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
,
Dec
em
b
er
201
6
,
p
p
.
2
7
8
1
~
2
7
8
8
I
SS
N:
2
0
8
8
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v
6i
6
.
1
2
0
0
6
2781
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
R
ecog
nition
using
Multi R
eg
io
n P
ro
m
inen
t
LB
P
R
epresen
tatio
n
Srini
v
a
s
a
Reddy
K
1
,
Vij
a
y
a
K
u
m
a
r
V
2
,
Ve
n
k
a
t
a
K
rish
na
3
1
De
p
a
rte
m
e
n
t
o
f
Co
m
p
u
ter S
c
ien
c
e
a
n
d
En
g
in
e
e
ri
n
g
,
In
stit
u
te o
f
A
e
ro
n
a
u
ti
c
a
l
E
n
g
in
e
e
rin
g
,
Hy
d
e
ra
b
a
d
,
In
d
ia
2
De
p
a
rte
m
e
n
t
o
f
Co
m
p
u
ter S
c
ien
c
e
a
n
d
En
g
in
e
e
ri
n
g
,
A
n
u
ra
g
G
ro
u
p
o
f
In
stit
u
ti
o
n
s,
Hy
d
e
ra
b
a
d
,
In
d
i
a
3
De
p
a
rte
m
e
n
t
o
f
Co
m
p
u
ter S
c
ien
c
e
a
n
d
En
g
in
e
e
ri
n
g
,
Vid
y
a
J
y
o
th
i
In
stit
u
te
o
f
T
e
c
h
n
o
lo
g
y
,
H
y
d
e
ra
b
a
d
,
In
d
ia
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Mar
29
,
2
0
1
6
R
ev
i
s
ed
A
u
g
2
5
,
2
0
1
6
A
cc
ep
ted
Sep
08
,
2
0
1
6
V
a
rio
u
s
f
a
c
e
re
c
o
g
n
it
io
n
m
e
th
o
d
s
a
re
d
e
riv
e
d
u
sin
g
lo
c
a
l
fe
a
tu
re
s
a
m
o
n
g
th
e
m
th
e
L
o
c
a
l
Bin
a
r
y
P
a
tt
e
rn
(L
BP
)
a
p
p
r
o
a
c
h
is
v
e
r
y
f
a
m
o
u
s.
T
h
e
h
isto
g
ra
m
tec
h
n
iq
u
e
s
b
a
se
d
o
n
L
BP
is
a
c
o
m
p
lex
tas
k
.
L
a
ter
Un
ifo
rm
L
o
c
a
l
Bin
a
ry
P
a
tt
e
rn
(UL
BP
)
is
d
e
riv
e
d
o
n
L
BP
,
b
a
se
d
o
n
th
e
b
i
tw
ise
tran
siti
o
n
s
a
n
d
UL
BP
‟s
a
re
trea
ted
a
s
th
e
fu
n
d
a
m
e
n
tal
p
ro
p
e
rty
o
f
tex
tu
re
.
T
h
e
U
L
BP
a
p
p
ro
a
c
h
trea
ted
a
ll
N
o
n
-
U
n
if
o
rm
L
o
c
a
l
Bin
a
ry
P
a
tt
e
rn
s‟
(
NU
L
BP
)
in
to
o
n
e
m
is
c
e
ll
a
n
e
o
u
s
lab
e
l.
Re
c
e
n
tl
y
w
e
h
a
v
e
d
e
riv
e
d
P
ro
m
in
e
n
t
LBP
(
P
L
B
P
),
M
a
x
i
m
u
m
P
L
BP
(M
P
L
BP
)
a
n
d
S
m
a
ll
e
st P
L
BP
(S
P
L
BP
).
T
h
e
P
L
BP
c
o
n
sists
o
f
th
e
m
a
jo
rit
y
o
f
th
e
UL
BP
‟s
a
n
d
so
m
e
o
f
th
e
NU
L
BP
‟s.
T
h
e
b
a
sic
d
isa
d
v
a
n
tag
e
o
f
th
e
se
v
a
rio
u
s
v
a
rian
ts
o
f
L
BP
‟s
is
th
e
y
a
r
e
b
a
si
c
a
ll
y
lo
c
a
l
a
p
p
ro
a
c
h
e
s
a
n
d
c
o
m
p
lete
ly
f
a
il
e
d
in
re
p
re
se
n
ti
n
g
f
e
a
tu
re
s
d
e
riv
e
d
f
ro
m
larg
e
re
g
io
n
s
o
r
m
a
c
ro
stru
c
tu
re
s,
w
h
i
c
h
a
re
v
e
r
y
m
u
c
h
e
s
se
n
ti
a
l
f
o
r
f
a
c
e
s.
T
h
is
p
a
p
e
r
d
e
riv
e
s
P
L
B
P
‟s
o
n
t
h
e
la
rg
e
re
g
io
n
.
T
h
e
re
c
tan
g
u
lar
re
g
i
o
n
o
f
t
h
is
p
a
p
e
r
is
a
ss
u
m
e
d
w
it
h
a
siz
e
o
f
m
u
lt
ip
les
o
f
th
re
e
a
n
d
P
L
BP
s
a
r
e
e
v
a
lu
a
ted
o
n
d
iv
i
d
in
g
e
a
c
h
re
g
io
n
in
t
o
m
u
lt
ip
le
re
g
io
n
s.
T
h
e
p
ro
p
o
se
d
M
u
lt
i
Re
g
io
n
-
P
L
B
P
(M
R
-
P
L
B
P
)
a
p
p
r
o
a
c
h
is
tes
ted
o
n
th
re
e
f
a
c
ial
d
a
tab
a
se
s
n
a
m
e
l
y
Y
a
le,
In
d
ian
a
n
d
A
T&
T
ORL
.
T
h
e
e
x
p
e
ri
m
e
n
tal
re
su
lt
s
sh
o
w
th
e
p
ro
p
o
se
d
a
p
p
ro
a
c
h
sig
n
if
ica
n
tl
y
o
u
tp
e
rf
o
rm
s
th
e
o
th
e
r
L
B
P
b
a
se
d
f
a
c
e
r
e
c
o
g
n
it
io
n
m
e
th
o
d
s.
K
ey
w
o
r
d
:
L
o
ca
l a
n
d
m
ac
r
o
s
tr
u
ct
u
r
es
L
o
ca
l b
in
ar
y
p
atter
n
No
n
u
n
i
f
o
r
m
lo
ca
l b
in
ar
y
p
atter
n
R
ec
o
g
n
itio
n
Un
i
f
o
r
m
lo
ca
l b
in
ar
y
p
atter
n
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
:
Srin
i
v
asa
R
ed
d
y
,
Dep
ar
te
m
en
t o
f
C
o
m
p
u
ter
Sci
en
ce
an
d
E
n
g
i
n
ee
r
in
g
,
I
n
s
tit
u
te
o
f
A
er
o
n
a
u
tica
l E
n
g
i
n
ee
r
in
g
,
Du
n
d
ig
al,
H
y
d
er
ab
ad
-
5
0
0
0
4
3
,
T
S,
I
n
d
ia.
E
m
ail:
k
o
n
d
s
r
ee
n
u
@
g
m
a
il.c
o
m
1.
I
NT
RO
D
UCT
I
O
N
Face
r
ec
o
g
n
itio
n
is
o
n
e
o
f
t
h
e
s
ig
n
i
f
ica
n
t a
n
d
p
r
o
m
i
n
en
t
r
ese
ar
ch
to
p
ics
f
o
r
t
h
e
la
s
t t
w
o
d
e
ca
d
es,
d
u
e
to
its
p
o
ten
t
ial
ap
p
licatio
n
s
a
n
d
ch
al
len
g
es.
Face
r
ec
o
g
n
iti
o
n
d
ea
ls
w
i
th
r
ec
o
g
n
itio
n
o
f
a
test
f
ac
ial
i
m
ag
e
f
r
o
m
a
f
ea
tu
r
e
lib
r
ar
y
o
r
d
ata
s
et
o
f
f
ac
ial
i
m
ag
e
s
.
T
h
e
r
esear
ch
er
s
i
n
i
m
ag
e
p
r
o
ce
s
s
i
n
g
a
n
d
f
ac
ial
i
m
a
g
e
an
al
y
tics
o
f
b
ig
d
ata
ar
e
s
h
o
w
i
n
g
tr
e
m
en
d
o
u
s
i
n
ter
est
b
ec
au
s
e
o
f
its
w
id
e
r
an
g
e
o
f
ap
p
licatio
n
s
lik
e
h
u
m
an
f
ac
e
i
n
ter
ac
tio
n
,
s
u
r
v
e
illa
n
ce
in
p
u
b
lic
a
n
d
p
r
iv
ate
m
ilit
ar
y
a
n
d
o
th
er
ar
ea
s
i
n
v
ar
io
u
s
leg
al
a
n
d
p
r
o
p
er
t
y
is
s
u
es
to
id
en
ti
f
y
t
h
e
p
r
o
p
er
p
er
s
o
n
.
T
h
e
f
ac
e
id
en
ti
f
icatio
n
ac
cu
r
ac
y
s
i
g
n
i
f
ican
tl
y
d
ec
r
e
ases
w
h
en
o
b
tai
n
ed
i
m
a
g
es d
o
n
o
t
h
av
e
ad
eq
u
ate
q
u
alit
y
e
ith
er
d
u
e
to
a
v
ar
iet
y
o
f
f
ac
ial
ex
p
r
ess
io
n
s
,
s
u
b
j
ec
t‟
s
alig
n
m
e
n
t p
r
o
b
le
m
to
th
e
ca
m
er
a,
g
az
e
d
ev
ia
tio
n
s
o
r
f
ac
ial
h
air
[
1
-
5
]
.
T
h
e
h
u
m
a
n
f
ac
e
ca
n
b
e
id
en
t
if
ied
b
ased
o
n
t
h
e
lo
ca
l
a
n
d
r
eg
io
n
al
attr
ib
u
tes.
T
o
d
ay
,
t
h
e
c
h
allen
g
e
i
n
f
ac
e
r
ec
o
g
n
itio
n
is
,
h
o
w
to
i
d
en
tify
an
d
ca
p
tu
r
e
th
i
s
in
f
o
r
m
atio
n
.
T
h
e
L
o
ca
l
B
in
ar
y
P
atter
n
(
L
B
P
)
is
a
p
o
w
er
f
u
l
lo
ca
l
d
escr
ip
to
r
f
o
r
tex
tu
r
e
a
n
al
y
s
i
s
an
d
al
s
o
ex
te
n
s
iv
el
y
u
s
ed
i
n
v
ar
io
u
s
ap
p
licat
io
n
s
[
6
]
.
T
h
e
lo
ca
l
in
f
o
r
m
atio
n
d
er
iv
ed
f
r
o
m
L
B
P
is
u
s
ed
w
id
el
y
i
n
f
ac
ia
l
an
al
y
s
is
e
s
p
ec
iall
y
f
o
r
th
e
p
u
r
p
o
s
e
o
f
f
ac
e
r
ec
o
g
n
itio
n
[
7
-
1
3
]
ag
e
cla
s
s
i
f
icatio
n
[
1
4
]
,
[
1
5
-
1
7
]
an
d
f
ac
i
al
ex
p
r
es
s
io
n
r
ec
o
g
n
i
tio
n
[
1
8
]
.
