
s
tead
o
f
tak
i
n
g
th
e
co
n
tr
ast
as
i
n
n
o
r
m
al
SI
FT
f
ea
tu
r
e.
I
n
t
h
e
n
o
r
m
al
SIFT
f
ea
t
u
r
e,
th
e
i
m
a
g
e
s
ca
le
s
p
ac
e
ex
t
r
e
m
e
is
d
etec
ted
i
n
th
e
f
ir
s
t
s
tep
.
Fro
m
t
h
at
it
i
s
k
n
o
w
n
t
h
at
t
h
e
L
ap
lacia
n
is
a
s
ec
o
n
d
-
o
r
d
er
d
er
iv
ativ
e
w
h
ic
h
is
v
er
y
s
e
n
s
i
tiv
e
to
n
o
is
e.
V
-
SIFT
[
2
8
]
[
2
9
]
ar
e
im
p
le
m
en
ted
to
o
v
er
co
m
e
th
i
s
l
i
m
itat
io
n
an
d
also
it
ca
n
co
m
p
letel
y
r
e
m
o
v
e
t
h
e
u
n
r
eliab
le
k
e
y
p
o
in
ts
.
I
n
t
h
is
te
ch
n
iq
u
e
h
er
e
is
s
o
m
e
u
n
iq
u
e
m
ax
i
m
u
m
o
v
er
s
ca
le
s
in
,
2
2
2
1
2
2
2
2
2
)
(
,
0
,
0
(
)
,
0
,
0
(
L
O
(
4
)
I
n
t
h
is
m
et
h
o
d
t
h
e
v
a
lu
e
o
f
)
,
,
(
2
y
x
O
in
E
q
u
atio
n
(
4
)
s
h
o
u
ld
b
e
ca
lcu
lated
at
ea
c
h
k
e
y
p
o
in
t
lo
ca
tio
n
an
d
if
t
h
e
v
alu
e
i
s
b
el
o
w
t
h
e
th
r
es
h
o
ld
,
th
e
co
r
r
esp
o
n
d
in
g
k
e
y
p
o
in
t
s
ar
e
r
e
m
o
v
ed
.
I
n
th
e
n
o
r
m
al
s
h
i
f
t
alg
o
r
ith
m
,
t
h
e
k
e
y
p
o
in
t
s
ar
e
r
e
m
o
v
ed
b
ased
o
n
t
h
e
co
n
tr
ast
v
alu
e
at
t
h
e
lo
ca
tio
n
o
f
t
h
e
k
e
y
p
o
in
t
o
f
t
h
e
i
m
a
g
e.
B
u
t in
V
-
SI
FT
,
it is
b
a
s
ed
o
n
th
e
v
o
l
u
m
e
o
f
t
h
e
s
tr
u
c
tu
r
e
w
h
ic
h
is
e
s
ti
m
ated
as
f
o
ll
o
w
s
,
)
,
(
*
)
,
,
(
)
,
,
(
2
4
2
2
2
y
x
I
G
y
x
O
y
x
V
(
5
)
I
n
E
q
u
atio
n
(
5
)
,
is
th
e
s
ca
le
o
f
th
e
co
r
r
esp
o
n
d
in
g
k
e
y
p
o
in
t.
W
e
ca
n
d
er
iv
e
th
e
ab
o
v
e
eq
u
a
tio
n
as,
)
,
,
(
(
)
,
,
(
2
1
1
2
1
2
1
1
y
x
O
y
x
V
(
6
)
)
,
,
(
(
)
,
,
(
2
2
2
2
2
2
2
2
y
x
O
y
x
V
(
7
)
Fro
m
E
q
u
atio
n
(
6
)
an
d
(
7
)
,
t
h
e
ex
tr
e
m
a
v
al
u
e
o
f
2
1
1
,
,
(
y
x
O
)
an
d
)
,
,
(
2
2
1
y
x
O
is
s
a
m
e
b
u
t
th
e
ex
tr
e
m
a
o
f
)
,
,
(
2
1
1
y
x
V
an
d
)
,
,
(
2
2
2
y
x
V
ar
e
d
if
f
er
en
t.
T
h
er
e
is
a
r
u
le
i
n
th
is
ca
s
e
w
h
ic
h
is
i
n
E
q
u
atio
n
(
1
4
)
.
e
x
t
r
e
m
a
e
x
t
r
e
m
a
y
x
V
y
x
V
)
,
,
(
)
,
,
(
2
2
2
2
1
1
(
8
)
I
n
V
-
SIFT
ap
p
r
o
ac
h
all
ca
s
es,
f
ir
s
t
w
e
s
h
o
u
ld
ev
al
u
ate
t
h
e
v
a
lu
e
o
f
)
,
,
(
2
y
x
V
at
ea
ch
ca
n
d
id
ate
k
e
y
p
o
in
t lo
ca
tio
n
an
d
r
e
m
o
v
e
if
it
is
b
elo
w
s
o
m
e
th
r
e
s
h
o
ld
v
al
u
e.
c.
E
n
tr
o
p
y
b
ased
Feat
u
r
e
Descr
i
p
to
r
E
n
tr
o
p
y
i
s
th
e
m
ea
s
u
r
e
o
f
u
n
p
r
ed
ictab
ilit
y
o
f
t
h
e
co
n
te
n
t
o
f
th
e
i
n
f
o
r
m
atio
n
i
n
an
i
m
ag
e
.
I
t
is
th
e
s
tatis
t
ical
m
ea
s
u
r
e
o
f
r
an
d
o
m
n
es
s
w
h
i
c
h
ca
n
b
e
u
s
ed
to
c
h
a
r
ac
ter
ize
th
e
te
x
t
u
r
e
o
f
t
h
e
i
n
p
u
t
i
m
a
g
e.
Si
n
ce
th
e
en
tr
o
p
y
is
th
e
h
ig
h
er
o
r
d
er
s
tatis
tical
f
ea
tu
r
e
it
ca
n
h
av
e
least
ef
f
ec
t
o
n
u
n
ce
r
tai
n
v
ar
i
atio
n
s
i
n
th
e
f
ac
e.
E
n
tr
o
p
y
b
ased
f
ea
tu
r
e
d
escr
ip
t
o
r
is
d
em
o
n
s
tr
ated
in
f
o
llo
w
i
n
g
s
tep
s
.
St
ep
1
:
T
h
e
v
o
lu
m
e
o
f
t
h
e
i
m
a
g
e
is
ca
lcu
lated
as
i
n
V
-
SIFT
f
o
r
m
u
latio
n
an
d
i
t
w
ill
b
e
in
t
h
e
f
o
r
m
o
f
m
atr
i
x
w
h
ic
h
is
d
eter
m
i
n
ed
in
E
q
u
atio
n
(
9
)
.
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.
7
,
No
.
4
,
A
u
g
u
s
t
2017
:
1
9
2
3
–
1
9
3
3
1926
)
,
(
j
i
V
)
,
(
.
)
,
(
)
,
(
.
.
.
.
)
,
(
.
)
,
(
)
,
(
)
,
(
.
)
,
(
)
,
(
2
1
2
2
2
1
2
1
2
1
1
1
n
m
m
m
n
n
j
i
v
j
i
v
j
i
v
j
i
v
j
i
v
j
i
v
j
i
v
j
i
v
j
i
v
(
9
)
Step
2
:
T
h
e
in
f
o
r
m
atio
n
s
o
u
r
ce
is
s
ta
tio
n
ar
y
an
d
m
e
m
o
r
y
l
ess
.
T
h
e
v
o
l
u
m
e
o
f
t
h
e
s
tr
u
ct
u
r
e
in
E
V
-
SIFT
an
al
y
s
i
s
w
h
ich
i
s
th
e
p
r
o
b
ab
ilit
y
f
u
n
ctio
n
i
s
r
ep
r
esen
te
d
in
E
q
u
atio
n
(
1
0
)
.
i
j
p
j
i
V
j
i
V
j
i
V
)
,
(
)
,
(
)
,
(
(
1
0
)
Step
3
:
Fro
m
t
h
e
v
o
l
u
m
e
o
f
t
h
e
s
tr
u
ctu
r
e,
t
h
e
en
tr
o
p
y
is
ca
lcu
lated
.
T
h
e
en
tr
o
p
y
ca
lc
u
lat
io
n
f
o
r
th
e
EV
-
SIFT
p
r
o
ce
s
s
is
d
eter
m
in
e
d
as f
o
llo
w
s
,
)
,
(
l
o
g
)
,
(
)
(
j
i
V
j
i
V
V
E
p
p
(
1
1
)
I
n
E
q
u
atio
n
(
1
1
)
,
if
)
(
V
E
is
h
i
g
h
en
tr
o
p
y
,
th
e
n
th
e
v
o
lu
m
e
is
f
r
o
m
u
n
i
f
o
r
m
d
is
tr
ib
u
tio
n
a
n
d
if
)
(
V
E
is
lo
w
e
n
tr
o
p
y
,
t
h
en
t
h
e
v
o
l
u
m
e
i
s
f
r
o
m
v
ar
ied
d
i
s
tr
ib
u
ti
o
n
.
T
h
er
ef
o
r
e,
th
e
o
b
tain
ed
f
i
n
al
E
V
-
SIFT
d
escr
ip
to
r
o
f
th
e
w
h
o
le
d
ata
b
ase
is
d
en
o
ted
as
D
i
F
.
