I
nte
rna
t
io
na
l J
o
urna
l o
f
I
nfo
rm
a
t
ics a
nd
Co
m
m
un
ica
t
io
n T
ec
hn
o
lo
g
y
(
I
J
-
I
CT
)
Vo
l.
13
,
No
.
3
,
Dec
em
b
er
20
24
,
p
p
.
3
3
3
~
3
4
3
I
SS
N:
2252
-
8
7
7
6
,
DOI
:
1
0
.
1
1
5
9
1
/iji
ct
.
v
13
i
3
.
pp
333
-
3
4
3
333
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ict.
ia
esco
r
e.
co
m
Therm
a
l
ima
g
ing
-
ba
sed iden
tif
ica
ti
o
n of f
a
cia
l f
ea
tur
es in nois
y
env
iro
nment
P
a
la
k
M
a
ha
j
a
n
1
,
P
a
wa
nes
h Abro
l
2
,
P
a
rv
ee
n K
um
a
r
L
eha
na
3
1
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
C
e
n
t
r
a
l
U
n
i
v
e
r
si
t
y
o
f
J
a
mm
u
,
Ja
mm
u
a
n
d
K
a
s
h
mi
r
(
J&K
)
,
I
n
d
i
a
2
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
IT
,
U
n
i
v
e
r
si
t
y
o
f
J
a
mm
u
,
Jamm
u
a
n
d
K
a
s
h
m
i
r
(
J&
K
)
,
I
n
d
i
a
3
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
o
n
i
c
s
,
U
n
i
v
e
r
s
i
t
y
o
f
Jam
mu
,
Jam
mu
a
n
d
K
a
s
h
mi
r
(
J&
K
)
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
an
1
6
,
2
0
2
4
R
ev
is
ed
Ma
y
1
5
,
2
0
2
4
Acc
ep
ted
J
u
n
1
8
,
2
0
2
4
F
a
c
e
id
e
n
ti
fica
ti
o
n
is
a
m
o
n
g
s
t
th
e
m
o
st
e
ffic
a
c
io
u
s
a
n
d
e
x
ten
siv
e
a
p
p
li
c
a
ti
o
n
s
i
n
b
i
o
m
e
tri
c
s
in
v
o
l
v
in
g
e
x
trac
ti
o
n
a
n
d
lo
c
a
ti
n
g
fa
c
ial
fe
a
tu
re
s.
Wi
th
id
e
n
t
ifi
c
a
ti
o
n
b
e
in
g
m
o
n
o
to
n
o
u
s
tas
k
a
tt
rib
u
tab
le
to
re
li
a
n
c
e
o
n
p
a
ra
m
e
ters
li
k
e
v
a
ried
c
a
m
e
ra
s,
flu
c
tu
a
ti
n
g
b
a
c
k
g
ro
u
n
d
s,
a
n
d
e
x
p
o
su
re
to
th
e
e
n
v
iro
n
m
e
n
t
in
wh
ich
a
n
i
n
d
i
v
id
u
a
l
is
p
re
se
n
t.
T
h
e
rm
a
l
i
m
a
g
in
g
is
e
n
d
e
a
v
o
ri
n
g
to
re
so
lv
e
t
h
e
a
c
c
u
ra
c
y
issu
e
o
f
a
p
p
a
re
n
t
ima
g
in
g
,
su
c
h
a
s
li
g
h
t
in
g
a
n
d
b
ri
g
h
t
n
e
ss
in
ten
sit
y
,
a
m
o
n
g
a
ll
b
io
m
e
tri
c
v
a
riab
les
.
Th
is
p
a
p
e
r
p
re
se
n
ts
a
st
u
d
y
o
f
th
e
rm
a
l
ima
g
in
g
a
n
d
e
ffe
c
ti
v
e
m
e
th
o
d
s
i
n
v
o
lv
e
d
i
n
t
h
e
fe
a
tu
re
e
x
trac
ti
o
n
p
r
o
c
e
s
s
fo
r
fa
c
i
a
l
fe
a
tu
re
s
with
th
e
rm
a
l
ima
g
in
g
u
n
d
e
r
t
h
e
in
flu
e
n
c
e
o
f
v
a
ried
n
o
ise
.