Fe
w
r
esear
c
h
er
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.
6
,
No
.
6
,
Dec
em
b
er
2
0
1
6
:
2
7
8
1
–
2
7
8
8
2782
u
s
ed
p
r
e
-
p
r
o
ce
s
s
i
n
g
m
et
h
o
d
s
to
i
m
p
r
o
v
e
class
if
ica
tio
n
p
er
f
o
r
m
a
n
ce
[
1
2
]
,
[
1
9
]
o
n
L
B
P
b
ec
au
s
e
it
i
s
m
o
r
e
p
r
o
n
e
to
n
o
is
e.
T
h
e
f
u
n
d
a
m
e
n
tal
L
B
P
o
p
er
at
o
r
h
as
s
m
all
s
p
a
tial
s
u
p
p
o
r
t
ar
ea
an
d
th
e
y
co
m
p
letel
y
f
a
il
in
ca
p
tu
r
in
g
lar
g
e
s
ca
le
s
tr
u
ct
u
r
e
(
m
ac
r
o
s
tr
u
ctu
r
e)
th
at
m
a
y
b
e
t
h
e
d
o
m
in
an
t
f
ea
t
u
r
e
o
f
f
ac
es.
I
n
th
e
lite
r
atu
r
e,
th
e
U
n
i
f
o
r
m
L
o
ca
l
B
in
ar
y
P
atter
n
(
UL
B
P
)
ar
e
u
s
ed
f
o
r
class
if
icatio
n
an
d
r
ec
o
g
n
it
io
n
p
u
r
p
o
s
e
b
ec
au
s
e
t
h
e
y
co
n
tai
n
f
u
n
d
a
m
en
ta
l
p
r
o
p
er
ties
o
f
th
e
tex
tu
r
e
an
d
i
g
n
o
r
ed
th
e
m
aj
o
r
ity
o
f
No
n
Un
if
o
r
m
L
o
c
al
B
in
ar
y
P
atter
n
(
NUL
B
P
)
as
m
is
ce
llan
eo
u
s
.
T
h
e
UL
B
P
s
alo
n
e
m
a
y
n
o
t
d
escr
ib
e
s
to
ch
a
s
tic
a
ttrib
u
te
s
a
n
d
ch
ar
ac
ter
i
s
tics
o
f
tex
t
u
r
e
ef
f
icien
tl
y
.
R
ec
e
n
tl
y
w
e
h
a
v
e
p
r
o
p
o
s
ed
a
n
e
w
v
ar
ian
t
o
f
L
B
P
ca
lled
Pro
m
i
n
en
t
L
B
P
(
P
L
B
P
)
th
at
ca
p
tu
r
es
a
s
et
o
f
UL
B
P
s
an
d
a
s
et
o
f
NUL
B
P
s
.
T
o
ca
p
tu
r
e
lo
ca
l
an
d
r
eg
io
n
al
in
f
o
r
m
at
io
n
o
f
f
ac
es
w
it
h
a
u
n
i
f
o
r
m
it
y
o
f
r
eg
io
n
a
n
d
s
u
b
-
r
eg
io
n
s
ize
t
h
e
p
r
esen
t
p
ap
er
p
r
o
p
o
s
es
MR
-
P
L
B
P
.
I
n
th
e
p
r
o
p
o
s
e
d
m
et
h
o
d
,
t
h
e
av
er
ag
e
v
alu
e
o
f
ea
ch
s
u
b
-
r
e
g
io
n
is
co
n
v
er
ted
in
to
th
e
g
r
e
y
lev
el
v
a
lu
e
o
f
n
ei
g
h
b
o
r
in
g
p
ix
els.
T
h
at‟
s
w
h
y
t
h
e
p
r
o
p
o
s
ed
m
et
h
o
d
is
m
o
r
e
r
o
b
u
s
t.
T
h
e
p
r
o
p
o
s
ed
MR
-
P
L
B
P
w
it
h
v
ar
io
u
s
v
ar
ian
ts
o
f
P
L
B
P
is
d
if
f
er
en
t
f
r
o
m
th
e
o
th
er
m
u
l
ti
-
b
lo
c
k
ap
p
r
o
ac
h
e
s
b
ec
au
s
e
th
e
y
h
a
v
e
u
s
ed
u
n
if
o
r
m
a
n
d
o
th
er
v
ar
ian
ts
o
f
L
B
P
f
ea
tu
r
es [
2
0
]
.
T
h
e
p
r
esen
t
p
ap
er
is
o
r
g
an
ize
d
as
f
o
llo
w
s
.
Sect
io
n
2
d
escr
ib
es
th
e
r
elate
d
w
o
r
k
.
T
h
e
s
ec
t
io
n
3
an
d
4
p
r
esen
ts
t
h
e
m
e
th
o
d
o
lo
g
y
a
n
d
r
esu
lt
s
an
d
d
is
c
u
s
s
io
n
.
Sectio
n
5
p
r
esen
ts
th
e
co
n
cl
u
s
io
n
s
.
2.
RE
L
AT
E
D
WO
RK
2
.
1
.
L
o
ca
l
B
ina
ry
P
a
t
t
er
n (
L
B
P
)
I
n
t
h
e
o
r
ig
i
n
al
L
B
P
,
in
tr
o
d
u
ce
d
b
y
t
h
e
Oj
ala
[
6
]
a
th
r
e
s
h
o
ld
in
g
p
r
o
ce
s
s
b
et
w
ee
n
th
e
g
r
e
y
le
v
el
v
alu
e
s
o
f
th
e
ce
n
tr
al
p
i
x
el
a
n
d
ea
ch
o
f
t
h
e
n
eig
h
b
o
r
h
o
o
d
p
ix
els
o
n
a
3
*
3
w
i
n
d
o
w
co
n
v
e
r
ts
th
e
n
ei
g
h
b
o
r
in
g
p
ix
el
v
al
u
es
in
to
a
b
i
n
ar
y
v
al
u
e.
T
h
e
b
in
ar
y
w
eig
h
t
s
ar
e
m
u
ltip
lied
w
i
th
t
h
e
b
i
n
ar
y
v
al
u
e
an
d
s
u
m
o
f
th
e
s
e
v
alu
e
s
r
es
u
lts
a
s
L
B
P
co
d
e
o
r
w
ei
g
h
t a
s
s
h
o
w
n
in
F
ig
u
r
e
1
.
68
94
55
A
f
ter
T
h
r
esh
o
ld
in
g
0
1
0
R
ep
r
esen
tat
io
n
o
f
B
in
ar
y
W
eig
h
ts
2
0
2
1
2
2
(
6
6
)
10
20
74
10
0
0
2
7
2
3
74
30
42
1
0
0
2
6
2
5
2
4
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
1
.
R
ep
r
esen
tatio
n
o
f
Ba
s
ic
L
B
P
co
d
e
(
a)
3
*
3
Neig
h
b
o
r
h
o
o
d
,
(
b
)
L
B
P Va
lu
es a
f
ter
T
h
r
esh
o
ld
in
g
,
(
c)
R
ep
r
esen
tatio
n
o
f
L
B
P
W
eig
h
ts
,
(
d
)
L
B
P
C
o
d
e
T
h
e
L
B
P
C
o
d
e
ca
n
also
b
e
d
e
r
iv
ed
f
r
o
m
t
h
e
eq
u
atio
n
1
[
2
1
]
∑
(
1
)
w
h
er
e
g
i
a
n
d
g
c
r
ep
r
esen
t
th
e
g
r
e
y
le
v
el
v
al
u
es
o
f
t
h
e
n
ei
g
h
b
o
r
in
g
an
d
ce
n
tr
al
p
ix
el
o
n
a
3
*
3
n
ei
g
h
b
o
r
h
o
o
d
,
S
r
ep
r
esen
ts
th
e
s
i
g
n
f
u
n
ctio
n
,
wh
er
e
{
T
h
e
Un
i
f
o
r
m
L
o
ca
l
B
i
n
ar
y
P
a
tter
n
(
U
L
B
P
)
is
d
er
iv
ed
o
n
t
h
e
L
B
P
to
i
m
p
r
o
v
e
t
h
e
p
er
f
o
r
m
an
ce
an
d
to
r
ed
u
ce
o
v
er
all
f
ea
tu
r
es.
T
h
e
U
L
B
P
p
r
o
v
id
es
a
m
aj
o
r
ity
o
f
p
atter
n
s
:
9
0
%
f
o
r
(
8
,
2
)
t
y
p
e
L
B
P
;
7
0
%
f
o
r
(
1
6
,
2
)
ty
p
e
L
B
P
[
2
1
]
,
th
at‟
s
w
h
y
t
h
e
y
ar
e
tr
ea
ted
a
s
f
u
n
d
a
m
en
tal
lo
ca
l
te
x
t
u
r
e
p
r
o
p
er
ties
o
f
t
h
e
i
m
a
g
e
a
n
d
r
esear
ch
er
s
co
n
s
id
er
ed
th
e
NUL
B
P
s
as
m
is
ce
lla
n
eo
u
s
.
T
h
er
e
w
ill
b
e
5
8
UL
B
P
s
an
d
1
9
8
NUL
B
P
s
o
n
a
b
asi
c
L
B
P
i.e
.
(
8
,
1
)
.
T
h
is
n
u
m
b
er
o
f
N
UL
B
P
s
i
n
cr
ea
s
e
s
d
r
asti
ca
l
l
y
a
s
w
e
in
cr
ea
s
e
t
h
e
s
ize
o
f
t
h
e
n
ei
g
h
b
o
r
h
o
o
d
o
r
t
h
e
n
u
m
b
er
n
ei
g
h
b
o
r
in
g
p
ix
el
s
.
2
.
2
.
P
r
o
m
ine
nt
L
o
ca
l B
ina
ry
P
a
t
t
en
(
P
L
B
P
)
Ma
n
y
r
esear
ch
er
s
e
x
p
r
ess
ed
t
h
eir
v
ie
w
s
o
n
th
e
ca
p
ab
ilit
y
o
f
U
L
B
P
an
d
NU
L
B
P
in
ter
m
s
o
f
tex
tu
r
e
i
m
a
g
e
an
a
l
y
s
is
,
r
ec
o
g
n
itio
n
et
c.
,
s
o
m
e
r
esear
ch
er
s
[
2
1
-
2
5
]
ex
p
lo
r
ed
ex
ten
s
iv
e
l
y
NU
L
B
P
‟
s
an
d
d
er
iv
ed
a
f
e
w
NUL
B
P
an
d
UL
B
P
in
a
f
ea
tu
r
e
v
ec
to
r
an
d
co
n
d
u
cted
ex
p
er
i
m
e
n
ts
.
H.
Z
h
o
u
et
al.
[
2
3
]
s
u
g
g
e
s
ted
th
at
U
L
B
P
alo
n
e
d
o
n
o
t
d
escr
ib
e
th
e
s
to
ch
asti
c
at
tr
ib
u
tes
o
f
tex
tu
r
e
e
f
f
i
cien
tl
y
.