T
h
e
g
r
ad
i
en
t
m
ag
n
it
u
d
es
a
n
d
o
r
ien
tatio
n
s
alo
n
g
w
i
th
v
o
lu
m
e
an
d
e
n
tr
o
p
y
d
escr
ip
to
r
s
o
f
th
e
i
m
a
g
e
is
s
a
m
p
led
ar
o
u
n
d
th
e
co
r
r
esp
o
n
d
in
g
k
e
y
p
o
in
t
lo
ca
tio
n
u
s
i
n
g
th
e
s
ca
le
o
f
k
e
y
p
o
in
t
s
in
o
r
d
er
to
s
elec
t
th
e
lev
el
o
f
Ga
u
s
s
ian
b
l
u
r
o
f
th
e
i
m
ag
e.
Fir
s
t
s
a
m
p
le
w
i
th
8
*
8
n
eig
h
b
o
u
r
w
in
d
o
w
w
h
ich
is
c
en
tr
ed
o
n
th
e
k
e
y
p
o
in
t
a
n
d
th
en
d
iv
id
e
th
e
n
ei
g
h
b
o
u
r
in
to
4
*
4
ch
ild
w
in
d
o
w
.
Fro
m
t
h
is
ea
c
h
c
h
ild
w
i
n
d
o
w,
ca
lcu
late
t
h
e
g
r
ad
ien
t
o
r
ien
t
atio
n
h
is
to
g
r
a
m
w
it
h
eig
h
t
b
in
s
.
E
ac
h
d
e
s
cr
ip
to
r
co
n
tain
s
a
4
*
4
ar
r
a
y
o
f
h
is
to
g
r
a
m
s
ar
o
u
n
d
ea
c
h
k
e
y
p
o
in
t
an
d
ea
ch
h
i
s
to
g
r
a
m
co
n
tai
n
s
8
b
in
s
.
T
h
er
ef
o
r
e
4
*
4
*
8
=1
2
8
d
i
m
e
n
s
io
n
i
s
o
b
tain
ed
w
h
ich
is
th
e
f
ea
t
u
r
e
v
ec
to
r
.
T
h
e
s
tr
u
ctu
r
e
o
f
t
h
e
p
r
o
p
o
s
ed
r
ec
o
g
n
it
io
n
s
y
s
te
m
i
s
ex
p
lai
n
ed
in
Fig
u
r
e
1
.
Fig
u
r
e
1
.
B
lo
ck
d
iag
r
a
m
o
f
E
V
-
SIFT
b
ased
p
last
ic
s
u
r
g
er
y
f
ac
e
r
ec
o
g
n
i
tio
n
Scal
e
spa
ce
e
xt
r
em
e
det
e
ct
i
on
D
i
st
i
nc
t
po
i
nt
s
DOG
sc
al
e
spa
c
e
K
ey
poi
n
t
ext
r
ac
t
i
on
Edg
e poi
nt
s
el
i
m
i
na
t
i
on
Extr
em
e
poi
n
t
s
el
i
m
i
na
t
i
on
O
r
i
en
t
a
t
i
on
and g
r
ad
i
en
t
m
odul
e t
o e
ac
h k
ey
poi
nt
SV
M
cl
as
s
i
f
i
er
Post
s
ur
g
er
y
t
es
t
i
ng
R
ec
og
ni
z
ed f
ac
e
D
et
er
m
i
ni
ng
v
ol
um
e
pai
r
s
V
ol
um
e
f
unc
t
i
on
Entr
opy
f
unc
t
i
on
Surr
ou
ndi
ng
pat
ch
es
Pre
su
r
g
er
y
t
r
a
i
n
i
ng
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
EV
-
S
I
F
T
-
A
n
E
xten
d
ed
S
c
a
le
I
n
va
r
ia
n
t F
a
ce
R
ec
o
g
n
itio
n
fo
r
P
la
s
tic
S
u
r
g
ery
….
(
A
r
ch
a
n
a
H.
S
a
b
le
)
1927
4.
RE
SU
L
T
AND
DI
SCUS
SI
O
N
4
.
1
.
Ex
peri
m
e
nta
l Set
up
T
h
e
ex
p
er
i
m
e
n
t
o
f
f
ac
e
r
ec
o
g
n
itio
n
i
s
co
n
d
u
cted
u
s
i
n
g
p
r
e
-
s
u
r
g
er
y
a
n
d
p
o
s
t
s
u
r
g
er
y
f
ac
es
.
T
h
e
d
ata
b
ase
f
o
r
t
h
e
p
r
e
-
s
u
r
g
er
y
a
n
d
p
o
s
t
s
u
r
g
er
y
f
ac
es
ar
e
d
o
w
n
lo
ad
ed
f
r
o
m
th
e
UR
L
:
h
ttp
://
www
.
lo
ca
tead
o
c.
co
m
/p
i
ctu
r
es/.
I
n
t
h
e
co
r
r
esp
o
n
d
in
g
d
ata
b
ase,
th
e
p
r
e
-
s
u
r
g
er
y
a
n
d
p
o
s
t
s
u
r
g
er
y
f
ac
e
s
o
f
5
1
5
p
er
s
o
n
s
ar
e
p
r
esen
t.
T
h
e
s
a
m
p
le
i
m
ag
e
s
o
f
t
h
e
d
atab
ase
ar
e
s
h
o
w
n
i
n
F
i
g
u
r
e
2
.
Fig
u
r
e
2
.
Sa
m
p
le
i
m
ag
e
s
o
f
t
h
e
d
ata
b
ase
(
a)
P
r
e
-
s
u
r
g
er
y
f
ac
es (
b
)
Po
s
t su
r
g
er
y
f
ac
e
s
T
h
e
p
r
e
-
s
u
r
g
er
y
f
ac
es
ar
e
tak
en
as
th
e
tr
ai
n
i
n
g
f
ac
es
an
d
p
o
s
t
s
u
r
g
er
y
f
ac
es
ar
e
tak
e
n
as
test
i
n
g
f
ac
es.
T
h
e
E
V
-
s
i
f
t
f
ea
t
u
r
e
is
c
alcu
lated
f
o
r
b
o
th
th
e
tr
ain
i
n
g
an
d
test
i
n
g
f
ac
es.
T
h
en
t
h
e
s
e
f
ea
tu
r
es
ar
e
ap
p
lied
to
SVM
clas
s
i
f
ier
an
d
i
t
g
iv
e
s
th
e
o
p
ti
m
u
m
m
a
tch
i
n
g
o
f
t
h
e
co
n
ce
r
n
ed
f
ac
e
s
.
T
h
e
clas
s
if
ica
tio
n
i
s
ca
r
r
ied
b
ased
o
n
th
e
p
ar
a
m
eter
s
s
u
c
h
as
T
r
u
e
P
o
s
itiv
e
(
T
P
)
,
T
r
u
e
Neg
ati
v
e
(
T
N)
,
Fals
e
P
o
s
itiv
e
(
FP
)
,
an
d
Fals
e
Neg
ati
v
e
(
FN)
.
U
s
i
n
g
th
e
s
e
p
a
r
a
m
eter
s
,
t
h
e
p
er
f
o
r
m
a
n
ce
m
e
asu
r
es
s
u
ch
as
A
cc
u
r
ac
y
,
Se
n
s
itiv
it
y
,
Sp
ec
if
ici
t
y
,
P
r
ec
is
io
n
,
Fals
e
P
o
s
iti
v
e
R
ate
(
FP
R
)
,
Fals
e
Ne
g
ati
v
e
R
ate
(
FNR
)
,
Ne
g
ativ
e
P
r
ed
ictio
n
Valu
e
(
NP
V)
,
Fals
e
Dis
co
v
er
y
R
ate
(
FD
R
)
,
F1
_
Sco
r
e
an
d
Ma
tth
e
w
s
C
o
r
r
elatio
n
C
o
ef
f
icie
n
t
(
MC
C
)
.
T
h
e
clea
r
d
escr
ip
tio
n
o
f
th
e
an
al
y
s
is
i
n
S
VM
class
if
ier
is
e
x
p
lain
ed
i
n
th
e
n
ex
t sectio
n
.
4
.
2
.
Rec
o
g
nizin
g
Su
rg
er
y
P
o
rt
io
ns
T
h
e
p
las
tic
s
u
r
g
er
y
p
o
r
tio
n
s
a
r
e
r
ec
o
g
n
ized
b
y
ca
lc
u
lati
n
g
t
h
e
SIFT
f
ea
tu
r
es.
Her
e
th
e
f
e
atu
r
es
ar
e
d
eter
m
in
ed
f
o
r
t
h
e
p
r
e
s
u
r
g
er
y
a
n
d
p
o
s
t
s
u
r
g
er
y
f
ac
e
s
.
I
n
t
h
is
d
eter
m
i
n
atio
n
,
m
an
y
k
e
y
p
o
in
ts
ar
e
o
b
tai
n
ed
.
Fu
r
t
h
er
b
o
th
th
e
p
r
e
s
u
r
g
er
y
an
d
p
o
s
t
s
u
r
g
er
y
f
ac
e
s
ar
e
m
atch
ed
b
y
t
h
e
s
titc
h
in
g
p
r
o
ce
s
s
.
Her
e
th
e
p
o
in
t
s
w
h
ic
h
ar
e
n
o
t
m
atc
h
ed
ar
e
tak
en
w
h
ic
h
ca
n
b
e
co
n
s
id
er
ed
as
th
e
p
last
ic
s
u
r
g
er
y
p
o
r
tio
n
s
o
f
t
h
e
f
ac
e.
T
h
e
r
ec
o
g
n
ized
p
o
r
tio
n
o
f
p
l
asti
c
s
u
r
g
er
y
f
ac
e
is
s
h
o
w
n
i
n
F
i
g
u
r
e
3
.
Fig
u
r
e
3
.
R
ec
o
g
n
iz
in
g
p
last
ic
s
u
r
g
er
y
p
o
r
tio
n
s
4
.
3
.