A
n
o
v
e
l
fa
c
e
d
a
tas
e
t
is
c
re
a
ted
TID
c
o
m
p
risin
g
2
7
th
e
rm
a
l
ima
g
e
s
a
n
d
it
s
c
o
rre
sp
o
n
d
in
g
v
isu
a
l
b
a
n
d
ima
g
e
u
si
n
g
F
l
u
k
e
4
8
0
Ti
P
ro
c
a
m
e
ra
.
Th
e
stu
d
y
a
n
a
ly
se
s
d
e
tec
ti
o
n
e
fficie
n
c
y
o
f
six
fe
a
tu
re
e
x
trac
ti
o
n
tec
h
n
iq
u
e
s
i
n
v
isib
le
a
n
d
th
e
rm
a
l
b
a
n
d
s
i
n
fa
c
ial
fe
a
tu
re
s
i
d
e
n
ti
fica
ti
o
n
.
Also
,
t
h
e
in
fl
u
e
n
c
e
o
f
n
o
ise
in
th
e
th
e
rm
a
l
b
a
n
d
with
in
th
e
re
g
i
o
n
o
f
in
tere
st
u
sin
g
fe
a
tu
re
p
o
i
n
ts
F
IN
,
F
OU
T
h
a
s
b
e
e
n
e
stim
a
ted
.
Th
ro
u
g
h
o
u
t
TI
D
d
a
tas
e
t,
O
RB
e
x
trac
ti
o
n
tec
h
n
i
q
u
e
h
a
s
b
e
e
n
a
b
le
t
o
i
d
e
n
ti
f
y
str
o
n
g
e
st
in
l
ier
fe
a
tu
re
s
F
IN
t
o
a
m
a
x
imu
m
e
x
te
n
t
wit
h
d
e
tec
ti
o
n
a
ro
u
n
d
t
h
e
n
o
se
,
e
y
e
s,
a
n
d
m
o
u
t
h
.
F
u
rth
e
r,
r
e
su
lt
s
i
n
d
ica
te
fe
a
tu
re
d
e
tec
ti
o
n
i
n
th
e
rm
a
l
ima
g
e
s
b
e
in
g
in
v
a
rian
t
to
e
ffe
c
t
o
f
n
o
ise
f
o
r
d
e
tec
t
i
n
g
fa
c
ial
fe
a
tu
re
s.
K
ey
w
o
r
d
s
:
Featu
r
e
ex
tr
ac
tio
n
Gau
s
s
ian
n
o
is
e
Salt a
n
d
p
ep
p
e
r
n
o
is
e
T
h
er
m
al
im
ag
in
g
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Palak
Ma
h
ajan
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
an
d
E
n
g
in
ee
r
in
g
,
C
en
tr
al
Un
iv
er
s
ity
o
f
J
am
m
u
J
am
m
u
an
d
Kash
m
ir
(
J
&
K)
,
I
n
d
ia
E
m
ail:
p
alak
.
cse@
cu
jam
m
u
.
a
c.
in
1.
I
NT
RO
D
UCT
I
O
N
An
em
er
g
en
t
ar
ea
o
f
co
m
p
u
te
r
v
is
io
n
is
f
ac
e
r
ec
o
g
n
itio
n
,
m
o
r
e
ex
p
licitly
f
ac
e
d
etec
tio
n
in
th
er
m
al
im
ag
es.
T
h
e
im
ag
e
ac
q
u
is
itio
n
p
r
o
ce
s
s
co
m
p
r
is
es
ca
p
tu
r
in
g
im
ag
es
f
r
o
m
a
ca
m
er
a.
T
h
ese
im
ag
es
ca
n
b
e
ca
p
tu
r
ed
in
v
ar
io
u
s
wav
elen
g
t
h
s
u
s
in
g
d
if
f
e
r
en
t
ca
m
er
as.
T
h
e
im
ag
es
ca
p
tu
r
e
d
in
th
e
v
is
ib
le
b
an
d
k
n
o
wn
as
v
is
u
al
im
ag
es
ca
n
b
e
ca
p
tu
r
ed
b
y
o
p
tical
ca
m
er
as,
wh
ile
th
eir
th
er
m
al
co
m
p
o
n
en
t
ca
n
f
u
r
t
h
er
b
e
ca
p
tu
r
e
d
b
y
in
f
r
ar
ed
ca
m
e
r
as.
Sp
ec
ial
in
f
r
ar
ed
v
is
io
n
d
ev
ices
ar
e
u
s
ed
to
ca
p
tu
r
e
th
e
h
ea
t
em
itted
b
y
th
ese
o
b
jects
to
p
r
o
d
u
ce
a
t
h
er
m
al
im
a
g
e.