A
s
a
r
es
u
lt,
te
x
tu
r
e
p
r
i
m
iti
v
e
i
n
f
o
r
m
atio
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
F
a
ce
R
ec
o
g
n
itio
n
u
s
in
g
Mu
lti R
eg
io
n
P
r
o
min
en
t LB
P
R
ep
r
esen
ta
tio
n
(
S
r
in
iva
s
a
R
ed
d
y
)
2783
r
ep
r
esen
ted
b
y
t
h
ese
p
atter
n
s
i
s
lo
s
t,
e
s
p
ec
iall
y
w
h
e
n
lar
g
e
n
eig
h
b
o
r
h
o
o
d
s
ar
e
co
n
s
id
er
ed
.
T
h
is
s
i
n
g
le
p
atter
n
m
ak
e
s
t
h
e
u
n
i
f
o
r
m
p
atter
n
s
s
en
s
it
iv
e
to
n
o
i
s
e.
A
n
e
x
ten
d
e
d
v
er
s
io
n
o
f
L
B
P
o
p
er
ato
r
(
L
B
P
ex
ten
d
)
is
also
p
r
o
p
o
s
ed
in
th
e
liter
at
u
r
e
[
2
3
]
,
w
h
ich
tr
ied
to
u
s
e
m
o
r
e
th
a
n
o
n
e
b
in
f
o
r
d
escr
ib
in
g
n
o
n
-
u
n
i
f
o
r
m
p
atter
n
s
an
d
to
r
ed
u
ce
th
e
ef
f
ec
t
o
f
n
o
is
e.
Sev
er
al
o
th
er
atte
m
p
t
s
w
er
e
also
m
ad
e
i
n
th
e
liter
at
u
r
e
to
u
s
e
n
o
n
-
u
n
i
f
o
r
m
p
atter
n
s
to
o
v
er
co
m
e
th
e
l
i
m
itatio
n
o
f
t
h
e
s
ta
n
d
ar
d
L
B
P
[
2
6
-
2
7
]
,
[
9
]
,
[
2
2
]
,
[
1
3
]
.
So
m
e
o
f
th
e
m
et
h
o
d
s
ex
tr
ac
ted
r
o
tatio
n
in
v
ar
ian
t
n
o
n
-
u
n
i
f
o
r
m
p
atter
n
s
[
2
7
]
,
[
9
]
,
[
2
2
]
.
T
h
e
p
r
esen
t
r
esear
c
h
ar
g
u
es
t
h
at
s
o
m
e
u
s
ef
u
l
in
f
o
r
m
a
tio
n
ca
n
b
e
o
b
tain
ed
b
y
u
s
i
n
g
NU
L
B
P
s
.
T
h
e
m
aj
o
r
p
r
o
b
lem
i
s
w
h
a
t
k
i
n
d
o
r
t
y
p
e
o
f
NU
L
B
P
s
to
b
e
s
elec
ted
f
r
o
m
th
e
lar
g
e
s
et
o
f
NU
L
B
P
s
.
So
f
ar
th
er
e
is
n
o
m
ec
h
an
is
m
t
h
at
d
er
iv
es
t
h
e
m
aj
o
r
it
y
o
f
th
e
U
L
B
P
s
a
n
d
a
f
e
w
o
f
NU
L
B
P
s
as
o
n
e
s
et.
A
ll
t
h
e
ab
o
v
e
r
esear
ch
er
s
co
n
s
id
er
ed
s
o
m
e
NUL
B
P
in
t
h
eir
o
w
n
w
a
y
.
T
h
is
h
as
lead
lo
t
a
m
b
ig
u
it
y
.
T
o
o
v
er
co
m
e
t
h
i
s
a
m
b
ig
u
it
y
a
n
d
to
g
i
v
e
a
s
y
s
te
m
atic
w
a
y
o
f
s
elec
t
in
g
N
UL
B
P
s
t
h
e
p
r
ese
n
t
p
ap
er
u
t
ilized
o
u
r
p
r
ev
io
u
s
d
er
iv
atio
n
ca
lled
P
r
o
m
i
n
en
t
L
o
ca
l
B
in
ar
y
P
atter
n
(
P
L
B
P
)
[
8
]
.
T
h
e
in
ter
esti
n
g
f
ea
t
u
r
e
o
f
P
L
B
P
is
t
h
at
it
co
n
tain
s
a
s
e
t
o
f
UL
B
P
s
a
n
d
NUL
B
P
s
.
T
h
e
P
r
o
m
i
n
en
t
L
o
ca
l
B
in
ar
y
P
atter
n
(
P
L
B
P
)
co
n
s
id
er
s
th
e
tr
an
s
itio
n
th
at
o
cc
u
r
s
a
f
ter
t
w
o
o
r
m
o
r
e
co
n
s
ec
u
tiv
e
ze
r
o
s
i
m
m
ed
iate
l
y
f
o
llo
w
ed
b
y
t
w
o
o
r
m
o
r
e
co
n
s
ec
u
ti
v
e
o
n
e
s
a
n
d
v
ice
v
er
s
a,
in
a
cir
c
u
lar
m
a
n
n
er
.
Fo
r
ex
a
m
p
le,
th
e
L
B
P
co
d
e
3
5
co
n
s
titu
te
s
th
e
P
L
B
P
an
d
th
e
L
B
P
co
d
e
9
6
f
o
r
m
s
a
No
n
P
r
o
m
i
n
e
n
t L
o
ca
l
B
in
ar
y
P
atter
n
(
N
P
L
B
P
)
.
T
h
e
P
L
B
P
co
n
tain
s
a
to
tal
o
f
9
2
p
atter
n
s
o
u
t
o
f
2
5
6
o
f
L
B
P
o
n
a
(
P
,
R
)
=
(
8
,
1
)
.
T
h
e
P
L
B
P
c
o
n
tain
s
4
0
UL
B
P
s
o
u
t
o
f
5
8
an
d
5
2
NUL
B
P
s
o
u
t
o
f
1
9
8
.
T
h
is
m
ea
n
s
th
e
P
L
B
P
d
is
ca
r
d
s
1
8
UL
B
Ps
an
d
1
4
6
NUL
B
P
s
in
to
o
n
e
lab
el
ca
lled
“m
i
s
ce
lla
n
eo
u
s
”.
Fo
r
th
e
e
f
f
icien
t f
ac
e
r
ec
o
g
n
itio
n
s
y
s
te
m
,
t
h
e
p
r
ese
n
t
p
ap
er
d
er
iv
ed
tw
o
v
ar
ian
t
s
o
f
P
L
B
P
,
n
a
m
el
y
Ma
x
i
m
u
m
P
r
o
m
i
n
e
n
t
L
o
ca
l
B
in
a
r
y
P
atter
n
(
MP
L
B
P
)
f
o
r
m
ed
f
r
o
m
P
L
B
P
U
U
L
B
P
an
d
S
m
al
l
P
r
o
m
i
n
en
t
L
o
ca
l
B
in
ar
y
P
at
ter
n
(
SP
L
B
P
)
f
o
r
m
ed
f
r
o
m
P
L
B
P
∩
UL
B
P
o
n
f
ac
ial
i
m
ag
e
s
.
T
h
e
SP
L
B
P
an
d
MP
L
B
P
c
o
n
tain
a
to
tal
o
f
4
0
an
d
1
1
0
p
att
er
n
s
r
esp
ec
tiv
e
l
y
o
n
a
3
x
3
n
eig
h
b
o
r
h
o
o
d
.
T
h
e
f
o
llo
w
in
g
T
ab
le
1
s
h
o
w
s
t
h
e
r
elatio
n
s
h
ip
b
et
w
ee
n
v
ar
io
u
s
v
ar
ia
n
ts
o
f
L
B
P
i.e
.
UL
B
P
,
NUL
B
P
,
P
L
B
P
,
NP
L
B
P
,
MP
L
B
P
an
d
SP
L
B
P
.
T
ab
le
1
.
R
elatio
n
s
h
ip
b
et
w
ee
n
Var
io
u
s
Var
ian
t
s
o
f
L
B
P
S.
N
o
.
B
I
N
A
R
Y
P
A
TT
ER
N
D
EC
I
M
A
L
V
A
L
U
E
U
L
B
P
N
U
L
B
P
P
L
B
P
N
P
L
B
P
M
P
L
B
P
S
P
L
B
P
1
0
0
0
0
0
0
0
0
0
X
X
X
2
0
0
0
0
0
1
1
0
6
X
X
3
0
0
1
0
0
0
1
1
35
X
X
X
4
0
0
1
1
1
1
1
0
60
X
X
5
0
1
1
0
1
1
1
1
1
1
1
X
X
X
X
6
1
0
1
1
0
0
0
1
1
7
7
X
X
X
7
1
1
1
1
1
0
1
1
2
5
1
X
X
X
8
1
1
1
1
1
1
1
1
2
5
5
X
X
X
T
h
e
u
n
io
n
o
f
P
L
B
P
a
n
d
U
L
B
P
(
P
L
B
P
U
UL
B
P
)
co
n
tain
s
a
to
tal
o
f
1
1
0
p
atter
n
s
o
u
t
o
f
wh
ich
5
8
ar
e
UL
B
P
s
an
d
5
2
ar
e
NU
L
B
P
s
.
T
h
is
s
et
i
s
n
a
m
ed
a
s
Ma
x
i
m
u
m
P
r
o
m
in
e
n
t
L
o
ca
l
B
i
n
ar
y
P
a
tter
n
(
MP
L
B
P
)
[
8
]
.
T
h
e
MP
L
B
P
tr
ea
ts
th
e
r
e
m
ai
n
in
g
1
4
6
NUL
B
P
s
as
a
m
i
s
ce
ll
an
eo
u
s
s
et.
T
h
e
in
ter
s
ec
tio
n
o
f
P
L
B
P
an
d
UL
B
P
(
P
L
B
P
∩
UL
B
P
)
is
n
a
m
ed
as
t
h
e
S
m
alle
s
t
P
r
o
m
i
n
en
t
L
o
ca
l
B
in
ar
y
P
atter
n
(
SP
L
B
P
)
[
8
]
.
T
h
e
SP
L
B
P
co
n
tain
s
a
to
tal
o
f
4
0
p
atter
n
s
o
u
t
o
f
w
h
ic
h
all
4
0
ar
e
UL
B
P
s
an
d
it
co
n
tain
s
ze
r
o
NUL
B
P
s
.
T
h
e
SP
L
B
P
tr
ea
ts
th
e
r
e
m
ain
in
g
2
1
6
L
B
P
s
(
w
h
ic
h
c
o
n
tain
1
8
UL
B
P
s
an
d
1
9
8
NUL
B
P
s
)
as a
m
is
ce
l
lan
eo
u
s
s
et.
3.
M
UL
T
I
RE
G
I
O
N
P
RO
M
I
NE
NT
L
O
CA
L
B
I
NARY
P
A
T
T
E
RN
(
M
R
-
P
L
B
P
)
T
h
e
b
asic
L
B
P
o
p
er
ato
r
s
w
it
h
an
y
(
P
,
R
)
(
w
h
er
e
P
co
r
r
esp
o
n
d
s
to
th
e
n
u
m
b
er
o
f
n
e
ig
h
b
o
r
in
g
p
i
x
els
o
n
a
cir
cle
o
f
r
a
d
iu
s
o
f
R
)
o
n
ly
ca
p
ab
le
o
f
ex
tr
ac
ti
n
g
f
ea
tu
r
es
o
n
s
m
all
s
p
atial
n
eig
h
b
o
r
h
o
o
d
i.e
.
m
icr
o
lev
el
f
ea
t
u
r
es
an
d
th
u
s
th
e
y
f
a
il
in
c
ap
tu
r
in
g
lar
g
er
s
ca
le
s
tr
u
ct
u
r
e
s
o
r
m
ac
r
o
s
tr
u
ctu
r
e
s
w
h
ic
h
ar
e
also
d
o
m
i
n
an
t
o
n
f
ac
es.