Sta
t
is
t
ica
l A
na
ly
s
is
T
h
e
s
tatis
tical
an
al
y
s
is
o
f
th
e
p
last
ic
s
u
r
g
er
y
f
ac
e
r
ec
o
g
n
iti
o
n
d
escr
ib
es
th
e
co
m
p
ar
is
o
n
o
f
f
ea
t
u
r
es
s
u
c
h
as
P
r
in
cip
le
C
o
m
p
o
n
e
n
t
An
al
y
s
i
s
(
P
C
A
)
,
SIFT
,
Vo
lu
m
e
SIFT
an
d
th
e
p
r
o
p
o
s
ed
m
eth
o
d
E
V
-
SIFT
.
T
h
e
p
er
f
o
r
m
a
n
ce
m
ea
s
u
r
es
ar
e
a
n
a
l
y
s
ed
f
o
r
all
th
e
s
e
f
ea
t
u
r
es.
T
h
e
an
a
l
y
s
is
o
f
t
h
e
clas
s
i
f
ier
s
s
u
ch
a
s
li
n
ea
r
SVM,
q
u
ad
r
atic
SVM,
R
B
F
SVM
a
n
d
ML
P
SVM
f
o
r
b
ef
o
r
e
an
d
af
ter
p
la
s
tic
s
u
r
g
er
y
i
s
ill
u
s
tr
ate
d
in
T
ab
le
1
,
T
ab
le
2
,
T
ab
le
3
a
n
d
T
ab
le
4
.
T
h
e
r
an
k
i
n
g
o
f
ea
c
h
m
ea
s
u
r
e
is
m
e
n
tio
n
ed
i
n
b
r
ac
k
et.
Se
n
s
i
tiv
i
t
y
is
th
e
m
ea
s
u
r
e
o
f
th
e
m
et
h
o
d
to
co
r
r
ec
tly
id
en
ti
f
y
t
h
e
p
o
s
iti
v
e
s
a
m
p
le
s
w
h
i
le
th
e
s
en
s
iti
v
it
y
is
t
h
e
m
ea
s
u
r
e
o
f
th
e
m
et
h
o
d
to
co
r
r
ec
tly
id
e
n
ti
f
y
th
e
n
e
g
ati
v
e
s
a
m
p
le
s
.
P
r
ec
is
io
n
ca
n
g
i
v
e
t
h
e
r
atio
o
f
p
o
s
iti
v
e
ag
a
in
s
t
all
th
e
p
o
s
iti
v
e
r
es
u
lts
.
(
a)
(
b
)
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.
7
,
No
.
4
,
A
u
g
u
s
t
2017
:
1
9
2
3
–
1
9
3
3
1928
FP
R
,
FNR
,
NP
V
an
d
FDR
ca
n
co
r
r
ec
tly
p
r
ed
ict
th
e
in
co
r
r
ec
t
id
en
tif
icat
io
n
an
d
co
r
r
ec
t
id
en
tif
icat
io
n
.
T
h
e
co
r
r
ec
tn
ess
o
f
th
e
cla
s
s
i
f
icatio
n
alg
o
r
it
h
m
a
n
d
th
e
ef
f
icac
y
o
f
b
in
ar
y
cla
s
s
clas
s
i
f
icat
io
n
ca
n
b
e
d
eter
m
i
n
ed
b
y
F1
_
Sco
r
e
an
d
MCC
.
I
n
T
ab
le
1
an
d
T
ab
le
2
,
w
h
ic
h
i
s
t
h
e
lin
ea
r
SVM
an
d
q
u
ad
r
atic
SVM,
th
e
ac
cu
r
ac
y
i
s
b
e
tter
f
o
r
t
h
e
P
C
A
w
h
ile
th
e
s
en
s
iti
v
it
y
a
n
d
th
e
s
p
ec
i
f
icit
y
ar
e
b
etter
f
o
r
t
h
e
E
V
-
SIFT
f
ea
t
u
r
e
f
o
r
p
last
i
c
s
u
r
g
er
y
f
ac
es.
B
u
t
h
er
e
all
th
e
m
ea
s
u
r
es a
r
e
b
ette
r
f
o
r
t
h
e
E
V
-
SIFT
f
ea
t
u
r
e.
T
h
e
r
an
k
in
g
o
f
all
t
h
e
m
ea
s
u
r
es
i
s
ca
lc
u
lated
a
n
d
th
e
f
i
n
al
r
an
k
is
b
est
f
o
r
E
V
SIFT
Featu
r
e
w
h
en
co
m
p
ar
ed
to
th
e
o
th
er
f
ea
t
u
r
e
ex
tr
ac
tio
n
m
et
h
o
d
s
in
li
n
ea
r
SV
M
an
d
q
u
ad
r
atic
SVM.
I
n
tab
le
I
I
I
,
it
d
esc
r
ib
es
th
e
an
al
y
s
is
f
o
r
R
B
F
SVM.
Her
e
all
th
e
m
ea
s
u
r
es
ar
e
b
etter
f
o
r
P
C
A
,
SIFT
AND
V
-
SI
FT
w
h
ile
t
h
e
p
r
o
p
o
s
ed
EV
-
SIFT
f
e
atu
r
e
s
h
o
w
s
les
s
p
er
f
o
r
m
an
ce
.
B
y
a
n
al
y
s
i
n
g
t
h
e
r
an
k
,
P
C
A
is
b
etter
th
a
n
o
th
er
m
e
th
o
d
s
in
R
B
F
SVM.
I
n
T
a
b
le
4
,
all
th
e
m
ea
s
u
r
es
s
h
o
w
b
ette
r
p
er
f
o
r
m
an
c
e
w
h
ile
u
s
i
n
g
E
V
-
SI
FT
f
ea
tu
r
e
.
So
b
y
ex
a
m
i
n
i
n
g
t
h
e
o
v
er
a
ll
an
al
y
s
i
s
,
it
is
clea
r
th
at
t
h
e
E
V
-
SIFT
f
ea
tu
r
e
ex
tr
ac
tio
n
i
s
b
etter
f
o
r
th
e
p
las
tic
s
u
r
g
er
y
f
ac
e
r
ec
o
g
n
i
tio
n
p
u
r
p
o
s
e.
T
ab
le
1
.
E
x
p
er
im
e
n
tal
E
v
al
u
at
io
n
o
n
SVM
w
it
h
L
i
n
ea
r
Ker
n
el
Fu
n
ctio
n
W
i
t
h
o
u
t
P
l
a
st
i
c
S
u
r
g
e
r
y
W
i
t
h
P
l
a
st
i
c
S
u
r
g
e
r
y
P
C
A
S
I
F
T
V
S
I
F
T
EV
S
I
F
T
P
C
A
S
I
F
T
V
S
I
F
T
EV
S
I
F
T
A
c
c
u
r
a
c
y
0
.
8
3
(
1
)
0
.
7
5
(
3
)
0
.
6
9
(
4
)
0
.
8
2
(
2
)
0
.
8
8
(
1
)
0
.
8
1
(
2
)
0
.
8
(
3
)
0
.
7
9
(
4
)
S
e
n
si
t
i
v
i
t
y
0
.
1
8
(
4
)
0
.
2
7
(
2
)
0
.
2
3
(
3
)
0
.
3
3
(
1
)
0
.
1
7
(
3
)
0
.
2
0
(
2
)
0
.
1
7
(
4
)
0
.
2
3
(
1
)
S
p
e
c
i
f
i
c
i
t
y
0
.
9
1
(
4
)
0
.
9
9
(
1
)
0
.
9
9
(
2
)
0
.
9
7
(
3
)
0
.
9
4
(
4
)
0
.
9
7
(
2
)
0
.
9
6
(
3
)
0
.
9
9
(
1
)
P
r
e
c
i
si
o
n
0
.
2
(
4
)
0
.
9
(
1
)
0
.
9
(
1
)
0
.
8
(
3
)
0
.
2
(
4
)
0
.
6
(
2
)
0
.
5
3
(
3
)
0
.
9
3
(
1
)
F
P
R
0
.
0
9
(
4
)
0
.
0
1
(
1
)
0
.
0
2
(
2
)
0
.
0
3
(
3
)
0
.
0
6
(
4
)
0
.
0
3
(
2
)
0
.
0
4
(
3
)
0
.
0
1
(
1
)
F
N
R
0
.
8
2
(
4
)
0
.
7
2
(
2
)
0
.
7
7
(
3
)
0
.
6
7
(
1
)
0
.
8
2
(
3
)
0
.
8
0
(
2
)
0
.
8
3
(
4
)
0
.
7
7
(
1
)
N
P
V
0
.
9
1
(
4
)
0
.
9
9
(
1
)
0
.
9
8
(
2
)
0
.
9
7
(
3
)
0
.
9
4
(
4
)
0
.
9
7
(
2
)
0
.
9
6
(
3
)
0
.
9
9
(
1
)
F
D
R
0
.
8
(
4
)
0
.
1
(
1
)
0
.
1
(
1
)
0
.
2
(
3
)
0
.
8
(
4
)
0
.
4
(
2
)
0
.
4
7
(
3
)
0
.
0
7
(
1
)
F
1
_
S
c
o
r
e
0
.
1
9
(
4
)
0
.
4
1
(
2
)
0
.
3
7
(
3
)
0
.
4
7
(
1
)
0
.
1
9
(
4
)
0
.
3
0
(
2
)
0
.
2
6
(
3
)
0
.
3
7
(
1
)
M
C
C
0
.
1
0
(
4
)
0
.
4
0
2
)
0
.
3
5
(
3
)
0
.
4
4
(
1
)
0
.
1
3
(
4
)
0
.
2
6
(
2
)
0
.
2
2
(
3
)
0
.
4
0
(
1
)
A
v
g
_
r
a
n
k
3
.
7
1
.
6
2
.
4
2
.
1
3
.
5
2
3
.
2
1
.