T
h
er
m
al
im
ag
in
g
ca
n
b
e
d
ef
in
e
d
as
th
e
ac
q
u
is
itio
n
o
f
elec
tr
o
m
ag
n
etic
e
n
er
g
y
g
en
er
ated
o
r
r
ef
lecte
d
b
y
an
o
b
ject
in
th
e
in
f
r
a
r
ed
b
an
d
o
f
t
h
e
elec
tr
o
m
a
g
n
etic
s
p
ec
tr
u
m
[
1
]
.
E
x
tr
ac
tin
g
an
d
d
etec
tin
g
f
ac
ial
ch
ar
ac
ter
is
tics
in
v
is
ib
le
b
an
d
im
ag
er
y
is
th
e
m
o
s
t
ef
f
ec
tiv
e
m
eth
o
d
f
o
r
f
ac
e
r
ec
o
g
n
itio
n
.
Ho
wev
er
,
f
ac
e
r
ec
o
g
n
itio
n
in
th
e
th
er
m
al
s
p
ec
tr
u
m
p
r
o
v
id
e
s
s
ig
n
if
ican
t
ad
v
an
tag
es
o
v
e
r
f
ac
e
r
ec
o
g
n
itio
n
in
th
e
v
is
ib
le
s
p
ec
tr
u
m
.
R
ec
en
t
d
ev
elo
p
m
en
ts
in
tech
n
o
l
o
g
y
,
th
e
av
ailab
ilit
y
o
f
af
f
o
r
d
ab
le
s
e
n
s
o
r
s
,
an
d
a
v
ar
iety
o
f
u
s
es
f
o
r
s
u
ch
im
a
g
er
y
,
h
av
e
in
cr
ea
s
ed
in
ter
est
in
th
er
m
al
im
ag
in
g
s
u
ch
as
in
m
o
n
ito
r
in
g
an
d
s
u
r
v
eillan
ce
s
y
s
tem
s
.
Mo
s
t
o
f
th
e
wo
r
k
d
o
n
e
in
im
ag
e
p
r
o
ce
s
s
in
g
h
as
b
e
en
d
o
n
e
in
th
e
v
is
ib
le
s
p
ec
tr
u
m
o
f
th
e
im
ag
e
[
2
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
13
,
No
.
3
,
Dec
em
b
er
20
24
:
333
-
3
4
3
334
Ho
wev
er
,
th
e
v
is
ib
le
s
p
ec
tr
u
m
m
ay
n
o
t
p
er
f
o
r
m
well
in
p
o
o
r
illu
m
in
atio
n
s
itu
atio
n
s
o
r
en
v
ir
o
n
m
en
tal
co
n
d
itio
n
s
lik
e
f
o
g
an
d
h
az
e
[
3
]
.
Fo
r
f
ac
e
r
ec
o
g
n
itio
n
ap
p
lic
atio
n
s
,
lo
w
illu
m
in
atio
n
h
i
n
d
e
r
s
th
e
p
e
r
f
o
r
m
an
ce
o
f
th
e
v
is
i
b
le
s
p
ec
tr
u
m
,
th
er
e
b
y
p
r
ev
e
n
tin
g
it
f
r
o
m
f
u
n
ctio
n
i
n
g
as
in
ten
d
e
d
.
T
h
e
r
m
al
im
ag
in
g
wh
er
ea
s
is
less
s
u
s
ce
p
tib
le
to
illu
m
in
atio
n
ch
an
g
es
h
en
ce
,
th
e
v
ar
iatio
n
in
f
ac
ial
f
ea
tu
r
es
id
en
tific
atio
n
d
u
e
to
ch
a
n
g
es
in
lig
h
tin
g
is
less
n
o
ticea
b
le
in
t
h
er
m
al
im
ag
es.
I
n
f
r
a
r
ed
r
ad
ia
tio
n
is
em
itted
b
y
ev
er
y
o
b
ject.
T
h
e
r
a
d
iatio
n
is
s
en
s
ed
as
h
ea
t
in
th
e
f
ar
in
f
r
ar
ed
r
eg
io
n
.