F
u
r
t
h
er
,
th
e
g
r
e
y
le
v
el
co
m
p
ar
is
o
n
m
a
y
p
r
o
n
e
to
n
o
i
s
e
w
h
e
n
d
i
f
f
er
e
n
ce
s
o
f
g
r
e
y
v
alu
es
o
f
p
ix
e
ls
ar
e
v
er
y
s
m
a
ll
o
r
eq
u
al
[
2
8
-
2
9
]
.
T
o
o
v
er
co
m
e
th
i
s
p
r
esen
t
p
ap
er
p
r
o
p
o
s
es
Mu
lti
R
e
g
io
n
-
P
r
o
m
in
en
t
L
o
ca
l
B
in
ar
y
P
atter
n
(
MR
-
P
L
B
P
)
.
T
h
e
m
ai
n
ad
v
a
n
ta
g
e
o
f
th
e
MR
-
P
L
B
P
is
th
e
s
ize
o
f
t
h
e
b
lo
ck
an
d
s
u
b
-
b
lo
ck
s
ar
e
f
i
x
ed
(
m
u
l
tip
les
o
f
3
)
.
T
h
e
p
r
ev
io
u
s
r
esear
ch
er
s
d
er
iv
ed
o
n
l
y
L
B
P
o
r
UL
B
P
f
ea
tu
r
es
ar
e
d
er
iv
ed
f
r
o
m
m
u
lti
b
lo
ck
s
b
y
tr
ea
tin
g
all
NU
L
B
P
S
o
f
t
h
e
r
eg
io
n
a
s
m
is
ce
lla
n
eo
u
s
.
I
n
th
e
p
r
o
p
o
s
ed
MR
-
P
L
B
P
th
e
s
ize
o
f
t
h
e
r
eg
io
n
is
f
i
x
ed
as
R
*
S
w
h
er
e
R
a
n
d
S
s
h
o
u
ld
b
e
m
u
ltip
le
s
o
f
t
h
r
ee
.
T
h
e
r
eg
io
n
o
f
s
ize
R
*
S
is
s
u
b
d
iv
id
ed
in
to
n
in
e
s
u
b
-
r
eg
io
n
s
o
f
s
ize
N
*
M
w
h
er
e
N
=R/3
an
d
M=
S/3
.
T
h
is
g
i
v
es
t
h
e
u
n
if
o
r
m
it
y
i
n
th
e
f
o
r
m
atio
n
o
f
MR
-
P
L
B
P
.
T
h
e
o
th
er
ad
v
an
ta
g
e
o
f
t
h
e
p
r
esen
t
m
eth
o
d
is
it e
s
ti
m
ates t
h
e
v
ar
i
o
u
s
v
ar
ia
n
ts
o
f
P
L
B
P
o
n
m
u
l
ti
r
eg
io
n
s
.
T
h
e
ea
ch
m
u
lt
i
r
eg
io
n
i
s
r
ep
r
esen
ted
b
y
a
s
in
g
le
v
al
u
e
a
n
d
th
e
v
alu
e
o
f
ea
c
h
m
u
lti
r
e
g
io
n
is
th
e
av
er
ag
e
g
r
e
y
lev
el
v
al
u
e
o
f
th
e
p
ix
els
o
f
th
at
s
u
b
-
r
eg
io
n
.
A
b
in
ar
y
co
d
e
is
d
er
iv
ed
f
r
o
m
ea
ch
s
u
b
-
r
eg
io
n
b
y
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
,
Dec
em
b
er
2
0
1
6
:
2
7
8
1
–
2
7
8
8
2784
co
m
p
ar
i
n
g
its
v
al
u
e
w
i
th
th
e
v
al
u
e
o
f
th
e
m
id
d
le
s
u
b
-
r
eg
io
n
.
T
h
e
cir
cu
lar
b
it
w
is
e
tr
an
s
it
io
n
o
f
t
h
i
s
s
u
b
-
r
eg
io
n
n
ei
g
h
b
o
r
h
o
o
d
d
er
iv
es
P
L
B
P
s
o
n
th
e
m
u
lti
r
e
g
i
o
n
.
T
h
e
s
ca
lar
v
al
u
es
i.e
.
a
v
er
ag
e
p
ix
el
g
r
e
y
lev
e
l
v
alu
e
s
o
f
ea
ch
s
u
b
-
r
e
g
io
n
o
f
s
ize
N*
M
ca
n
b
e
co
m
p
u
t
ed
v
er
y
ef
f
icie
n
tl
y
f
r
o
m
th
e
i
n
teg
r
al
i
m
a
g
e
[
3
0
]
.
T
h
er
ef
o
r
e
MR
-
P
L
B
P
f
ea
t
u
r
es
ex
tr
ac
tio
n
p
r
o
ce
s
s
i
s
v
er
y
f
ast
.
Ho
w
e
v
er
,
it
o
n
l
y
i
n
cu
r
s
a
litt
le
m
o
r
e
co
s
t
w
h
e
n
co
m
p
ar
ed
to
b
asic
L
B
P
o
p
er
ato
r
(
8
,
1
)
.
E
v
en
as
„
P
‟
i
n
cr
e
ases
t
h
e
b
asic
L
B
P
f
ea
t
u
r
e
e
x
tr
ac
tio
n
b
ec
o
m
e
s
co
s
tlier
.
T
h
e
b
asic
p
ar
am
eter
s
R
an
d
S
o
f
MR
-
P
L
B
P
in
f
lu
e
n
ce
th
e
o
v
er
all
s
tr
u
c
tu
r
e
o
f
t
h
e
f
ea
t
u
r
es.
I
f
R
an
d
S
ar
e
s
m
al
l
t
h
en
MR
-
P
L
B
P
ca
p
tu
r
es
o
n
l
y
t
h
e
lo
ca
l
f
ea
t
u
r
es
an
d
w
h
en
R
a
n
d
S
ar
e
lar
g
e
(
esp
ec
iall
y
R
a
n
d
S>=9
)
th
e
M
R
-
P
L
B
P
ca
p
tu
r
es
b
o
th
m
icr
o
an
d
m
ac
r
o
s
tr
u
ct
u
r
e
f
ea
t
u
r
es
a
n
d
esp
ec
iall
y
th
e
av
er
ag
e
g
r
e
y
le
v
el
v
alu
e
s
o
f
s
u
b
-
r
e
g
io
n
s
N
*
M
o
v
er
co
m
es
t
h
e
n
o
i
s
e
ef
f
ec
t,
m
a
k
es
M
R
-
P
L
B
P
as
r
o
b
u
s
t,
an
d
p
r
o
v
id
es
lar
g
e
s
ca
le
in
f
o
r
m
atio
n
i
n
ad
d
itio
n
to
m
ic
r
o
lev
el
i
n
f
o
r
m
atio
n
.
T
h
e
MR
-
P
L
B
P
m
ec
h
an
is
m
o
n
a
r
e
g
io
n
s
iz
e
„
S
‟
o
f
9
*
9
is
s
h
o
w
n
in
F
ig
u
r
e
2
,
th
e
b
lo
ck
s
izes a
r
e
3
*
3
.
25
24
50
45
22
33
18
19
24
55
40
65
73
17
56
31
32
12
80
75
45
55
45
68
28
27
34
14
21
22
24
50
40
29
24
13
19
23
24
40
35
23
17
28
11
18
28
29
43
25
26
45
19
39
31
32
12
24
49
26
22
25
45
28
27
34
73
87
56
37
55
53
22
30
45
20
12
13
45
80
34
51
46
25
22
34
25
29
40
44
1
1
0
0
0
0
1
1
(
5
1
)
10
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
2
.
Mu
lti R
e
g
io
n
P
r
o
m
i
n
en
t
L
o
ca
l B
in
ar
y
P
atter
n
(
MR
-
P
L
B
P
)
C
o
d
e
Ge
n
er
atio
n
,
(
a)
Div
is
io
n
o
f
R
eg
io
n
o
f
S
ize
9
*
9
i
n
to
„
9
‟
S
u
b
R
eg
io
n
s
o
f
3
*
3
(
b
)
R
ep
r
esen
tatio
n
o
f
Av
er
ag
e
Val
u
es
o
f
„
9
‟
S
u
b
-
R
eg
io
n
o
f
3
*
3
(
c)
R
ep
r
esen
ta
tio
n
r
-
Su
b
-
R
eg
io
n
s
w
i
th
B
in
ar
y
Val
u
es
(
d
)
MR
-
P
L
B
P
C
o
d
e
T
h
e
MR
-
P
L
B
P
co
d
e
is
ev
alu
a
ted
in
th
e
s
a
m
e
w
a
y
a
s
r
ep
r
esen
ted
in
E
q
u
a
tio
n
1
.
T
h
e
Fig
u
r
e
2
clea
r
ly
s
h
o
w
s
t
h
e
r
ep
r
esen
tatio
n
o
f
lar
g
e
s
tr
u
ct
u
r
es
o
r
m
ac
r
o
s
tr
u
ct
u
r
es
b
y
MR
-
P
L
B
P
.
T
h
e
r
esu
ltin
g
b
in
ar
y
p
atter
n
s
as
f
ea
t
u
r
es
o
f
MR
-
P
L
B
P
ca
n
d
etec
t
d
iv
er
s
e
i
m
ag
e
s
tr
u
c
t
u
r
es
s
u
c
h
as
lin
e
s
,
ed
g
es,
s
p
o
ts
,
co
r
n
er
s
at
d
if
f
er
en
t
s
ca
le
a
n
d
lo
ca
tio
n
.
T
h
er
e
w
il
l
b
e
f
e
w
er
n
u
m
b
er
s
o
f
M
R
-
P
L
B
P
co
d
e
f
ea
tu
r
es
w
h
e
n
co
m
p
a
r
ed
to
b
asic
L
B
P
.
A
b
asic
L
B
P
w
ill
g
e
n
er
ate
(
N
-
1
)
*
(
M
-
1
)
L
B
P
co
d
es,
w
h
er
ea
s
a
n
M
R
-
P
L
B
P
w
it
h
a
r
eg
io
n
s
iz
e
o
f
R
*
S
g
en
er
ate
s
a
to
tal
n
u
m
b
er
o
f
(
N
*
M)
/(
R
*
S
)
L
B
P
co
d
es in
a
n
o
n
-
o
v
er
lap
p
ed
m
a
n
n
er
.
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
e
p
r
o
p
o
s
ed
MR
-
P
L
B
P
u
s
ed
th
r
ee
d
if
f
er
en
t
d
atab
ases
i.e
.
Yale
,
I
n
d
ian
an
d
A
T
&
T
OR
L
.