3
F
i
n
a
l
_
r
a
n
k
4
1
3
2
4
2
3
1
T
ab
le
2
.
E
x
p
er
im
e
n
tal
E
v
al
u
at
io
n
o
n
SVM
w
it
h
Q
u
ad
r
atic
Ker
n
el
F
u
n
ct
io
n
W
i
t
h
o
u
t
P
l
a
st
i
c
S
u
r
g
e
r
y
W
i
t
h
P
l
a
st
i
c
S
u
r
g
e
r
y
P
C
A
S
I
F
T
V
S
I
F
T
EV
S
I
F
T
P
C
A
S
I
F
T
V
S
I
F
T
EV
S
I
F
T
A
c
c
u
r
a
c
y
0
.
7
5
(
4
)
0
.
8
2
(
2
)
0
.
8
6
(
1
)
0
.
8
2
(
2
)
0
.
8
8
(
1
)
0
.
8
3
(
4
)
0
.
8
6
(
3
)
0
.
8
7
(
2
)
S
e
n
si
t
i
v
i
t
y
0
.
1
1
(
4
)
0
.
3
5
(
2
)
0
.
4
(
1
)
0
.
3
2
(
3
)
0
.
0
7
(
4
)
0
.
1
3
(
2
)
0
.
1
3
(
3
)
0
.
2
2
(
1
)
S
p
e
c
i
f
i
c
i
t
y
0
.
9
0
(
4
)
0
.
9
9
(
1
)
0
.
9
8
(
2
)
0
.
9
6
(
3
)
0
.
9
3
(
4
)
0
.
9
4
(
2
)
0
.
9
4
(
3
)
0
.
9
5
(
1
)
P
r
e
c
i
si
o
n
0
.
2
(
4
)
0
.
9
(
1
)
0
.
8
(
2
)
0
.
7
(
3
)
0
.
0
7
(
4
)
0
.
2
7
(
2
)
0
.
2
(
3
)
0
.
4
(
1
)
F
P
R
0
.
0
9
(
4
)
0
.
0
1
(
1
)
0
.
0
2
(
2
)
0
.
0
4
(
3
)
0
.
0
7
(
4
)
0
.
0
6
(
2
)
0
.
0
6
(
3
)
0
.
0
4
(
1
)
F
N
R
0
.
8
9
(
4
)
0
.
6
5
(
1
)
0
.
6
(
2
)
0
.
6
8
(
3
)
0
.
9
3
(
4
)
0
.
8
8
(
2
)
0
.
8
7
(
3
)
0
.
7
8
(
1
)
N
P
V
0
.
9
0
(
4
)
0
.
9
9
(
2
)
0
.
9
8
(
1
)
0
.
9
6
(
3
)
0
.
9
3
(
4
)
0
.
9
4
(
2
)
0
.
9
4
(
3
)
0
.
9
5
(
1
)
F
D
R
0
.
8
(
4
)
0
.
1
(
1
)
0
.
2
(
2
)
0
.
3
(
3
)
0
.
9
3
(
4
)
0
.
7
3
(
2
)
0
.
8
(
3
)
0
.
6
(
1
)
F
1
_
S
c
o
r
e
0
.
1
4
(
4
)
0
.
5
(
2
)
0
.
5
3
(
1
)
0
.
4
4
(
3
)
0
.
0
7
(
4
)
0
.
1
7
(
2
)
0
.
1
6
(
3
)
0
.
2
9
(
1
)
M
C
C
0
.
0
1
(
4
)
0
.
4
9
(
2
)
0
.
5
(
1
)
0
.
3
9
(
3
)
0
(
4
)
0
.
1
0
2
)
0
.
0
9
(
3
)
0
.
2
3
(
1
)
A
v
g
_
r
a
n
k
4
1
.
5
1
.
5
2
.
9
3
.
7
2
.
4
2
.
8
1
.
1
F
i
n
a
l
_
r
a
n
k
4
1
2
3
4
2
3
1
T
ab
le
3
.
E
x
p
er
im
e
n
tal
E
v
al
u
at
io
n
o
n
SVM
w
it
h
R
B
F Ke
r
n
el
Fu
n
ctio
n
W
i
t
h
o
u
t
P
l
a
st
i
c
S
u
r
g
e
r
y
W
i
t
h
P
l
a
st
i
c
S
u
r
g
e
r
y
P
C
A
S
I
F
T
V
S
I
F
T
EV
S
I
F
T
P
C
A
S
I
F
T
V
S
I
F
T
EV
S
I
F
T
A
c
c
u
r
a
c
y
0
.
9
1
(
4
)
0
.
9
2
(
2
)
0
.
9
4
(
1
)
0
.
9
2
(
2
)
0
.
8
7
(
1
)
0
.
8
3
(
4
)
0
.
8
6
(
3
)
0
.
8
6
(
2
)
S
e
n
si
t
i
v
i
t
y
1
(
1
)
1
(
1
)
1
(
1
)
1
(
1
)
0
.
0
7
(
4
)
0
.
1
3
(
2
)
0
.
1
3
(
3
)
0
.
2
2
1
)
S
p
e
c
i
f
i
c
i
t
y
0
.
9
1
(
4
)
0
.
9
2
(
2
)
0
.
9
4
(
1
)
0
.
9
2
(
2
)
0
.
9
3
(
4
)
0
.
9
4
(
2
)
0
.
9
4
(
3
)
0
.
9
5
(
1
)
P
r
e
c
i
si
o
n
0
.
1
(
4
)
0
.
2
(
2
)
0
.
4
(
1
)
0
.
2
(
2
)
0
.
0
7
(
4
)
0
.
2
7
(
2
)
0
.
2
(
3
)
0
.
4
(
1
)
F
P
R
0
.
0
9
(
4
)
0
.
0
8
(
2
)
0
.
0
6
(
1
)
0
.
0
8
(
2
)
0
.
0
7
(
4
)
0
.
0
6
(
2
)
0
.
0
6
(
3
)
0
.
0
4
(
1
)
F
N
R
0
(
1
)
0
(
1
)
0
(
1
)
0
(
1
)
0
.
9
3
(
4
)
0
.
8
8
(
2
)
0
.
8
7
(
3
)
0
.
7
8
(
1
)
N
P
V
0
.
9
0
(
4
)
0
.
9
2
(
2
)
0
.
9
4
(
1
)
0
.
9
2
(
2
)
0
.
9
3
(
4
)
0
.
9
4
(
2
)
0
.
9
4
(
3
)
0
.
9
5
(
1
)
F
D
R
0
.
9
(
4
)
0
.
8
(
2
)
0
.
6
(
1
)
0
.
8
(
2
)
0
.
9
3
(
4
)
0
.
7
3
(
2
)
0
.
8
(
3
)
0
.
6
(
1
)
F
1
_
S
c
o
r
e
0
.
1
8
(
4
)
0
.
3
3
(
2
)
0
.
5
7
(
1
)
0
.
3
3
(
2
)
0
.
0
7
(
4
)
0
.
1
7
(
2
)
0
.
1
6
(
3
)
0
.
2
8
(
1
)
M
C
C
0
.
3
0
(
4
)
0
.
4
3
(
2
)
0
.
6
1
(
1
)
0
.
4
3
(
2
)
0
(
4
)
0
.
0
9
(
2
)
0
.
0
9
(
3
)
0
.
2
3
(
1
)
A
v
g
_
r
a
n
k
3
.
4
1
.
8
1
1
.
8
3
.
7
2
.
4
2
.
8
1
.
1
F
i
n
a
l
_
r
a
n
k
4
2
1
3
4
2
3
1
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
EV
-
S
I
F
T
-
A
n
E
xten
d
ed
S
c
a
le
I
n
va
r
ia
n
t F
a
ce
R
ec
o
g
n
itio
n
fo
r
P
la
s
tic
S
u
r
g
ery
….
(
A
r
ch
a
n
a
H.
S
a
b
le
)
1929
T
ab
le
4
.
E
x
p
er
im
e
n
tal
E
v
al
u
at
io
n
o
n
SVM
w
it
h
M
L
P
Ker
n
el
Fu
n
ctio
n
W
i
t
h
o
u
t
P
l
a
st
i
c
S
u
r
g
e
r
y
W
i
t
h
P
l
a
st
i
c
S
u
r
g
e
r
y
P
C
A
S
I
F
T
V
S
I
F
T
EV
S
I
F
T
P
C
A
S
I
F
T
V
S
I
F
T
EV
S
I
F
T
A
c
c
u
r
a
c
y
0
.
5
5
(
4
)
0
.
7
1
(
2
)
0
.
8
8
(
1
)
0
.
8
2
(
4
)
0
.
8
6
(
3
)
0
.
8
7
(
2
)
0
.
5
5
(
4
)
0
.
7
1
(
2
)
S
e
n
si
t
i
v
i
t
y
0
.
1
3
(
3
)
0
.
1
2
(
4
)
0
.
0
7
(
4
)
0
.
1
2
(
2
)
0
.
1
3
(
3
)
0
.
2
2
(
1
)
0
.
1
3
(
3
)
0
.
1
2
(
4
)
S
p
e
c
i
f
i
c
i
t
y
0
.
9
2
(
2
)
0
.
9
1
(
4
)
0
.
9
3
(
4
)
0
.
9
4
(
2
)
0
.
9
4
(
3
)
0
.
9
5
(
1
)
0
.
9
2
(
2
)
0
.
9
1
(
4
)
P
r
e
c
i
si
o
n
0
.
6
(
2
)
0
.
3
(
4
)
0
.
0
6
(
4
)
0
.
2
7
(
2
)
0
.
2
(
3
)
0
.
4
(
1
)
0
.
6
(
2
)
0
.
3
(
4
)
F
P
R
0
.
0
8
(
2
)
0
.
0
9
(
4
)
0
.