L
iv
i
n
g
cr
ea
tu
r
es
lik
e
h
u
m
an
s
a
n
d
an
im
als
in
g
en
er
al
em
it
m
o
r
e
in
f
r
ar
ed
r
a
d
iatio
n
th
a
n
th
eir
s
u
r
r
o
u
n
d
in
g
s
.
T
h
e
th
er
m
al
i
n
f
r
ar
ed
(
IR
)
s
en
s
o
r
m
o
n
ito
r
s
th
e
h
ea
t
en
er
g
y
e
m
itted
b
y
th
e
f
ac
e
in
s
tead
o
f
d
etec
tin
g
r
ef
lecte
d
lig
h
t.
T
h
e
h
ea
t
p
at
ter
n
s
r
elea
s
ed
b
y
an
o
b
ject
s
er
v
e
as
th
e
b
asis
f
o
r
th
er
m
al
im
ag
i
n
g
.
Dep
en
d
i
n
g
o
n
th
ei
r
h
ea
t
an
d
co
m
p
o
s
itio
n
,
v
ar
i
o
u
s
q
u
an
titi
es
o
f
I
R
r
ad
i
atio
n
ar
e
em
itted
b
y
o
b
jects.
T
h
er
m
al
ca
m
er
as
c
an
ca
p
tu
r
e
im
ag
es
in
c
o
m
p
lete
d
ar
k
n
ess
an
d
a
r
e
r
esis
tan
t
to
c
h
an
g
es
i
n
illu
m
in
atio
n
,
m
a
k
in
g
t
h
em
s
u
itab
le
f
o
r
r
esear
ch
ac
tiv
ities
s
u
ch
as
n
ig
h
t
v
is
io
n
s
u
r
v
eillan
c
e
s
y
s
tem
s
[
4
]
an
d
b
io
m
etr
ics
[
5
]
.
R
esear
ch
er
s
h
av
e
ex
p
lo
r
ed
t
h
er
m
al
im
ag
in
g
f
o
r
n
u
m
er
o
u
s
ap
p
licatio
n
s
[
6
]
.
A
YOL
Ov
5
b
ased
f
ac
e
d
etec
tio
n
f
r
a
m
ewo
r
k
h
as
b
ee
n
p
r
o
p
o
s
ed
b
y
u
s
in
g
th
e
r
m
al
f
ac
es
in
th
e
wild
(
T
FW
)
th
e
r
m
al
d
ataset
u
n
d
er
co
n
tr
o
lled
an
d
u
n
co
n
t
r
o
lled
en
v
ir
o
n
m
e
n
tal
f
ac
to
r
s
.
T
h
e
m
o
d
el
ap
p
lied
Sh
u
f
f
leNe
tv
2
as
its
b
ac
k
b
o
n
e
co
n
v
o
l
u
tio
n
al
n
eu
r
al
n
etwo
r
k
s
(
C
NN
)
with
p
r
e
-
tr
ain
ed
weig
h
ts
.
T
h
e
m
o
d
el
s
h
o
wca
s
ed
a
tr
ad
eo
f
f
b
etwe
en
s
p
ee
d
an
d
ac
cu
r
ac
y
wh
ile
c
o
n
s
id
er
i
n
g
v
iv
id
im
ag
e
r
eso
lu
tio
n
s
[
7
]
.
Ma
r
tin
ez
et
a
l.
[
8
]
,
a
m
eth
o
d
f
o
r
au
to
m
atica
lly
id
en
tify
in
g
ey
es
an
d
n
o
s
tr
ils
in
f
ac
e
th
er
m
al
im
ag
es
u
s
in
g
Haa
r
an
d
t
h
e
Gen
tle
B
o
o
s
t
alg
o
r
ith
m
h
as
b
ee
n
d
em
o
n
s
tr
ated
.
T
h
e
th
er
m
al
f
ac
ial
c
h
ar
ac
ter
is
tics
wer
e
lo
ca
ted
u
s
in
g
th
e
Har
r
is
in
t
er
est
p
o
in
t
d
etec
to
r
[
9
]
.
T
h
e
Har
r
is
in
ter
est
p
o
in
t
alg
o
r
ith
m
r
elies
o
n
in
ten
s
ity
.
I
t
u
s
es
g
r
ey
v
alu
es
to
id
en
tify
th
e
ex
is
ten
ce
o
f
an
in
ter
esti
n
g
p
o
in
t.