T
h
e
p
r
esen
t
p
ap
er
co
n
s
id
er
ed
1
2
0
f
ac
ial
i
m
ag
e
s
o
u
t
o
f
1
5
p
er
s
o
n
s
w
it
h
1
1
d
if
f
er
en
t
f
ac
ial
ex
p
r
ess
io
n
s
p
er
p
er
s
o
n
as
tr
ain
in
g
s
et
f
r
o
m
Yale
d
ata
b
ase
[
3
1
]
.
T
h
e
p
r
esen
t
p
ap
er
a
ls
o
co
n
s
id
er
ed
4
7
2
f
ac
ial
i
m
a
g
es
a
s
a
tr
ain
i
n
g
s
e
t
f
r
o
m
I
n
d
ian
d
atab
ase
[
3
2
]
.
T
h
ese
4
7
2
f
ac
ial
i
m
a
g
es
co
r
r
esp
o
n
d
to
5
9
d
if
f
er
e
n
t
i
n
d
iv
id
u
al
s
o
f
b
o
th
m
ale
a
n
d
f
e
m
ale,
a
n
d
o
n
ea
c
h
i
n
d
i
v
id
u
a
l 1
1
d
if
f
er
en
t e
x
p
r
ess
io
n
s
o
f
I
n
d
ian
d
atab
ase.
T
h
e
p
r
esen
t
p
ap
er
also
co
n
s
id
er
ed
3
2
0
f
ac
ial
im
a
g
es
as
a
tr
ain
i
n
g
s
et
f
r
o
m
A
T
&
T
OR
L
d
atab
ase
[
3
3
]
f
o
r
f
ac
e
r
ec
o
g
n
itio
n
.
T
h
e
p
r
esen
t
p
ap
e
r
p
er
f
o
r
m
ed
e
x
p
er
i
m
e
n
ts
b
y
co
n
s
id
er
in
g
t
w
o
ca
s
es
f
o
r
test
d
atab
ase.
T
est
C
ase
1
:
I
n
ca
s
e1
th
e
r
e
m
ain
i
n
g
le
f
to
v
er
f
ac
ial
i
m
ag
e
s
o
f
th
e
ab
o
v
e
th
r
ee
d
atab
ases
(
w
h
ic
h
ar
e
n
o
t
co
n
s
id
er
ed
f
o
r
th
e
tr
ain
in
g
s
et)
ar
e
co
n
s
id
er
ed
as
test
i
m
ag
es.
T
est
C
ase
2
:
I
n
t
h
e
s
ec
o
n
d
ca
s
e
th
e
p
r
esen
t
p
ap
er
co
n
s
id
er
ed
th
e
test
i
m
a
g
es a
s
a
co
m
b
in
at
io
n
o
f
le
f
to
v
e
r
an
d
tr
ain
in
g
d
atab
ase
i
m
a
g
e
s
.
Fo
r
ef
f
icie
n
t
f
ac
e
r
ec
o
g
n
itio
n
,
t
h
e
p
r
esen
t
p
ap
er
ev
al
u
at
ed
h
is
to
g
r
a
m
s
o
f
L
B
P
,
UL
B
P
,
P
L
B
P
,
MP
L
B
P
an
d
SP
L
B
P
w
it
h
d
i
f
f
er
en
t
r
eg
io
n
s
izes
o
n
ea
c
h
i
n
d
iv
id
u
al
f
ac
ia
l
i
m
a
g
e
a
n
d
p
la
ce
d
in
t
h
e
tr
ain
i
n
g
d
atab
ase.
I
n
a
s
i
m
ilar
w
a
y
th
e
ab
o
v
e
h
is
to
g
r
a
m
s
ar
e
ev
alu
ate
d
f
o
r
test
f
ac
ial
i
m
ag
e
a
n
d
th
e
f
ac
e
r
ec
o
g
n
i
tio
n
i
s
ev
alu
a
ted
b
ased
o
n
C
h
i
-
Sq
u
ar
e
d
is
tan
ce
m
eth
o
d
as g
iv
e
n
i
n
Eq
u
atio
n
2
.
∑
(
2
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
F
a
ce
R
ec
o
g
n
itio
n
u
s
in
g
Mu
lti R
eg
io
n
P
r
o
min
en
t LB
P
R
ep
r
esen
ta
tio
n
(
S
r
in
iva
s
a
R
ed
d
y
)
2785
w
h
er
e
d
,
t a
r
e
t
w
o
i
m
a
g
e
f
ea
tu
r
es (
h
is
to
g
r
a
m
v
ec
to
r
s
)
an
d
R
(
d
,
t)
is
th
e
h
i
s
to
g
r
a
m
d
i
s
ta
n
ce
f
o
r
r
ec
o
g
n
itio
n
.
Fig
u
r
e
3
.
Face
R
ec
o
g
n
itio
n
R
a
te
f
o
r
Yale
D
atab
ase
f
o
r
T
est C
ase
1
Fig
u
r
e
4
.
Face
R
ec
o
g
n
itio
n
R
a
te
f
o
r
A
T
&
T
OR
L
D
atab
ase
f
o
r
T
est C
ase
1
Fig
u
r
e
5
.
Face
R
ec
o
g
n
itio
n
R
a
te
f
o
r
I
n
d
ian
Da
tab
ase
f
o
r
T
est C
ase
1
Fig
u
r
e
6
.
Face
R
ec
o
g
n
itio
n
R
a
te
f
o
r
Yale
D
atab
ase
f
o
r
T
est C
ase
2
T
h
e
g
r
ap
h
s
o
f
Fig
u
r
e
3
,
Fi
g
u
r
e
4
,
an
d
Fig
u
r
e
5
s
h
o
w
s
t
h
e
f
a
ce
r
ec
o
g
n
itio
n
r
ate
f
o
r
Yale
,
AT
&
T
OR
L
an
d
I
n
d
ian
d
atab
ases
f
o
r
T
est
C
ase
1
w
it
h
d
if
f
er
en
t
r
eg
io
n
s
izes.
T
h
e
s
a
m
e
is
al
s
o
r
ep
r
esen
ted
in
g
r
ap
h
s
o
f
Fig
u
r
e
6
,
Fig
u
r
e
7
an
d
Fig
u
r
e
8
f
o
r
T
est C
ase
2
.
T
h
e
f
o
llo
w
i
n
g
f
ac
to
r
s
ar
e
n
o
ted
d
o
w
n
f
r
o
m
t
h
e
g
r
ap
h
s
o
f
f
ig
u
r
e
s
f
r
o
m
Fi
g
u
r
e
3
to
Fi
g
u
r
e
8
.
I
n
th
e
ab
o
v
e
g
r
ap
h
s
,
t
h
e
r
eg
io
n
s
ize
o
f
3
*
3
r
ep
r
esen
ts
t
h
e
b
asic r
ep
r
esen
tatio
n
o
f
L
B
P
w
it
h
(
8
,
1
)
.
1
.
As
t
h
e
m
ac
r
o
r
eg
io
n
i
n
cr
ea
s
es
t
h
e
f
ac
ia
l
r
ec
o
g
n
itio
n
r
ate
in
cr
ea
s
e
s
s
li
g
h
t
l
y
b
y
th
e
p
r
o
p
o
s
ed
MR
P
L
B
P
an
d
its
v
ar
ia
n
ts
i.e
.
M
R
-
MP
L
B
P
an
d
MR
-
SP
L
B
P
.
T
h
is
clea
r
l
y
r
e
f
lect
s
th
e
f
ac
t
t
h
at
m
ac
r
o
s
tr
u
ct
u
r
e
f
ea
t
u
r
es
ar
e
d
o
m
i
n
a
n
t
i
n
f
ac
ia
l
i
m
a
g
es
an
d
t
h
e
y
ar
e
w
ell
ca
p
tu
r
ed
b
y
th
e
p
r
o
p
o
s
ed
MR
-
P
L
B
P
,
MR
-
SP
L
B
P
,
an
d
MR
-
MP
L
B
P
.
2
.
T
h
e
f
ac
e
r
ec
o
g
n
itio
n
r
ate
f
o
r
A
T
&
T
OR
L
a
n
d
I
n
d
ia
n
d
a
tab
ases
ar
e
h
ig
h
w
h
e
n
co
m
p
ar
ed
to
Yale
d
atab
ase
f
o
r
ca
s
e
1
an
d
ca
s
e
2
.
T
h
is
is
b
ec
au
s
e
t
h
e
Yale
d
atab
ase
is
p
r
o
n
e
to
n
o
is
e
an
d
ill
u
m
i
n
atio
n
e
f
f
ec
ts
.
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
,
Dec
em
b
er
2
0
1
6
:
2
7
8
1
–
2
7
8
8
2786
Fig
u
r
e
7
.
Face
R
ec
o
g
n
itio
n
Ra
te
f
o
r
A
T
&
T
OR
L
Dat
ab
ase
f
o
r
T
est C
ase
2
Fig
u
r
e
8
.
Face
R
ec
o
g
n
itio
n
R
a
te
f
o
r
I
n
d
ian
D
atab
ase
f
o
r
T
est C
ase
2
5.
CO
NCLU
SI
O
NS
I
n
th
i
s
p
ap
er
,
w
e
p
r
o
p
o
s
ed
Mu
lti
R
eg
io
n
P
r
o
m
i
n
e
n
t
L
o
ca
l
B
in
ar
y
P
atter
n
(
MR
-
P
L
B
P
)
,
MR
-
MP
L
B
P
an
d
MR
-
SP
L
B
P
as
a
d
escr
ip
to
r
f
o
r
f
ac
e
r
ec
o
g
n
itio
n
to
r
ef
l
ec
t
th
e
u
n
i
f
o
r
m
ap
p
ea
r
an
ce
o
f
t
h
e
f
ac
ia
l
i
m
ag
e
s
.
T
h
e
L
o
ca
l
B
in
ar
y
P
atter
n
(
L
B
P
)
is
to
o
lo
ca
l
to
b
e
r
o
b
u
s
t.
U
n
i
f
o
r
m
p
atter
n
s
m
a
y
n
o
t
r
e
m
a
i
n
t
h
e
s
a
m
e
as
th
o
s
e
d
ef
in
ed
b
y
Oj
ala
et
al
[
6
]
d
u
e
to
n
o
is
e
an
d
t
h
e
y
m
a
y
n
o
t
r
ep
r
esen
t
p
r
o
p
er
ly
th
e
s
to
ch
a
s
tic
in
f
o
r
m
a
tio
n
o
f
tex
t
u
r
es.
T
o
d
escr
ib
e
f
u
n
d
am
en
tal
a
n
d
s
to
ch
a
s
tic
a
ttrib
u
tes
e
f
f
ic
ien
t
l
y
,
t
h
e
p
r
ese
n
t
p
ap
er
d
er
iv
ed
P
L
B
P
,
MP
L
B
P
an
d
S
P
L
B
P
o
n
th
e
m
ac
r
o
s
tr
u
ct
u
r
es.
Featu
r
e
ex
t
r
ac
tio
n
f
o
r
MR
-
P
L
B
P
is
v
er
y
f
ast
u
s
in
g
in
teg
r
a
l
i
m
a
g
es.