0
6
(
4
)
0
.
0
6
(
2
)
0
.
0
6
(
3
)
0
.
0
4
(
1
)
0
.
0
8
(
2
)
0
.
0
9
(
4
)
F
N
R
0
.
8
7
(
3
)
0
.
8
8
(
4
)
0
.
9
3
(
4
)
0
.
8
8
(
2
)
0
.
8
7
(
3
)
0
.
7
8
(
1
)
0
.
8
7
(
3
)
0
.
8
8
(
4
)
N
P
V
0
.
9
2
(
2
)
0
.
9
1
(
4
)
0
.
9
3
(
4
)
0
.
9
4
(
2
)
0
.
9
4
(
3
)
0
.
9
5
(
1
)
0
.
9
2
(
2
)
0
.
9
1
(
4
)
F
D
R
0
.
4
(
2
)
0
.
7
(
4
)
0
.
9
3
(
4
)
0
.
7
3
(
2
)
0
.
8
(
3
)
0
.
6
(
1
)
0
.
4
(
2
)
0
.
7
(
4
)
F
1
_
S
c
o
r
e
0
.
2
1
(
3
)
0
.
1
7
(
4
)
0
.
0
6
(
4
)
0
.
1
7
(
2
)
0
.
1
6
(
3
)
0
.
2
9
(
1
)
0
.
2
1
(
3
)
0
.
1
7
(
4
)
M
C
C
0
.
0
9
(
3
)
0
.
0
4
(
4
)
0
(
4
)
0
.
0
9
(
2
)
0
.
0
9
(
3
)
0
.
2
3
(
1
)
0
.
0
9
(
3
)
0
.
0
4
(
4
)
A
v
g
_
r
a
n
k
2
.
6
3
.
8
3
.
7
2
.
4
2
.
8
1
.
1
2
.
6
3
.
8
F
i
n
a
l
_
r
a
n
k
3
4
4
2
3
1
3
4
4
.
4
.
I
m
pa
ct
o
f
SI
F
T
di
m
en
s
io
n
I
n
SVM
class
if
ier
,
t
w
o
m
ai
n
p
ar
am
eter
s
s
u
c
h
as
r
ad
iu
s
a
n
d
E
n
lar
g
e
Facto
r
(
E
F)
ar
e
p
r
esen
t.
T
h
ese
f
ac
to
r
s
ar
e
v
ar
ied
an
d
t
h
e
p
er
f
o
r
m
a
n
ce
an
al
y
s
is
i
s
d
o
n
e.
T
h
e
d
iag
r
a
m
m
a
tic
r
ep
r
esen
tati
o
n
o
n
th
e
i
m
p
ac
t
o
f
SIFT
f
ea
tu
r
e
i
s
s
h
o
w
n
i
n
F
ig
u
r
e
4
.
T
h
e
v
al
u
e
f
o
r
r
ad
iu
s
i
s
v
ar
ied
f
r
o
m
2
.
5
,
5
,
1
0
,
1
5
an
d
2
0
w
h
er
e
t
h
e
co
r
r
esp
o
n
d
in
g
v
al
u
e
o
f
E
F i
s
v
ar
ied
f
r
o
m
0
.
5
,
1
,
1
.
3
,
1
.
5
an
d
1
.
7
.
I
n
Fig
u
r
e
4
(
a)
,
th
e
ac
cu
r
ac
y
i
s
b
etter
f
o
r
th
e
r
ad
iu
s
2
.
5
an
d
E
F
0
.
5
.
T
h
e
s
p
ec
if
icit
y
i
s
b
etter
f
o
r
th
e
r
ad
iu
s
2
.
5
an
d
E
F
1
as
s
h
o
w
n
in
Fig
u
r
e
4
(
b
)
.
I
n
Fig
u
r
e
4
(
c)
,
s
p
ec
if
icit
y
r
a
n
g
e
i
s
b
etter
f
o
r
th
e
r
ad
iu
s
1
0
an
d
th
e
E
F
v
al
u
e
1
.
7
.
B
u
t
th
e
r
a
n
g
e
o
f
f
1
_
s
co
r
e
is
h
i
g
h
e
s
t
f
o
r
th
e
r
ad
iu
s
1
0
an
d
E
F
1
.
T
h
e
m
e
asu
r
es
s
u
c
h
a
s
FD
R
an
d
FN
R
ar
e
b
etter
f
o
r
t
h
e
E
F
1
.
3
.
B
u
t
h
er
e
t
h
e
r
ad
iu
s
is
d
if
f
er
en
t
f
o
r
t
w
o
m
ea
s
u
r
es
s
u
c
h
as
2
.
5
in
FD
R
an
d
5
in
FN
R
.
T
h
e
r
an
g
e
o
f
FP
R
is
h
ig
h
w
i
th
r
ad
i
u
s
2
.
5
an
d
E
F
1
.
3
.
T
h
e
m
ea
s
u
r
es
s
u
ch
a
s
MC
C
,
NP
V
an
d
p
r
ec
is
io
n
s
h
o
w
b
etter
p
er
f
o
r
m
an
ce
w
it
h
r
ad
iu
s
1
0
an
d
E
F
1
.
7
.
Fro
m
t
h
is
i
n
s
p
ec
tio
n
,
it
is
clea
r
th
e
th
er
e
n
o
f
i
x
ed
v
al
u
e
f
o
r
r
ad
iu
s
a
n
d
E
F a
n
d
th
e
p
er
f
o
r
m
a
n
ce
o
f
SV
M
class
i
f
ier
b
etter
w
h
ile
v
ar
y
i
n
g
th
ese
f
ac
to
r
s
.
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.
7
,
No
.
4
,
A
u
g
u
s
t
2017
:
1
9
2
3
–
1
9
3
3
1930
(
a
)
(
b
)
(
c
)
(
d
)
(
e
)
(
f
)
(g
)
(h
)
(
i
)
(
j)
Fig
u
r
e
4
.
Gr
ap
h
ical
r
ep
r
esen
ta
tio
n
o
n
th
e
i
m
p
ac
t o
f
SIFT
f
ea
tu
r
es (
a)
A
cc
u
r
ac
y
(
b
)
Sen
s
iti
v
it
y
(
c)
Sp
ec
if
ici
t
y
(
d
)
F1
Sco
r
e
(
e)
FDR
(
f
)
FNR
(
g
)
FP
R
(
h
)
MC
C
(
i)
NP
V(
j)
Pr
ec
is
io
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
EV
-
S
I
F
T
-
A
n
E
xten
d
ed
S
c
a
le
I
n
va
r
ia
n
t F
a
ce
R
ec
o
g
n
itio
n
fo
r
P
la
s
tic
S
u
r
g
ery
….
(
A
r
ch
a
n
a
H.
S
a
b
le
)
1931
4
.
5
.
Sens
it
iv
it
y
t
o
P
la
s
t
ic
Su
rg
er
y
F
a
ce
s
T
h
e
g
r
ap
h
ical
r
ep
r
esen
tat
io
n
o
f
th
e
r
an
k
i
n
g
o
f
d
if
f
er
e
n
t
k
er
n
el
o
f
S
VM
clas
s
i
f
ier
l
ik
e
li
n
ea
r
,
q
u
ad
r
atic,
R
B
F
an
d
M
L
P
f
o
r
w
it
h
a
n
d
w
it
h
o
u
t
p
las
tic
s
u
r
g
er
y
is
s
h
o
w
n
i
n
Fi
g
u
r
e
5
.
Ge
n
er
all
y
,
p
r
e
-
s
u
r
g
er
y
an
d
p
o
s
t
-
s
u
r
g
er
y
f
ac
es
m
a
y
h
av
e
h
i
g
h
v
ar
iatio
n
in
t
h
eir
f
ea
tu
r
es.
I
n
Fig
u
r
e
5
,
th
e
r
an
k
o
f
p
r
e
-
s
u
r
g
er
y
f
ac
e
is
h
ig
h
w
h
er
e
as
th
e
r
an
k
o
f
p
o
s
t
s
u
r
g
er
y
f
ac
e
i
s
lo
w
.
T
h
er
ef
o
r
e,
it
i
s
clea
r
th
a
t
t
h
e
p
r
o
p
o
s
ed
E
V
-
SIFT
f
ea
tu
r
e
i
s
h
ig
h
l
y
s
e
n
s
itiv
e
to
p
last
ic
s
u
r
g
er
y
f
ac
es.
(
a)
(
b
)
Fig
u
r
e
5
.
R
an
k
i
n
g
o
f
d
i
f
f
er
en
t
SVM
k
er
n
el
u
s
in
g
P
C
A
,
SIFT
,
V
-
SIFT
an
d
E
V
-
SIFT
(
a)
W
i
th
o
u
t p
las
tic
s
u
r
g
er
y
(
b
)
W
ith
p
last
ic
s
u
r
g
er
y
5.
CO
NCLU
SI
O
N
T
h
is
p
ap
er
h
as
p
r
ese
n
ted
a
f
a
ce
r
ec
o
g
n
itio
n
tec
h
n
iq
u
e
t
h
at
u
s
e
s
d
er
iv
ed
f
ea
tu
r
e
s
b
ased
o
n
E
V
-
SIFT
ap
p
r
o
ac
h
.
T
h
e
co
r
r
esp
o
n
d
in
g
s
y
s
te
m
w
as
e
v
al
u
ated
u
s
i
n
g
p
last
ic
s
u
r
g
er
y
i
m
ag
e
d
atab
ase
o
f
5
15
s
u
b
j
ec
ts
w
h
er
e
it
co
n
tai
n
s
ea
c
h
i
m
a
g
e
o
f
p
r
e
-
s
u
r
g
er
y
a
n
d
p
o
s
t
s
u
r
g
er
y
f
ac
es.