E
v
en
w
h
en
t
h
e
lig
h
tin
g
,
r
o
tatio
n
,
an
d
s
ca
lin
g
wer
e
ch
a
n
g
ed
,
th
e
e
x
p
er
im
en
t
s
till
r
esu
lted
in
a
lar
g
e
n
u
m
b
er
o
f
i
n
ter
esti
n
g
p
o
in
ts
.
B
ased
o
n
v
is
ib
le
an
d
t
h
er
m
al
in
f
r
ar
ed
d
ata,
th
e
r
esear
ch
er
[
10
]
p
r
esen
ted
a
d
etailed
p
e
r
f
o
r
m
an
ce
e
v
alu
a
tio
n
o
f
d
if
f
e
r
en
t
ap
p
ea
r
an
ce
-
b
ased
f
ac
e
r
ec
o
g
n
itio
n
alg
o
r
ith
m
s
.
T
h
ey
c
o
m
p
ar
e
d
alg
o
r
ith
m
s
b
o
t
h
with
in
th
e
s
am
e
im
ag
in
g
m
o
d
ality
as
well
as
ac
r
o
s
s
th
em
.
T
h
e
f
in
d
in
g
s
s
h
o
w
th
at
em
p
lo
y
in
g
th
er
m
al
in
f
r
ar
ed
im
ag
e
r
y
p
r
o
d
u
ce
s
b
etter
r
esu
lts
in
m
an
y
ca
s
es,
b
u
t
p
er
f
o
r
m
an
ce
in
b
o
th
m
o
d
alities
is
co
m
p
ar
ab
le
in
o
t
h
er
s
.
W
h
en
v
is
ib
le
an
d
th
er
m
al
in
f
r
a
r
ed
p
ictu
r
es
ar
e
m
er
g
ed
,
p
er
f
o
r
m
an
ce
im
p
r
o
v
es
ev
en
f
u
r
th
er
.
I
t
also
o
f
f
er
ed
a
p
a
r
tial
ex
p
lan
atio
n
f
o
r
th
e
n
u
m
er
o
u
s
co
n
tr
ad
icto
r
y
s
tatem
en
ts
ab
o
u
t
th
e
ef
f
ec
tiv
e
n
ess
o
f
v
ar
io
u
s
al
g
o
r
ith
m
s
o
n
ac
c
ess
ib
le
d
ata
s
ets.
Gu
zm
an
et
a
l.
[
11
]
,
ap
p
ly
i
n
g
Gab
o
r
f
il
ter
s
o
n
im
ag
es
o
f
d
if
f
er
en
t
s
izes
an
d
o
r
ien
tatio
n
s
ex
p
an
d
ed
f
ac
e
au
th
en
ticatio
n
.
B
y
d
iv
id
in
g
th
e
im
ag
es
in
to
tin
y
m
o
d
u
les
th
at
lo
ca
lize
f
ac
ial
ch
an
g
es,
th
e
v
a
r
iatio
n
s
in
th
e
f
ac
es
h
av
e
b
ee
n
d
is
co
v
er
e
d
.
T
h
e
b
e
h
av
io
r
o
f
tr
ad
itio
n
al
f
ea
t
u
r
e
p
o
in
t
d
escr
ip
to
r
s
in
im
ag
es
f
r
o
m
th
e
lo
n
g
-
wav
e
in
f
r
a
r
ed
s
p
ec
tr
al
r
eg
io
n
is
ev
alu
ated
in
[1
2
]
,
an
d
th
e
r
esu
lts
ar
e
co
m
p
ar
ed
to
th
o
s
e
o
b
tain
ed
in
th
e
v
is
ib
le
s
p
ec
tr
u
m
.
A
s
tate
-
of
-
th
e
-
ar
t
m
eth
o
d
o
lo
g
y
is
u
s
ed
to
ass
ess
r
o
b
u
s
tn
ess
to
ch
an
g
es
in
s
ca
lin
g
,
n
o
is
e,
b
lu
r
,
an
d
r
o
tatio
n
.
I
n
an
o
th
er
wo
r
k
[1
3
]
,
a
Har
r
is
o
p
er
at
or
-
b
ased
in
ter
est
p
o
in
t
d
etec
to
r
f
o
r
3
D
o
b
jects
h
av
e
b
ee
n
in
tr
o
d
u
ce
d
,
wh
ich
h
as
b
ee
n
s
u
cc
ess
f
u
lly
u
s
ed
in
co
m
p
u
ter
v
is
io
n
ap
p
licatio
n
s
.