As
t
h
e
m
ac
r
o
r
eg
io
n
i
n
cr
ea
s
es
th
e
f
ac
ial
r
ec
o
g
n
itio
n
r
ate
i
n
cr
ea
s
es
s
li
g
h
tl
y
b
y
t
h
e
p
r
o
p
o
s
ed
MR
-
P
L
B
P
an
d
its
v
ar
ia
n
t
s
.
T
h
is
clea
r
l
y
r
ef
lect
s
t
h
e
f
ac
t
m
ac
r
o
s
tr
u
ct
u
r
e
f
ea
tu
r
e
s
ar
e
d
o
m
in
a
n
t
in
f
ac
ia
l
i
m
a
g
es
a
n
d
t
h
e
y
p
la
y
a
cr
u
ci
al
r
o
le
in
f
ac
e
r
ec
o
g
n
it
io
n
t
h
an
m
icr
o
s
tr
u
ct
u
r
e
f
ea
t
u
r
es.
Mo
r
eo
v
er
,
o
u
r
f
ac
e
r
ec
o
g
n
itio
n
ap
p
r
o
ac
h
MR
-
SP
L
B
P
h
as
s
h
o
w
n
v
er
y
g
o
o
d
p
er
f
o
r
m
an
ce
o
n
all
d
atab
ase
s
w
it
h
f
e
w
er
u
n
if
o
r
m
p
atter
n
s
an
d
t
h
er
ef
o
r
e
it is
m
o
r
e
s
u
itab
le
f
o
r
r
ea
l ti
m
e
ap
p
lic
atio
n
s
.
RE
F
E
R
E
NC
E
S
[1
]
P
.
Dre
u
w
,
P
.
S
tein
g
ru
b
e
,
H.
Ha
n
s
e
lma
n
n
,
H.
Ne
y
,
a
n
d
G
.
A
a
c
h
e
n
,
"
S
URF
-
F
a
c
e
:
F
a
c
e
Re
c
o
g
n
it
io
n
u
n
d
e
r
V
i
e
w
p
o
in
t
Co
n
siste
n
c
y
Co
n
stra
in
ts
"
,
In
BM
VC
,
p
p
.
1
-
1
1
,
2
0
0
9
.
[2
]
X
.
L
u
,
a
n
d
A
.
K.
Ja
in
,
"
In
teg
ra
ti
n
g
ra
n
g
e
a
n
d
tex
tu
re
in
f
o
rm
a
ti
o
n
f
o
r
3
D
f
a
c
e
re
c
o
g
n
it
io
n
"
,
A
p
p
l
ica
t
io
n
o
f
C
o
mp
u
ter
Vi
sio
n
,
S
e
v
e
n
t
h
IEE
E
W
o
rk
sh
o
p
,
V
o
l
.
1
,
p
p
.
1
5
6
-
1
6
3
,
Ja
n
.
2
0
0
5
.
[3
]
J.
Ru
iz
-
d
e
l
-
S
o
lar,
R.
V
e
rsc
h
a
e
,
a
n
d
M
.
Co
rre
a
,
"
Re
c
o
g
n
it
io
n
o
f
f
a
c
e
s in
u
n
c
o
n
s
train
e
d
e
n
v
ir
o
n
m
e
n
ts:
a
c
o
m
p
a
ra
ti
v
e
stu
d
y
"
,
EURA
S
IP
J
o
u
r
n
a
l
o
n
A
d
v
a
n
c
e
s in
S
i
g
n
a
l
Pro
c
e
ss
in
g
,
V
o
l.
2
0
0
9
,
N
o
.
1
,
p
p
.
1
-
2
0
,
M
a
r.
2
0
0
9
.
[4
]
J.
W
rig
h
t,
a
n
d
G
.
Hu
a
,
"
Im
p
li
c
it
e
las
ti
c
m
a
t
c
h
in
g
w
it
h
ra
n
d
o
m
p
ro
jec
ti
o
n
s
f
o
r
p
o
se
-
v
a
rian
t
f
a
c
e
re
c
o
g
n
it
io
n
"
,
Co
m
p
u
ter
Vi
si
o
n
a
n
d
Pa
t
ter
n
Rec
o
g
n
it
io
n
-
IEE
E
Co
n
fer
e
n
c
e
,
p
p
.
1
5
0
2
-
1
5
0
9
,
Ju
n
.
2
0
0
9
.
[5
]
G
.
Zh
a
o
,
a
n
d
M
.
P
iet
ik
a
in
e
n
,
"
Dy
n
a
m
ic
te
x
tu
re
re
c
o
g
n
it
io
n
u
sin
g
l
o
c
a
l
b
in
a
ry
p
a
tt
e
rn
s
w
it
h
a
n
a
p
p
l
i
c
a
ti
o
n
t
o
f
a
c
ial
e
x
p
re
ss
io
n
s"
,
Pa
tt
e
rn
An
a
lys
is
a
n
d
M
a
c
h
in
e
In
telli
g
e
n
c
e
,
IE
EE
T
ra
n
sa
c
ti
ons
,
v
o
l.
2
9
,
n
o
.
6
,
p
p
.
9
1
5
-
9
2
8
,
Ju
n
.
2
0
0
7
.
[6
]
T
.
Oja
la,
M
.
P
ietik
a
in
e
n
,
a
n
d
D.
Ha
r
w
o
o
d
,
"
A
c
o
m
p
a
ra
ti
v
e
stu
d
y
o
f
te
x
tu
re
m
e
a
su
re
s
w
it
h
c
la
ss
i
f
ic
a
ti
o
n
b
a
se
d
o
n
f
e
a
tu
re
d
d
istri
b
u
ti
o
n
s”
,
Pa
tt
e
rn
re
c
o
g
n
i
ti
o
n
,
Vo
l.
2
9
,
No
.
1
,
p
p
.
5
1
-
5
9
,
Ja
n
.
1
9
9
6
.
[7
]
T
.
A
h
o
n
e
n
,
A
.
H
a
d
id
,
a
n
d
M
.
P
ietik
a
in
e
n
,
"
F
a
c
e
d
e
sc
rip
ti
o
n
w
it
h
lo
c
a
l
b
in
a
ry
p
a
tt
e
rn
s:
A
p
p
li
c
a
ti
o
n
t
o
f
a
c
e
re
c
o
g
n
it
io
n
"
,
Pa
tt
e
rn
An
a
lys
is
a
n
d
M
a
c
h
in
e
In
telli
g
e
n
c
e
,
IEE
E
T
r
a
n
sa
c
ti
o
n
s
,
v
o
l
.
2
8
n
o
.
1
2
,
p
p
.
2
0
3
7
-
2
0
4
1
,
De
c
.
2
0
0
6
.
[8
]
V
.
V
ij
a
y
a
Ku
m
a
r,
K.
S
rin
iv
a
sa
R
e
d
d
y
,
a
n
d
V
.
V
e
n
k
a
ta
Krish
n
a
.
"
F
a
c
e
Re
c
o
g
n
it
io
n
u
sin
g
P
ro
m
in
e
n
t
L
BP
M
o
d
e
l"
,
In
ter
n
a
t
io
n
a
l
J
o
u
rn
a
l
o
f
A
p
p
li
e
d
En
g
i
n
e
e
rin
g
Res
e
a
rc
h
,
V
o
l
.
1
0
,
N
o
.
2
,
p
p
.
4
3
7
3
-
4
3
8
4
,
M
a
r.
2
0
1
5
.
[9
]
S
.
L
iao
,
a
n
d
A
.
C.
Ch
u
n
g
,
"
F
a
c
e
re
c
o
g
n
it
io
n
b
y
u
sin
g
e
lo
n
g
a
te
d
lo
c
a
l
b
i
n
a
ry
p
a
tt
e
rn
s
w
it
h
a
v
e
ra
g
e
m
a
x
i
m
u
m
d
istan
c
e
g
ra
d
ien
t
m
a
g
n
it
u
d
e
"
,
Co
mp
u
ter
Vi
si
o
n
–
S
p
rin
g
e
r
,
p
p
.
6
7
2
-
6
7
9
,
No
v
.
2
0
0
7
.
[1
0
]
L
.
Na
n
n
i,
a
n
d
A
.
L
u
m
in
i,
"
Re
g
io
n
Bo
o
st
lea
rn
i
n
g
f
o
r
2
D+
3
D
b
a
se
d
f
a
c
e
re
c
o
g
n
it
io
n
"
,
Pa
tt
e
rn
Rec
o
g
n
it
i
o
n
L
e
tt
e
rs
,
V
o
l
.
2
8
,
N
o
.
1
5
,
p
p
.
2
0
6
3
-
2
0
7
0
,
No
v
.
2
0
0
7
.
[1
1
]
L
.
Na
n
n
i,
S
.
Bra
h
n
a
m
,
a
n
d
A
.
L
u
m
in
i,
"
A
lo
c
a
l
a
p
p
ro
a
c
h
b
a
s
e
d
o
n
a
L
o
c
a
l
Bin
a
ry
P
a
tt
e
rn
s
v
a
rian
t
tex
tu
re
d
e
sc
rip
to
r
f
o
r
c
las
sify
in
g
p
a
in
sta
tes
"
,
Exp
e
rt
S
y
ste
ms
wit
h
Ap
p
li
c
a
ti
o
n
s
,
V
o
l.
3
7
,
No
.
1
2
,
p
p
.
7
8
8
8
-
7
8
9
4
,
De
c
.
2
0
1
0
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
F
a
ce
R
ec
o
g
n
itio
n
u
s
in
g
Mu
lti R
eg
io
n
P
r
o
min
en
t LB
P
R
ep
r
esen
ta
tio
n
(
S
r
in
iva
s
a
R
ed
d
y
)
2787
[1
2
]
W
.
Zh
a
n
g
,
S
.
S
h
a
n
,
W
.
G
a
o
,
X
.
Ch
e
n
,
a
n
d
H.
Zh
a
n
g
,
"
L
o
c
a
l
g
a
b
o
r
b
i
n
a
ry
p
a
tt
e
rn
h
ist
o
g
ra
m
s
e
q
u
e
n
c
e
(lg
b
p
h
s):
A
n
o
v
e
l
n
o
n
-
sta
ti
stica
l
m
o
d
e
l
f
o
r
fa
c
e
re
p
re
se
n
tatio
n
a
n
d
re
c
o
g
n
it
i
o
n
"
,
Co
mp
u
ter
Vi
sio
n
,
T
e
n
th
IEE
E
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
,
V
o
l
.
1
,
p
p
.
7
8
6
-
7
9
1
,
Oc
t.
2
0
0
5
.
[1
3
]
S
a
b
in
a
Ya
s
m
in
,
M
d
.
M
a
su
d
Ra
n
a
,
“
P
e
rf
o
r
m
a
n
c
e
S
tu
d
y
o
f
S
o
f
t
L
o
c
a
l
Bin
a
r
y
P
a
tt
e
rn
o
v
e
r
L
o
c
a
l
Bin
a
r
y
P
a
tt
e
rn
u
n
d
e
r
No
isy
Im
a
g
e
s”
,
In
ter
n
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
El
e
c
trica
l
a
n
d
Co
mp
u
ter
E
n
g
i
n
e
e
rin
g
(
IJ
ECE
)
,
Vo
l.
6
,
No
.
3
,
p
p
.