T
h
e
p
r
o
p
o
s
ed
E
V
-
SIFT
ap
p
r
o
ac
h
h
as
o
b
tain
ed
th
e
v
o
lu
m
e
o
f
t
h
e
s
tr
u
ctu
r
e
a
n
d
th
e
co
n
tr
ast
o
f
t
h
e
i
m
a
g
e
an
d
also
it
h
as
r
e
m
o
v
e
d
all
th
e
u
n
w
a
n
ted
k
e
y
p
o
in
ts
e
f
f
ec
tiv
el
y
.
T
h
e
ex
tr
ac
ted
f
ea
tu
r
es
w
er
e
ap
p
lied
to
th
e
SVM
clas
s
i
f
ier
f
o
r
th
e
r
ec
o
g
n
itio
n
p
u
r
p
o
s
e.
T
h
e
p
er
f
o
r
m
an
ce
m
ea
s
u
r
es
wer
e
an
al
y
ze
d
in
d
if
f
er
e
n
t
k
er
n
el
o
f
SVM
w
ith
d
i
f
f
er
e
n
t
ex
is
tin
g
f
ea
t
u
r
es.
Her
e
EV
-
SIFT
f
ea
t
u
r
e
w
a
s
ef
f
ec
ti
v
e
f
o
r
p
r
o
d
u
cin
g
th
e
b
es
t
p
er
f
o
r
m
an
ce
.
T
h
e
p
ar
am
e
ter
s
o
f
S
VM
class
i
f
ier
s
u
c
h
as
r
ad
iu
s
an
d
en
lar
g
e
f
ac
t
o
r
w
a
s
v
ar
ied
.
Fro
m
t
h
e
an
al
y
s
i
s
it
w
as
clea
r
th
at,
th
e
p
er
f
o
r
m
an
ce
w
a
s
b
etter
f
o
r
v
ar
ied
v
al
u
es
o
f
r
ad
iu
s
an
d
E
F
an
d
it
w
as
n
o
t
f
i
x
ed
.
So
h
e
r
e
a
p
r
o
p
er
tu
n
in
g
w
as
n
ee
d
ed
f
o
r
o
b
tain
in
g
t
h
e
f
i
x
ed
v
al
u
e.
I
n
f
u
tu
r
e
w
o
r
k
,
th
e
a
n
al
y
s
is
b
a
s
ed
o
n
th
e
tu
n
in
g
p
r
o
ce
s
s
w
i
ll
b
e
d
o
n
e
to
h
av
e
th
e
ac
c
u
r
at
e
r
ec
o
g
n
itio
n
o
f
p
last
ic
s
u
r
g
er
y
f
ac
e.
ACK
NO
WL
E
D
G
E
M
E
NT
S
First
o
f
all,
I
t
h
an
k
L
o
r
d
Sh
r
i
Gaj
an
an
Ma
h
ar
aj
f
o
r
g
i
v
in
g
m
e
s
tr
en
g
th
an
d
ab
ilit
y
to
co
m
p
lete
t
h
i
s
s
tu
d
y
.
F
u
t
h
er
m
o
r
e
,
I
w
o
u
ld
e
x
p
r
ess
m
y
g
r
ati
tu
d
e
to
m
y
r
es
ea
r
ch
g
u
id
e
P
r
o
f
.
S.N.
T
al
b
ar
f
o
r
th
eir
p
o
w
er
f
u
l
g
u
id
a
n
ce
,
m
o
t
iv
atio
n
,
v
a
lu
ab
l
e
s
u
g
g
esti
o
n
s
,
s
u
p
p
o
r
t
an
d
at
ten
tio
n
th
r
o
u
g
h
o
u
t
th
i
s
r
esear
ch
.
I
th
a
n
k
all
m
y
f
r
ien
d
s
f
o
r
s
h
ar
in
g
t
h
eir
ex
p
er
ien
ce
s
an
d
k
n
o
w
led
g
e.
Sp
e
cia
l
th
an
k
to
M
y
Hu
s
b
an
d
f
o
r
p
r
o
v
id
in
g
m
o
r
al
s
u
p
p
o
r
t a
n
d
al
w
a
y
s
e
n
co
u
r
a
g
i
n
g
m
e
to
d
o
g
o
o
d
r
esear
ch
.
RE
F
E
R
E
NC
E
S
[
1
]
A
n
il
Ku
m
a
r
S
a
o
,
B.
Ye
g
n
a
n
a
r
a
y
a
n
a
,
“
F
a
c
e
V
e
ri
f
ica
ti
o
n
u
sin
g
T
e
m
p
late
M
a
tch
in
g
”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
In
fo
rm
a
t
io
n
F
o
re
n
sic
s a
n
d
S
e
c
u
rity
,
v
o
l.
2
,
n
o
.
3
,
p
p
.
6
3
6
-
6
4
1
,
S
e
p
.
2
0
0
7
.
[
2
]
W
il
li
a
m
Ro
b
so
n
S
c
h
w
a
rt
z
,
Hu
im
in
G
u
o
,
Jo
n
g
h
y
u
n
C
h
o
i,
L
a
rry
S
.
Da
v
is,
“
F
a
c
e
Id
e
n
ti
f
ica
ti
o
n
u
sin
g
L
a
r
g
e
F
e
a
tu
re
S
e
ts”
,
IEE
E
T
r
a
n
sa
c
ti
o
n
s
o
n
Ima
g
e
Pro
c
e
ss
in
g
,
v
o
l.
2
1
,
n
o
.
4
,
p
p
.
2
2
4
5
-
2
2
5
5
,
A
p
ril
2
0
1
2
.
[
3
]
M
.
A
.
T
u
rk
,
A
.
P
.
P
e
n
tl
a
n
d
,
“
Fa
c
e
Rec
o
g
n
it
i
o
n
u
sin
g
E
i
g
e
n
f
a
c
e
s”
,
In
P
ro
c
e
e
d
in
g
s
o
f
th
e
IEE
E
C
o
m
p
u
ter
S
o
c
iety
Co
n
f
e
re
n
c
e
o
n
Co
m
p
u
ter
V
isi
o
n
a
n
d
P
a
tt
e
rn
Re
c
o
g
n
it
i
o
n
,
M
a
u
i
,
H
I,
1
9
9
1
,
p
p
.
5
8
6
-
5
9
1
.
[
4
]
J.
Ru
iz
-
d
e
l
-
S
o
lar,
P
.
Na
v
a
rre
t
e
,
“
Ei
g
e
n
sp
a
c
e
-
b
a
se
d
fa
c
e
R
e
c
o
g
n
it
io
n
:
A
C
o
m
p
a
r
a
ti
v
e
S
tu
d
y
o
f
d
iffere
n
t
A
p
p
ro
a
c
h
e
s”
,
IEE
E
T
r
a
n
sa
c
t
io
n
s
o
n
S
y
ste
ms
,
M
a
n
,
a
n
d
Cy
b
e
rn
e
t
ics
,
Pa
rt
C
(
Ap
p
li
c
a
ti
o
n
s
a
n
d
Re
v
iews
)
,
v
o
l.
3
5
,
n
o
.
3
,
p
p
.
3
1
5
-
3
2
5
,
A
u
g
.
2
0
0
5
.
[
5
]
S
.
L
a
w
re
n
c
e
,
C.
L
.
G
il
e
s,
A
h
Ch
u
n
g
T
so
i,
A
.
D.
Ba
c
k
,
“
F
a
c
e
Re
c
o
g
n
it
io
n
:
A
C
o
n
v
o
l
u
ti
o
n
a
l
Ne
u
ra
l
-
Ne
t
w
o
rk
A
p
p
ro
a
c
h
”
,
IEE
E
T
ra
n
s
a
c
ti
o
n
s o
n
Ne
u
ra
l
Ne
two
rk
s
,
v
o
l.
8
,
n
o
.
1
,
p
p
.
9
8
-
1
1
3
,
Ja
n
.
1
9
9
7
.
[
6
]
X
iao
f
e
i
He
,
S
h
u
ich
e
n
g
Ya
n
,
Y
u
x
iao
Hu
,
P
.
Niy
o
g
i,
Ho
n
g
-
Jia
n
g
Zh
a
n
g
,
“
F
a
c
e
R
e
c
o
g
n
it
io
n
u
si
n
g
L
a
p
lac
ian
f
a
c
e
s”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Pa
t
ter
n
An
a
lys
is
a
n
d
M
a
c
h
in
e
In
telli
g
e
n
c
e
,
v
o
l.
2
7
,
n
o
.
3
,
p
p
.
3
2
8
-
3
4
0
,
M
a
rc
h
2
0
0
5
.
[
7
]
T
o
lg
a
In
a
n
,
Ug
u
r
Ha
li
c
i,
“
3
-
D
F
a
c
e
Re
c
o
g
n
it
io
n
W
it
h
L
o
c
a
l
S
h
a
p
e
De
sc
rip
to
rs”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
In
fo
rm
a
ti
o
n
Fo
re
n
sic
s a
n
d
S
e
c
u
rity
,
v
o
l.
7
,
n
o
.
2
,
p
p
.
5
7
7
-
5
8
7
,
A
p
ril
2
0
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.
7
,
No
.
4
,
A
u
g
u
s
t
2017
:
1
9
2
3
–
1
9
3
3
1932
[
8
]
Ze
L
u
,
Xu
d
o
n
g
Jia
n
g
,
A
lex
C.
Ko
t,
“
A
C
o
lo
r
C
h
a
n
n
e
l
F
u
si
o
n
A
p
p
ro
a
c
h
f
o
r
F
a
c
e
Re
c
o
g
n
it
io
n
”
,
IEE
E
S
ig
n
a
l
Pro
c
e
ss
in
g
L
e
tt
e
rs
,
v
o
l.