An
a
d
ap
t
iv
e
tech
n
iq
u
e
f
o
r
d
eter
m
in
i
n
g
a
v
er
tex
’
s
n
eig
h
b
o
r
h
o
o
d
h
as
b
ee
n
p
r
esen
ted
,
wh
ich
is
th
e
n
u
s
ed
to
ca
lcu
late
th
e
Ha
r
r
is
r
esp
o
n
s
e
f
o
r
t
h
at
v
er
tex
.
T
h
e
tec
h
n
iq
u
e
was
s
h
o
wn
to
b
e
r
esis
tan
t
to
a
v
ar
iety
o
f
c
h
an
g
es,
as
is
e
v
id
en
t
f
r
o
m
t
h
e
h
ig
h
r
e
p
ea
ta
b
ilit
y
v
alu
es
o
b
tain
ed
with
t
h
e
SHR
E
C
f
ea
tu
r
e
d
etec
tio
n
an
d
d
escr
ip
tio
n
b
en
ch
m
ar
k
.
Fu
r
th
e
r
m
o
r
e,
Har
r
is
3
D
s
u
r
p
ass
es
co
n
tem
p
o
r
ar
y
e
f
f
ec
tiv
e
tech
n
iq
u
es
s
u
ch
as h
ea
t k
er
n
el
s
ig
n
atu
r
es,
ac
co
r
d
in
g
to
r
esear
c
h
er
s
.
I
m
ag
in
g
f
ea
tu
r
es
p
r
o
d
u
ce
d
b
y
a
th
er
m
al
s
en
s
o
r
i
n
d
icate
th
e
th
er
m
al
p
r
o
p
er
ties
o
f
th
e
f
ac
i
al
p
atter
n
.
E
x
tr
ac
tin
g
co
n
to
u
r
s
in
th
er
m
a
l
im
ag
es
is
an
is
s
u
e
th
at
em
er
g
es
d
u
e
to
th
e
h
eter
o
g
en
e
o
u
s
v
iew
o
f
th
e
ab
s
en
ce
o
f
d
ef
in
ed
b
o
u
n
d
ar
ies
in
im
ag
es.
I
n
ad
d
itio
n
,
th
e
m
ajo
r
i
ty
o
f
th
e
im
ag
es
co
n
tain
les
s
s
ig
n
if
ican
t
im
ag
e
f
ea
tu
r
es,
p
ar
ticu
la
r
ly
ar
o
u
n
d
th
e
ey
es
a
n
d
m
o
u
t
h
ar
ea
.
I
n
th
is
p
ap
e
r
,
th
e
an
aly
s
is
f
o
r
in
v
esti
g
atin
g
th
e
id
en
tific
at
io
n
ca
p
ab
ilit
y
o
f
v
a
r
io
u
s
f
ea
tu
r
e
d
etec
tio
n
tech
n
i
q
u
es
h
as
b
ee
n
ca
r
r
ied
o
u
t
to
d
etec
t
f
ac
ial
f
ea
tu
r
e
p
o
in
ts
s
u
ch
as
ey
es,
n
o
s
e,
an
d
m
o
u
t
h
in
v
ar
ied
n
o
is
e
co
n
d
itio
n
s
.
T
h
e
in
f
lu
e
n
ce
o
f
t
wo
ty
p
es
o
f
n
o
is
e:
Gau
s
s
ian
n
o
is
e
[1
4
]
an
d
s
alt
an
d
p
ep
p
er
n
o
is
e
[
15
]
ha
v
e
b
ee
n
an
aly
ze
d
in
th
e
r
m
al
im
ag
es.
Ad
d
in
g
Gau
s
s
ian
n
o
is
e
in
th
e
im
ag
es
ch
an
g
es
th
e
v
alu
e
o
f
th
e
p
ix
el
b
y
av
er
ag
in
g
t
h
e
v
alu
es
o
f
its
s
u
r
r
o
u
n
d
in
g
p
ix
els.
T
h
e
Gau
s
s
ian
n
o
is
e
is
r
ep
r
es
en
ted
as
th
e
(
1
)
f
o
r
r
ep
r
esen
tin
g
th
e
n
o
is
e
f
ac
to
r
an
d
e
d
en
o
tin
g
th
e
E
u
le
r
n
u
m
b
er
with
(
,
)
s
y
m
b
o
lizin
g
h
o
r
izo
n
tal
an
d
v
er
tical
d
is
tan
ce
c
o
n
ce
r
n
in
g
th
e
ce
n
ter
p
ix
el
[1
6
]
.