1
1
6
1
-
1
1
6
7
,
J
u
n
e
2
0
1
6
.
[1
4
]
M
.
Ch
a
n
d
ra
M
o
h
a
n
,
V
.
V
ij
a
y
a
Ku
m
a
r,
a
n
d
B.
S
u
jath
a
,
"
Clas
sifica
ti
o
n
o
f
c
h
il
d
a
n
d
a
d
u
lt
b
a
se
d
o
n
g
e
o
m
e
tri
c
f
e
a
tu
re
s
o
f
fa
c
e
u
sin
g
li
n
e
a
r
wa
v
e
lets"
,
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
S
ig
n
a
l
Pro
c
e
ss
in
g
,
Ima
g
e
Pro
c
e
ss
in
g
a
n
d
P
a
tt
e
r
n
Rec
o
g
n
it
io
n
(
IJ
S
IP)
,
v
o
l.
1
,
n
o
.
3
,
p
p
.
2
1
1
-
2
2
0
,
M
a
y
2
0
1
0,
[1
5
]
A
.
G
u
n
a
y
,
a
n
d
V
.
V
.
Na
b
iy
e
v
,
"
Au
to
m
a
ti
c
a
g
e
c
las
si
f
ic
a
ti
o
n
w
it
h
L
BP
"
,
Co
mp
u
ter
a
n
d
I
n
f
o
rm
a
ti
o
n
S
c
ien
c
e
s,
2
3
r
d
In
ter
n
a
t
io
n
a
l
S
y
mp
o
si
u
m
,
p
p
.
1
-
4
,
Oc
t.
2
0
0
8
.
[1
6
]
V
.
V
.
Ku
m
a
r,
J.S
.
Kira
n
,
a
n
d
V
.
H
.
Ch
a
n
d
a
n
a
,
"
A
n
E
ff
e
c
ti
v
e
Ag
e
Clas
sif
ic
a
ti
o
n
Us
in
g
T
o
p
o
lo
g
ica
l
F
e
a
tu
re
s
Ba
s
e
d
o
n
Co
m
p
re
ss
e
d
a
n
d
Re
d
u
c
e
d
G
re
y
L
e
v
e
l
M
o
d
e
l
o
f
th
e
F
a
c
ial
S
k
in
"
,
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Ima
g
e
,
Gr
a
p
h
ics
a
n
d
S
ig
n
a
l
Pro
c
e
ss
in
g
(
IJ
IGS
P),
v
o
l
.
6
,
No
.
1
,
p
p
.
9
-
1
7
,
N
o
v
.
2
0
1
3
.
[1
7
]
B.
E.
Re
d
d
y
,
P
.
C.
S
.
Re
d
d
y
,
a
n
d
V
.
V
.
Ku
m
a
r,
"
Tex
to
n
Ba
se
d
S
h
a
p
e
F
e
a
tu
re
s
o
n
L
o
c
a
l
Bin
a
ry
P
a
tt
e
rn
f
o
r
Ag
e
Clas
sif
ic
a
ti
o
n
"
,
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Ima
g
e
,
Gr
a
p
h
ics
a
n
d
S
ig
n
a
l
Pr
o
c
e
ss
in
g
,
Vo
l.
4
,
N
o
.
7
,
p
p
.
54
-
6
0
,
J
u
l
.
2
0
1
2
.
[1
8
]
D.C.
He
,
a
n
d
L
.
W
a
n
g
,
"
T
e
x
tu
re
u
n
it
,
tex
tu
re
sp
e
c
tru
m
,
a
n
d
tex
tu
re
a
n
a
ly
sis"
,
Ge
o
sc
i
e
n
c
e
a
n
d
R
e
mo
te
S
e
n
sin
g
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
,
Vo
l.
2
8
,
n
o
.
4
,
p
p
.
5
0
9
-
5
1
2
,
1
9
9
0
.
[1
9
]
[1
9
]
L
.
Na
n
n
i,
a
n
d
A
.
L
u
m
in
i,
"
Lo
c
a
l
b
i
n
a
ry
p
a
tt
e
rn
s
f
o
r
a
h
y
b
rid
f
in
g
e
rp
rin
t
m
a
tch
e
r"
,
P
a
tt
e
rn
re
c
o
g
n
it
io
n
,
v
o
l
.
4
1
,
n
o
.
1
1
,
p
p
.
3
4
6
1
-
3
4
6
6
,
No
v
.
2
0
0
8
.
[2
0
]
T
.
A
h
o
n
e
n
,
A
.
Ha
d
id
,
a
n
d
M
.
P
i
e
ti
k
a
in
e
n
,
"
F
a
c
e
re
c
o
g
n
it
io
n
w
it
h
lo
c
a
l
b
in
a
ry
p
a
tt
e
rn
s"
,
In
Co
m
p
u
ter
v
isio
n
-
e
c
c
v
2
0
0
4
,
p
p
.
4
6
9
-
4
8
1
,
2
0
0
4
.
[2
1
]
Z.
G
u
o
,
L
.
Zh
a
n
g
,
D.Z
h
a
n
g
,
a
n
d
X
.
M
o
u
,
"
Hie
ra
rc
h
ica
l
m
u
lt
isc
a
le
L
BP
f
o
r
f
a
c
e
a
n
d
p
a
lm
p
rin
t
re
c
o
g
n
it
io
n
"
,
Im
a
g
e
Pro
c
e
ss
in
g
,
1
7
t
h
IEE
E
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
,
p
p
.
4
5
2
1
-
4
5
2
4
,
S
e
p
t.
2
0
1
0
,
[2
2
]
T
.
Oja
la,
M
.
P
ietik
a
in
e
n
,
a
n
d
T
.
M
a
e
n
p
a
a
,
"
M
u
lt
ires
o
l
u
ti
o
n
g
ra
y
-
sc
a
le
a
n
d
ro
tatio
n
i
n
v
a
rian
t
tex
tu
re
c
las
si
f
ica
ti
o
n
w
it
h
lo
c
a
l
b
in
a
ry
p
a
tt
e
rn
s"
,
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
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
,
Vo
l.
2
4
,
No
.
7
,
p
p
.
9
7
1
-
9
8
7
,
Ju
l
.
2
0
0
2
.
[2
3
]
S
.
L
iao
,
M
.
W
.
La
w
,
a
n
d
A
.
Ch
u
n
g
,
"
Do
m
in
a
n
t
lo
c
a
l
b
in
a
ry
p
a
tt
e
rn
s
f
o
r
tex
tu
re
c
la
ss
i
f
ica
ti
o
n
"
,
Ima
g
e
Pro
c
e
ss
in
g
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
,
Vo
l.
1
8
,
No
.
5
,
p
p
.
1
1
0
7
-
1
1
1
8
,
M
a
y
,
2
0
0
9
.
[2
4
]
H
.
Zh
o
u
,
R.
W
a
n
g
,
a
n
d
C.
W
a
n
g
,
"
A
n
o
v
e
l
e
x
ten
d
e
d
lo
c
a
l
-
b
in
a
ry
-
p
a
tt
e
rn
o
p
e
ra
t
o
r
f
o
r
t
e
x
tu
re
a
n
a
ly
sis"
,
In
fo
rm
a
t
io
n
S
c
ien
c
e
s
,
Vo
l.
1
7
8
,
No
.
2
2
,
p
p
.
4
3
1
4
-
4
3
2
5
,
N
o
v
.
2
0
0
8
.
[2
5
]
K.
S
rin
iv
a
sa
Re
d
d
y
,
V
.
V
e
n
k
a
ta
Krish
n
a
,
a
n
d
B.
Esw
a
ra
R
e
d
d
y
,
"
F
a
c
ial
R
e
c
o
g
n
it
io
n
u
si
n
g
th
e
c
o
m
b
in
a
ti
o
n
o
f
sig
n
if
ica
n
t
n
o
n
u
n
if
o
rm
lo
c
a
l
b
in
a
ry
p
a
tt
e
rn
s"
,
G
ra
p
h
ics
Vi
sio
n
a
n
d
Im
a
g
e
Pro
c
e
ss
in
g
,
V
o
l
.
1
5
,
No
.
1
,
p
p
1
-
8
,
Ju
n
.
2
0
1
5
,
[2
6
]
K.S
.
Re
d
d
y
,
V
.
V
.
Krish
n
a
,
a
n
d
V
.
V
.
Ku
m
a
r,
“
A
M
e
th
o
d
f
o
r
F
a
c
ial
Re
c
o
g
n
it
io
n
Ba
se
d
On
Lo
c
a
l
F
e
a
tu
re
s”
,
In
ter
n
a
t
io
n
a
l
J
o
u
rn
a
l
o
f
M
a
t
h
e
ma
ti
c
s
&
Co
mp
u
ta
ti
o
n
,
Vo
l.
2
7
,
Iss
.
3
,
p
p
.
9
8
-
1
0
9
,
2
0
1
5
.
[2
7
]
A
.
F
a
th
i,
a
n
d
A
.
R.
Na
g
h
sh
-
Nilch
i,
"
No
ise
to
lera
n
t
l
o
c
a
l
b
i
n
a
ry
p
a
tt
e
rn
o
p
e
ra
to
r
f
o
r
e
ff
icie
n
t
tex
tu
re
a
n
a
ly
sis
"
,
Pa
tt
e
rn
Rec
o
g
n
it
io
n
L
e
tt
e
rs
,
v
o
l
.
3
3
,
n
o
.
9
,
p
p
.
1
0
9
3
-
1
1
0
0
,
Ju
l
.
2
0
1
2
.
[2
8
]
M
.
He
ik
k
il
a
,
M
.
P
ietik
a
in
e
n
,
a
n
d
C.
S
c
h
m
id
,
"
De
sc
rip
ti
o
n
o
f
in
tere
st
re
g
io
n
s
w
it
h
lo
c
a
l
b
in
a
ry
p
a
tt
e
rn
s"
,
Pa
tt
e
r
n
re
c
o
g
n
it
i
o
n
,
v
o
l.
4
2
,
n
o
.
3
,
p
p
.
4
2
5
-
4
3
6
,
M
a
r.
2
0
0
9
.
[2
9
]
K.
S
rin
iv
a
sa
Re
d
d
y
,
V
.
V
i
jay
a
K
u
m
a
r,
B.
Es
w
a
ra
R
e
d
d
y
,
"
F
a
c
e
r
e
c
o
g
n
it
io
n
b
a
s
e
d
o
n
tex
tu
re
f
e
a
tu
re
s
u
sin
g
lo
c
a
l
tern
a
ry
p
a
tt
e
rn
s"
,
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Ima
g
e
,
Gr
a
p
h
ics
a
n
d
S
i
g
n
a
l
Pro
c
e
ss
in
g
,
Vo
l.
7
,
No
.
1
0
,
p
p
.
37
-
4
6
,
2
0
1
5
.
[3
0
]
X
.
T
a
n
,
a
n
d
B.
T
rig
g
s,
"
En
h
a
n
c
e
d
lo
c
a
l
tex
tu
re
f
e
a
tu
re
s
e
ts
f
o
r
fa
c
e
re
c
o
g
n
it
io
n
u
n
d
e
r
d
if
f
icu
lt
li
g
h
ti
n
g
c
o
n
d
i
ti
o
n
s"
,
Ima
g
e
Pro
c
e
ss
in
g
,
IEE
E
T
ra
n
sa
c
ti
o
n
s,
Vo
l.