2
2
,
n
o
.
1
1
,
p
p
.
1
8
3
9
-
1
8
4
3
,
No
v
.
2
0
1
5
.
[
9
]
Yo
n
g
X
u
,
X
iao
z
h
a
o
F
a
n
g
,
X
u
e
l
o
n
g
L
i,
Jia
n
g
Y
a
n
g
,
Ja
n
e
Y
o
u
,
Ho
n
g
L
iu
,
S
h
a
o
h
u
a
T
e
n
g
,
“
D
a
ta
Un
c
e
rtain
ty
in
F
a
c
e
Re
c
o
g
n
it
io
n
”
,
IEE
E
T
r
a
n
s
a
c
ti
o
n
s
o
n
Cy
b
e
rn
e
ti
c
s
,
v
o
l.
4
4
,
n
o
.
1
0
,
p
p
.
1
9
5
0
-
1
9
6
1
,
Oc
t.
2
0
1
4
.
[
1
0
]
A
n
il
K.
Ja
in
,
Bre
n
d
a
n
Kla
re
,
Un
sa
n
g
P
a
rk
,
“
Fa
c
e
Rec
o
g
n
it
io
n
:
S
o
me
Ch
a
ll
e
n
g
e
s
in
F
o
re
n
si
c
s”
,
2
0
1
1
IEE
E
In
tern
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
A
u
t
o
m
a
ti
c
F
a
c
e
&
Ge
stu
re
Re
c
o
g
n
i
ti
o
n
a
n
d
W
o
rk
sh
o
p
s
(F
G
2
0
1
1
),
S
a
n
ta
Ba
rb
a
ra
,
2
0
1
1
,
p
p
.
7
2
6
-
7
3
3
.
[
1
1
]
P
o
o
n
a
m
S
h
a
r
m
a
,
Ra
m
N.
Ya
d
a
v
,
Ka
r
m
v
e
e
r
V
.
A
r
y
a
,
“
P
o
se
-
In
v
a
rian
t
f
a
c
e
Re
c
o
g
n
it
io
n
u
si
n
g
Cu
rv
e
let
Ne
u
ra
l
N
e
tw
o
rk
”
,
IET
Bi
o
me
trics
,
v
o
l.
3
,
n
o
.
3
,
p
p
.
1
2
8
-
1
3
8
,
S
e
p
.
2
0
1
4
.
[
1
2
]
S
a
n
g
-
He
o
n
L
e
e
,
Do
n
g
-
Ju
Kim
,
J
in
-
Ho
C
h
o
,
“
Ill
u
m
in
a
ti
o
n
-
R
o
b
u
st
f
a
c
e
Re
c
o
g
n
it
io
n
S
y
ste
m
b
a
se
d
o
n
Dif
f
e
re
n
ti
a
l
C
o
m
p
o
n
e
n
ts”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
C
o
n
s
u
me
r E
lec
tro
n
ics
,
v
o
l
.
5
8
,
n
o
.
3
,
p
p
.
9
6
3
-
9
7
0
,
A
u
g
2
0
1
2
.
[
1
3
]
W
il
m
a
n
W
.
W
.
Zo
u
,
P
o
n
g
C.
Y
u
e
n
,
“
V
e
ry
L
o
w
Re
so
lu
ti
o
n
F
a
c
e
Re
c
o
g
n
it
io
n
P
r
o
b
lem
”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
I
ma
g
e
Pro
c
e
ss
in
g
,
v
o
l.
2
1
,
n
o
.
1
,
p
p
.
3
2
7
-
3
4
0
,
Ja
n
2
0
1
2
.
[
1
4
]
Un
sa
n
g
P
a
rk
,
Yi
y
in
g
T
o
n
g
,
A
n
il
K.
Ja
in
,
“
A
g
e
-
In
v
a
rian
t
F
a
c
e
Re
c
o
g
n
it
io
n
”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s o
n
Pa
tt
e
rn
An
a
lys
is
a
n
d
M
a
c
h
i
n
e
In
telli
g
e
n
c
e
,
v
o
l.
3
2
,
n
o
.
5
,
p
p
.
9
4
7
-
9
5
4
,
M
a
y
2
0
1
0
.
[
1
5
]
15.
S
iv
a
ra
m
P
ra
sa
d
M
u
d
u
n
u
r
i,
S
o
m
a
Biswa
s,
“
L
o
w
Re
so
lu
ti
o
n
F
a
c
e
Re
c
o
g
n
it
io
n
A
c
ro
ss
V
a
riatio
n
s
i
n
P
o
se
a
n
d
Ill
u
m
in
a
ti
o
n
”
,
IE
EE
T
ra
n
sa
c
ti
o
n
s
o
n
P
a
tt
e
rn
A
n
a
lys
is
a
n
d
M
a
c
h
in
e
In
telli
g
e
n
c
e
,
v
o
l.
3
8
,
n
o
.
5
,
p
p
.
1
0
3
4
-
1
0
4
0
,
M
a
y
2
0
1
6
.
[
1
6
]
Rich
a
S
in
g
h
,
M
a
y
a
n
k
V
a
tsa
,
Af
z
e
l
No
o
re
,
“
Ef
fec
t
o
f
P
la
stic
S
u
rg
e
ry
o
n
fa
c
e
Rec
o
g
n
it
io
n
:
A
Pre
li
min
a
ry
S
t
u
d
y
”
,
2
0
0
9
IEE
E
Co
m
p
u
ter
S
o
c
iety
C
o
n
f
e
re
n
c
e
o
n
Co
m
p
u
ter
V
isio
n
a
n
d
P
a
tt
e
rn
Re
c
o
g
n
it
i
o
n
W
o
rk
sh
o
p
s,
M
iam
i,
F
L
,
2
0
0
9
,
p
p
.
7
2
-
7
7
.
[
1
7
]
Rich
a
S
in
g
h
,
M
a
y
a
n
k
V
a
tsa
,
Him
a
n
sh
u
S
.
Bh
a
tt
,
S
a
m
a
rth
Bh
a
ra
d
w
a
j,
Af
z
e
l
No
o
re
,
S
h
a
h
i
n
S
.
No
o
re
y
e
z
d
a
n
,
“
P
las
ti
c
S
u
rg
e
r
y
:
A
Ne
w
Di
m
e
n
sio
n
to
F
a
c
e
Re
c
o
g
n
it
io
n
”
,
IE
EE
T
ra
n
s
a
c
ti
o
n
s
o
n
In
f
o
rm
a
ti
o
n
Fo
re
n
sic
s
a
n
d
S
e
c
u
rity
,
v
o
l.
5
,
n
o
.
3
,
p
p
.
4
4
1
-
4
4
8
,
S
e
p
.
2
0
1
0
.
[
1
8
]
X
in
L
iu
,
S
h
ig
u
a
n
g
S
h
a
n
,
Xili
n
C
h
e
n
,
“
F
a
c
e
Re
c
o
g
n
it
i
o
n
a
f
ter
P
las
ti
c
S
u
rg
e
r
y
:
A
Co
m
p
re
h
e
n
siv
e
S
t
u
d
y
”
,
Co
mp
u
ter
Vi
sio
n
–
ACC
V
2
0
1
2
,
L
e
c
tu
re
No
t
e
s in
Co
mp
u
ter
S
c
ien
c
e
,
S
p
ri
n
g
e
r
,
v
o
l.
7
7
2
5
,
p
p
.
5
6
5
-
5
7
6
.
[
1
9
]
M
a
ria
De
M
a
rsi
c
o
,
M
ich
e
le
Na
p
p
i,
Da
n
iel
Ricc
io
,
Ha
rr
y
W
e
c
h
sle
r,
“
Ro
b
u
st
F
a
c
e
Re
c
o
g
n
it
io
n
a
f
te
r
P
las
ti
c
S
u
rg
e
ry
Us
in
g
L
o
c
a
l
Re
g
io
n
A
n
a
l
y
sis”
,
I
ma
g
e
An
a
lys
is
a
n
d
Rec
o
g
n
it
io
n
,
L
e
c
tu
re
No
tes
in
Co
m
p
u
ter
S
c
ien
c
e
,
S
p
rin
g
e
r
,
v
o
l.
6
7
5
4
,
p
p
.
1
9
1
-
2
0
0
.
[
2
0
]
N.
S
.
L
a
k
sh
m
ip
ra
b
h
a
,
S
.
M
a
j
u
m
d
e
r,
“
Fa
c
e
Rec
o
g
n
it
i
o
n
S
y
ste
m
In
v
a
ria
n
t
to
Pl
a
stic
S
u
rg
e
ry
”
,
2
0
1
2
1
2
t
h
In
tern
a
ti
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
In
tell
ig
e
n
t
S
y
ste
m
s De
si
g
n
a
n
d
A
p
p
li
c
a
ti
o
n
s (IS
DA
),
Ko
c
h
i
,
2
0
1
2
,
p
p
.
2
5
8
-
2
6
3
.
[
2
1
]
Ab
h
a
R.
G
u
lh
a
n
e
,
S
.
A
.
L
a
d
h
a
k
e
,
S
.
B.
Ka
stu
riw
a
la,
“
A
Rev
iew
o
n
S
u
rg
ica
l
ly
Al
ter
e
d
f
a
c
e
Ima
g
e
s
Rec
o
g
n
it
io
n
u
sin
g
M
u
lt
imo
d
a
l
B
io
-
me
tri
c
F
e
a
tu
re
s”
,
in
P
r
o
c
e
e
d
in
g
s
o
f
th
e
2
0
1
5
2
n
d
In
tern
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
El
e
c
tro
n
ics
a
n
d
Co
m
m
u
n
ica
ti
o
n
S
y
ste
m
s (IC
ECS
),
Co
im
b
a
to
re
,
2
0
1
5
,
p
p
.