(
,
)
=
1
2
2
−
(
2
+
2
)
2
2
⁄
(
1
)
T
h
e
s
alt
an
d
p
e
p
p
er
n
o
is
e
(
L
u
o
et
a
l.
[1
7
]
)
ca
n
b
e
r
ep
r
esen
te
d
b
y
(
2
)
g
iv
en
as:
(
)
=
{
,
=
,
=
0
,
ℎ
(
2
)
wh
er
e
th
e
in
ten
s
it
o
f
a
n
o
is
y
im
ag
e
’
s
p
ix
els
is
r
ep
r
esen
t
ed
b
y
.
Ad
d
itio
n
ally
,
an
d
s
tan
d
f
o
r
n
o
is
e
im
p
u
ls
es f
o
r
d
ar
k
an
d
lig
h
t p
o
in
ts
; w
h
en
>
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
Th
erma
l ima
g
in
g
-
b
a
s
ed
id
en
ti
fica
tio
n
o
f fa
cia
l fe
a
t
u
r
es in
n
o
is
y
en
viro
n
men
t
(
P
a
la
k
Ma
h
a
ja
n
)
335
I
n
th
is
wo
r
k
,
th
e
th
e
r
m
al
im
a
g
es
in
th
e
in
f
r
ar
ed
b
an
d
h
av
e
b
ee
n
ex
p
lo
r
e
d
f
o
r
th
e
f
ac
e
r
ec
o
g
n
itio
n
p
r
o
ce
s
s
.
A
n
ew
f
ac
e
d
ataset
h
as
b
ee
n
cr
ea
ted
,
co
m
p
r
is
in
g
2
7
th
er
m
al
im
ag
es
alo
n
g
with
th
eir
co
r
r
esp
o
n
d
in
g
v
is
u
al
b
an
d
im
ag
es.
Fo
r
th
e
id
en
tific
atio
n
,
s
ix
f
ea
tu
r
e
e
x
t
ra
ctio
n
tech
n
iq
u
es
:
f
ea
tu
r
es
f
r
o
m
ac
ce
ler
ated
s
eg
m
en
t
test
(
FAST
)
[1
8
]
,
h
is
to
g
r
am
o
f
o
r
ie
n
ted
g
r
a
d
ien
ts
(
HOG
)
[
19
]
,
o
r
ie
n
ted
FAST
an
d
r
o
tated
B
R
I
E
F
(
OR
B
)
[2
0
]
,
m
ax
im
ally
s
tab
le
ex
tr
e
m
al
r
eg
io
n
s
(
MSE
R
)
[
2
1]
,
Har
r
is
[
22
],
an
d
E
ig
en
[2
3
]
h
av
e
b
ee
n
ap
p
lied
.
T
ab
le
1
r
ep
r
esen
ts
a
b
r
ief
d
escr
ip
tio
n
o
f
th
ese
f
ea
tu
r
e
ex
tr
ac
tio
n
tech
n
iq
u
es
f
o
r
co
m
p
u
ter
v
is
io
n
task
s
.
A
co
m
p
ar
is
o
n
h
as
b
ee
n
d
o
n
e
to
an
aly
ze
th
e
p
er
f
o
r
m
an
ce
o
f
t
h
e
f
ea
tu
r
e
d
etec
tio
n
tech
n
i
q
u
e
in
th
e
th
er
m
al
a
n
d
v
is
ib
le
b
an
d
s
.
Als
o
,
th
eir
e
f
f
ic
ien
cy
h
as
f
u
r
th
e
r
b
ee
n
s
u
b
s
tan
tiated
in
d
etec
tin
g
f
ac
ial
f
ea
tu
r
es
in
th
e
p
r
esen
ce
o
f
ex
ter
n
al
n
o
is
e.
T
h
e
k
ey
c
o
n
tr
ib
u
tio
n
s
o
f
t
h
is
m
an
u
s
cr
ip
t a
r
e
as f
o
llo
ws:
−
A
n
o
v
el
f
ac
e
d
ataset
T
I
D
is
cr
ea
ted
co
m
p
r
is
in
g
o
f
th
er
m
a
l
im
ag
es
an
d
th
eir
c
o
r
r
esp
o
n
d
in
g
v
is
u
al
b
an
d
im
ag
es u
n
d
er
d
if
f
er
e
n
t e
n
v
i
r
o
n
m
en
tal
f
ac
to
r
s
.