1
9
,
No
.
6
,
p
p
.
1
6
3
5
-
1
6
5
0
,
Ju
n
.
2
0
1
0
.
[3
1
]
P
.
V
io
la,
a
n
d
M
.
Jo
n
e
s,
"
Ra
p
id
o
b
jec
t
d
e
tec
ti
o
n
u
si
n
g
a
b
o
o
ste
d
c
a
sc
a
d
e
o
f
si
m
p
le
f
e
a
tu
re
s
"
,
Co
m
p
u
ter
Visio
n
a
n
d
P
a
tt
e
r
n
Re
c
o
g
n
it
i
o
n
,
IEE
E
C
o
mp
u
ter
S
o
c
iety
C
o
n
fer
e
n
c
e
,
V
o
l
.
1
,
p
p
.
I
-
5
1
1
-
I
-
5
1
8
,
2
0
0
1
.
[3
2
]
P
.
N.
Be
lh
u
m
e
u
r,
a
n
d
D.J
.
Krie
g
m
a
n
,
"
T
h
e
Ya
le
f
a
c
e
d
a
tab
a
se
.
UR
L
:
h
tt
p
:
//
c
v
c
.
y
a
le.
e
d
u
/p
r
o
jec
ts/y
a
le
fa
c
e
s/
y
a
le
f
a
c
e
s.h
tm
l
"
,
v
o
l.
1
,
n
o
.
2
,
p
p
.
4
,
1
9
9
7
.
[3
3
]
V
.
Ja
i
n
,
a
n
d
A
.
M
u
k
h
e
rjee
,
"
T
h
e
In
d
ian
f
a
c
e
d
a
tab
a
se
"
,
2
0
0
2
.
B
I
O
G
RAP
H
I
E
S
O
F
AUTH
O
R
S
Dr
.
S
r
in
iv
a
sa
Re
d
d
y
K
re
c
e
iv
e
d
th
e
B
.
T
e
c
h
.
(CS
E)
d
e
g
re
e
f
ro
m
JN
T
U,
H
y
d
e
r
a
b
a
d
in
2
0
0
3
,
M
a
ste
rs
De
g
re
e
in
M
.
Tec
h
.
f
ro
m
S
RM
Un
iv
e
rsit
y
,
Ch
e
n
n
a
i,
T
.
N.,
In
d
ia
in
2
0
0
5
.
He
re
c
e
iv
e
d
P
h
.
D.
in
F
a
c
u
l
ty
o
f
Co
m
p
u
ter
S
c
ien
c
e
&
En
g
in
e
e
rin
g
f
ro
m
JN
T
U
A
,
A
n
a
n
tap
u
ra
m
,
A
.
P
.
,
In
d
ia,
in
2
0
1
6
.
A
t
p
re
se
n
t,
h
e
is
w
o
r
k
in
g
a
s
P
r
o
f
e
ss
o
r
in
I
n
stit
u
te
o
f
Ae
ro
n
a
u
ti
c
a
l
E
n
g
in
e
e
rin
g
,
Du
n
d
ig
a
l,
Hy
d
e
ra
b
a
d
,
T
.
S
,
In
d
i
a
.
His
re
se
a
rc
h
in
tere
sts
in
c
lu
d
e
Im
a
g
e
P
ro
c
e
ss
in
g
,
c
o
m
p
u
ter
v
isio
n
,
a
n
d
m
a
c
h
in
e
lea
r
n
in
g
.
He
h
a
s p
u
b
li
sh
e
d
re
se
a
rc
h
p
a
p
e
rs i
n
v
a
rio
u
s Na
ti
o
n
a
l,
I
n
tern
a
ti
o
n
a
l
c
o
n
f
e
re
n
c
e
s an
d
Jo
u
rn
a
ls.
He
is
a
li
f
e
m
e
m
b
e
r
o
f
IS
T
E,
CS
I.
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
,
Dec
em
b
er
2
0
1
6
:
2
7
8
1
–
2
7
8
8
2788
Dr
.
V
ij
a
y
a
K
u
m
a
r
V
is
w
o
rk
i
n
g
a
s
De
a
n
in
De
p
t.
o
f
CS
E
&
IT
a
n
d
Dire
c
to
r
Ce
n
tre
f
o
r
A
d
v
a
n
c
e
d
Co
m
p
u
tatio
n
a
l
Re
se
a
rc
h
(CACR)
a
t
A
n
u
ra
g
G
ro
u
p
o
f
In
stit
u
ti
o
n
s,
(AG
OI)
(A
u
to
n
o
m
o
u
s),
Hy
d
e
ra
b
a
d
.
He
re
c
e
iv
e
d
in
teg
ra
ted
M
.
S
.
En
g
g
,
i
n
CS
E
f
ro
m
USS
R
in
1
9
8
9
.
He
re
c
e
iv
e
d
h
is
P
h
.
D.
d
e
g
re
e
in
Co
m
p
u
ter
S
c
ien
c
e
f
ro
m
Ja
w
a
h
a
rlal
Ne
h
ru
T
e
c
h
n
o
l
o
g
ica
l
Un
iv
e
rsit
y
(JN
T
U),
H
y
d
e
ra
b
a
d
,
In
d
ia
in
1
9
9
8
a
n
d
g
u
id
e
d
2
4
re
se
a
rc
h
sc
h
o
lars
f
o
r
P
h
.
D.
He
h
a
s
se
rv
e
d
JN
T
Un
iv
e
rsit
y
f
o
r
1
3
y
e
a
rs
a
s
As
sis
tan
t
P
r
o
f
e
ss
o
r
a
n
d
A
ss
o
c
iate
P
ro
f
e
s
so
r.
He
h
a
s
re
c
e
iv
e
d
b
e
st
re
se
a
rc
h
e
r
a
n
d
b
e
st
tea
c
h
e
r
a
w
a
r
d
f
ro
m
JN
T
Un
iv
e
r
sit
y
,
Ka
k
in
a
d
a
,
In
d
ia.
His
re
se
a
rc
h
in
tere
sts
in
c
lu
d
e
Im
a
g
e
P
ro
c
e
ss
in
g
,
P
a
tt
e
rn
Re
c
o
g
n
it
io
n
,
Dig
it
a
l
W
a
ter
M
a
rk
in
g
,
Clo
u
d
Co
m
p
u
ti
n
g
,
Im
a
g
e
Re
tri
e
v
a
l
S
y
ste
m
s
a
n
d
im
a
g
e
a
n
a
l
y
ti
c
s
in
Big
Da
ta.
He
is
t
h
e
li
f
e
m
e
m
b
e
r
o
f
CS
I,
IS
CA
,
IS
T
E,
IE
(I),
IET
E,
A
CCS
,
CRS
I
,
IRS
,
a
n
d
REDCROS
S
.
He
p
u
b
l
ish
e
d
m
o
re
th
a
n
1
2
5
r
e
se
a
rc
h
p
u
b
li
c
a
ti
o
n
s
ti
ll
n
o
w
in
v
a
rio
u
s
Na
ti
o
n
a
l,
I
n
tern
a
ti
o
n
a
l
jo
u
rn
a
ls
a
n
d
c
o
n
f
e
re
n
c
e
s.
He
h
a
s
a
lso
e
sta
b
li
sh
e
d
a
n
d
a
lso
a
c
ted
a
s
a
He
a
d
,
S
rin
iv
a
sa
Ra
m
a
n
u
jan
Re
se
a
rc
h
F
o
ru
m
(S
RRF
)
a
t
G
IE
T
,
Ra
jah
m
u
n
d
ry
,
In
d
ia f
o
r
p
ro
m
o
ti
n
g
re
se
a
r
c
h
a
n
d
so
c
ial
a
c
ti
v
it
ies
.
Dr
.
V
e
n
k
a
t
a
K
r
ish
n
a
V
re
c
e
iv
e
d
t
h
e
B.
T
e
c
h
.
(ECE
)
d
e
g
re
e
f
ro
m
S
ri
V
e
n
k
a
tes
w
a
r
a
Un
iv
e
rsit
y
.
He
c
o
m
p
lete
d
h
is
M
.
T
e
c
h
.
(Co
m
p
u
ter
S
c
ien
c
e
)
f
ro
m
JN
T
Un
iv
e
rsit
y
.
He
re
c
e
iv
e
d
h
is
P
h
.
D.
d
e
g
re
e
in
Co
m
p
u
ter
S
c
ien
c
e
f
ro
m
Ja
wa
h
a
rlal
Ne
h
ru
T
e
c
h
n
o
lo
g
ica
l
Un
iv
e
rsi
ty
(
JN
T
U),
H
y
d
e
ra
b
a
d
,
In
d
ia
in
2
0
0
4
.
He
w
o
rk
e
d
a
s
P
ro
f
e
ss
o
r
a
n
d
He
a
d
o
f
th
e
De
p
a
rtm
e
n
t
f
o
r
1
0
y
e
a
r
s
in
M
a
h
a
tm
a
G
a
n
d
h
i
In
stit
u
te
o
f
T
e
c
h
n
o
lo
g
y
,
H
y
d
e
r
a
b
a
d
.
He
w
o
rk
e
d
a
s
P
rin
c
ip
a
l
f
o
r
V
i
d
y
a
V
ik
a
s
Co
ll
e
g
e
o
f
En
g
in
e
e
rin
g
,
JN
T
U,
H
y
d
e
r
a
b
a
d
,
Ch
a
it
a
n
y
a
In
stit
u
te
o
f
S
c
ien
c
e
&
Tec
h
n
o
lo
g
y
,
JN
TU,
Ka
k
in
a
d
a
.
G
o
d
a
v
a
ri
In
stit
u
te
o
f
En
g
in
e
e
rin
g
a
n
d
T
e
c
h
n
o
lo
g
y
,
Ra
jah
m
u
n
d
ry
,
JN
T
U,
Ka
k
in
a
d
a
.
A
t
p
re
se
n
t
He
is
w
o
rk
in
g
a
s
P
ro
f
e
ss
o
r,
CS
E
De
p
a
rtm
e
n
t
in
V
id
y
a
Jy
o
th
i
In
stit
u
te
o
f
T
e
c
h
n
o
lo
g
y
,
H
y
d
e
ra
b
a
d
,
In
d
ia.
He
is
a
n
a
d
v
is
o
ry
m
e
m
b
e
r
f
o
r
m
a
n
y
En
g
in
e
e
rin
g
c
o
ll
e
g
e
s.
He
P
u
b
li
sh
e
d
m
o
re
th
a
n
4
0
re
se
a
rc
h
p
u
b
l
ica
ti
o
n
s
i
n
v
a
rio
u
s
Na
ti
o
n
a
l,
In
tern
a
ti
o
n
a
l
jo
u
rn
a
ls
a
n
d
c
o
n
f
e
re
n
c
e
s.
G
u
id
e
d
4
re
se
a
rc
h
sc
h
o
lars
f
o
r
P
h
.
D.
He
is
a
li
f
e
m
e
m
b
e
r
o
f
IS
T
E
a
n
d
CS
I.
Evaluation Warning : The document was created with Spire.PDF for Python.