1
1
6
8
-
1
1
7
1
.
[
2
2
]
M
a
ria
De
M
a
rsic
o
,
M
ich
e
le
Na
p
p
i,
Da
n
iel
Ricc
io
,
Ha
rry
Wec
h
sle
r,
“
Ro
b
u
st
f
a
c
e
R
e
c
o
g
n
it
io
n
a
f
ter
P
las
ti
c
S
u
rg
e
r
y
u
sin
g
Re
g
io
n
-
b
a
se
d
A
p
p
ro
a
c
h
e
s”
,
Pa
tt
e
rn
Rec
o
g
n
it
i
o
n
,
v
o
l.
4
8
,
n
o
.
4
,
p
p
.
1
2
6
1
-
1
2
7
6
,
A
p
ril
2
0
1
5
.
[
2
3
]
Na
m
a
n
Ko
h
li
,
Da
k
sh
a
Y
a
d
a
v
,
Afz
e
l
No
o
re
,
“
M
u
lt
ip
le
P
ro
jec
ti
v
e
Dic
ti
o
n
a
ry
L
e
a
rn
in
g
to
De
tec
t
P
las
ti
c
S
u
rg
e
r
y
f
o
r
F
a
c
e
V
e
rif
ica
ti
o
n
”
,
IEE
E
Acc
e
ss
,
v
o
l.
3
,
p
p
.
2
5
7
2
-
2
5
8
0
,
2
0
1
5
.
[
2
4
]
Ch
o
ll
e
tt
e
C
.
Ch
u
d
e
-
Olisa
h
,
G
h
a
z
a
li
B.
S
u
lo
n
g
,
Uc
h
e
A
.
K.
Ch
u
d
e
-
Ok
o
n
k
w
o
,
S
it
i
Z.
M
.
Ha
sh
i
m
,
“
Ed
g
e
-
b
a
se
d
Rep
re
se
n
ta
ti
o
n
a
n
d
Rec
o
g
n
it
i
o
n
fo
r
S
u
rg
ic
a
ll
y
Al
ter
e
d
fa
c
e
I
ma
g
e
s”
,
in
P
r
o
c
e
e
d
in
g
s
o
f
th
e
2
0
1
3
7
th
In
ter
n
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
S
ig
n
a
l
P
ro
c
e
ss
in
g
a
n
d
C
o
m
m
u
n
ica
ti
o
n
S
y
ste
m
s (IC
S
P
CS
),
Ca
rra
ra
,
V
I
C,
2
0
1
3
,
p
p
.
1
-
7.
[
2
5
]
Ha
m
id
Ou
a
n
a
n
,
M
o
h
a
m
m
e
d
Ou
a
n
a
n
,
“
G
a
b
o
rHO
G
F
e
a
tu
re
s
b
a
se
d
f
a
c
e
Re
c
o
g
n
it
io
n
S
c
h
e
m
e
,
”
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
ol
.
15
, n
o
.
2
,
2
0
1
5
,
p
p
.
3
3
1
-
3
3
5
.
[
2
6
]
M
.
I.
Ou
l
o
u
l
,
Z.
M
o
u
tak
k
i
,
K.
A
fd
e
l,
A
.
Am
g
h
a
r,
“
A
n
Eff
icie
n
t
F
a
c
e
Re
c
o
g
n
it
io
n
u
sin
g
S
IF
T
De
sc
r
ip
to
r
in
RG
BD
Im
a
g
e
s,
”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
El
e
c
trica
l
a
n
d
C
o
mp
u
ter
En
g
i
n
e
e
rin
g
(
IJ
ECE
)
,
v
ol
.
5
,
n
o
.
6
,
2
0
1
5
.
[
2
7
]
Him
a
n
sh
u
S
.
B
h
a
tt
,
S
a
m
a
rth
Bh
a
ra
d
w
a
j,
Rich
a
S
in
g
h
,
M
a
y
a
n
k
V
a
tsa
,
“
Re
c
o
g
n
izin
g
S
u
rg
ica
ll
y
A
lt
e
re
d
F
a
c
e
I
m
a
g
e
s
Us
in
g
M
u
lt
io
b
jec
ti
v
e
Ev
o
lu
ti
o
n
a
ry
A
lg
o
rit
h
m
”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
In
f
o
rm
a
ti
o
n
F
o
re
n
sic
s
a
n
d
S
e
c
u
rity
,
v
o
l.
8
,
n
o
.
1
,
p
p
.
8
9
-
1
0
0
,
Ja
n
.
2
0
1
3
.
[
2
8
]
Dh
y
a
a
S
h
a
h
e
e
d
S
a
b
r
A
l
-
A
z
z
a
wy
,
“
Ei
g
e
n
f
a
c
e
a
n
d
S
IF
T
F
o
r
G
e
n
d
e
r
Clas
sif
ic
a
ti
o
n
”
,
W
a
siit
J
o
u
rn
a
ll
fo
r
S
c
ii
e
n
c
e
&
M
e
d
icin
e
,
v
o
l.
5
,
n
o
.
1
,
p
p
.
6
0
-
7
6
,
2
0
1
2
.
[
2
9
]
Co
n
g
G
e
n
g
,
X
u
d
o
n
g
Jia
n
g
,
“
Fa
c
e
Rec
o
g
n
it
i
o
n
u
si
n
g
S
IFT
Fea
tu
re
s”
,
1
6
t
h
IEE
E
I
n
tern
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
Im
a
g
e
P
r
o
c
e
ss
in
g
(ICI
P
),
p
a
g
e
s 3
3
1
3
-
3
3
1
6
,
2
0
0
9
.
BI
O
G
RAP
H
I
E
S
O
F
AUTH
O
RS
Ar
c
h
a
n
a
H
.
S
a
b
le,
sh
e
re
c
e
iv
e
d
M
E
d
e
g
re
e
in
c
o
m
p
u
ter
S
c
ien
c
e
&
e
n
g
in
e
e
rin
g
f
ro
m
M
.
G
.
M
’s
Co
ll
e
g
e
o
f
En
g
g
.
,
S
RT
M
Un
iv
e
rsit
y
,
Na
n
d
e
d
.
He
r
re
se
a
rc
h
in
tere
sts
a
re
I
m
a
g
e
p
ro
c
e
ss
in
g
,
P
a
tt
e
rn
re
c
o
g
n
izin
g
a
n
d
Co
m
p
u
ter
v
isio
n
.
S
h
e
is
d
o
i
n
g
P
h
.
D
(T
h
e
sis)
in
C
o
m
p
u
ter
En
g
g
f
ro
m
S
R
T
M
Un
iv
e
rsit
y
,
Na
n
d
e
d
.
He
r
e
m
a
il
a
d
d
re
ss
is
h
e
ll
o
a
rc
h
u
2
7
@g
m
a
il
.
c
o
m
.
S
a
n
ja
y
N.
T
a
lb
a
r
re
c
e
i
v
e
d
h
is
B
.
E
a
n
d
M
.
E
d
e
g
re
e
s
f
ro
m
S
GG
S
In
stit
u
te
o
f
T
e
c
h
n
o
lo
g
y
,
N
a
n
d
e
d
,
In
d
ia
in
1
9
8
5
a
n
d
1
9
9
0
re
sp
e
c
ti
v
e
ly
.
He
o
b
tain
e
d
h
is
P
h
D
f
ro
m
S
RTM
Un
iv
e
rsity
,
Na
n
d
e
d
,
In
d
ia
in
2
0
0
0
.
He
re
c
e
iv
e
d
th
e
“
Yo
u
n
g
S
c
ien
ti
st
A
w
a
rd
”
b
y
URSI,
Ital
y
in
2
0
0
3
.
H
e
h
a
d
Co
ll
a
b
o
ra
ti
v
e
re
se
a
rc
h
p
ro
g
ra
m
m
e
a
t
Ca
rd
iff
Un
iv
e
rsit
y
W
a
les
,
U
K.
P
re
se
n
tl
y
h
e
is
w
o
rk
in
g
a
s
P
r
o
f
e
ss
o
r
a
n
d
He
a
d
,
De
p
a
rtm
e
n
t
o
f
El
e
c
tro
n
ics
&
T
e
lec
o
m
m
u
n
ica
ti
o
n
En
g
g
.
,
S
G
G
S
In
stit
u
te
o
f
En
g
in
e
e
rin
g
&
T
e
c
h
n
o
lo
g
y
Na
n
d
e
d
,
In
d
ia.
He
h
a
s
p
u
b
li
sh
e
d
5
0
j
o
u
r
n
a
l
p
a
p
e
rs
,
1
0
b
o
o
k
s
a
n
d
m
o
re
th
a
n
1
2
5
p
a
p
e
rs
in
re
f
e
rre
d
Na
ti
o
n
a
l
a
s
w
e
ll
a
s
In
tern
a
ti
o
n
a
l
C
o
n
f
e
re
n
c
e
s.
His
re
se
a
rc
h
in
tere
sts
in
c
lu
d
e
s
Im
a
g
e
p
ro
c
e
ss
in
g
,
M
e
d
ica
l
Im
a
g
e
p
ro
c
e
ss
in
g
,
M
u
lt
im
e
d
ia
Co
m
p
u
ti
n
g
a
n
d
Em
b
e
d
d
e
d
S
y
ste
m
D
e
sig
n
.
He
is
a
m
e
m
b
e
r
o
f
IEE
E,
IET
,
IET
E,
A
M
P
I
,
IS
T
E,
a
n
d
w
o
rk
e
d
o
n
m
a
n
y
p
re
stig
io
u
s co
m
m
it
te
e
s in
a
c
a
d
e
m
ic f
i
e
ld
o
f
In
d
ia.
Evaluation Warning : The document was created with Spire.PDF for Python.