−
T
h
e
s
tu
d
y
a
n
aly
s
es th
e
d
etec
tio
n
ef
f
icien
c
y
o
f
f
ea
tu
r
e
e
x
tr
ac
tio
n
tech
n
iq
u
es in
v
is
ib
le
an
d
t
h
er
m
al
b
an
d
s
.
−
T
h
e
ass
ess
m
en
t
an
d
an
aly
s
is
o
f
n
o
is
e
im
p
ac
t:
g
au
s
s
ian
an
d
s
alt
an
d
p
ep
p
er
n
o
is
e
in
th
e
th
er
m
al
b
an
d
with
in
th
e
ar
ea
o
f
in
ter
est h
av
e
b
ee
n
co
n
d
u
cte
d
u
s
in
g
f
ea
tu
r
e
p
o
in
ts
.
−
Mo
r
eo
v
er
,
u
s
in
g
th
e
d
im
e
n
s
io
n
ality
r
ed
u
ctio
n
tech
n
iq
u
e
p
r
i
n
cip
al
co
m
p
o
n
e
n
ts
an
aly
s
is
(
PC
A
)
,
th
e
f
ea
tu
r
e
s
p
ac
e
is
o
p
tim
ized
b
y
c
o
n
s
id
er
in
g
th
e
r
elev
a
n
t f
ea
tu
r
es.
T
h
e
o
u
tlin
e
o
f
th
is
wo
r
k
is
s
tr
u
ctu
r
ed
as
f
o
llo
ws.
I
n
s
ec
tio
n
2
,
th
er
m
al
d
ataset
co
n
s
tr
u
ctio
n
h
as
b
ee
n
ex
p
lain
ed
an
d
th
e
o
v
er
v
iew
o
f
th
e
p
r
o
p
o
s
ed
f
r
am
ewo
r
k
i
s
d
is
cu
s
s
ed
.
T
h
e
an
aly
s
is
an
d
o
u
tco
m
es
o
f
th
e
ex
p
er
im
en
ts
ar
e
d
is
cu
s
s
ed
in
s
ec
tio
n
3
.
L
astl
y
,
th
e
co
n
clu
s
io
n
is
p
r
esen
ted
in
s
ec
tio
n
4
.
T
ab
le
1
.
Featu
r
e
e
x
tr
ac
tio
n
tec
h
n
iq
u
es
2.
RE
S
E
ARCH
M
E
T
H
O
D
T
h
e
f
ac
e
d
ataset
h
as
b
ee
n
cr
e
ated
at
Dep
a
r
tm
en
t
o
f
C
o
m
p
u
ter
Scien
ce
an
d
I
T
an
d
Dep
ar
tm
en
t
o
f
E
lectr
o
n
ics,
Un
iv
er
s
ity
o
f
J
am
m
u
,
J
&
K,
I
n
d
ia.
T
h
er
m
al
im
ag
es
h
av
e
b
ee
n
ca
p
tu
r
ed
u
s
in
g
a
Flu
k
e
4
8
0
T
i
Pro
I
n
f
r
ar
e
d
C
am
er
a
[
2
4
]
,
alo
n
g
with
th
e
co
r
r
esp
o
n
d
in
g
v
is
u
al
im
ag
es.
Un
d
er
v
a
r
ied
co
n
s
tr
ain
ts
,
im
ag
es
h
av
e
b
ee
n
co
llected
in
p
air
s
,
o
n
e
t
h
er
m
al
an
d
o
n
e
v
is
ib
le.
Fig
u
r
e
1
s
h
o
ws
th
e
ex
p
er
im
en
tal
c
o
n
f
ig
u
r
atio
n
o
f
t
h
e
im
ag
e
ca
p
tu
r
e
p
r
o
ce
s
s
.
As
illu
s
tr
ated
in
Fig
u
r
e
1
(
a)
,
t
h
e
ca
m
er
a
s
ettin
g
s
h
av
e
b
ee
n
ad
j
u
s
ted
f
o
r
ca
p
tu
r
in
g
th
e
im
ag
es
.
T
h
e
ca
m
er
a
h
as
b
ee
n
p
lace
d
ar
o
u
n
d
6
0
cm
awa
y
f
r
o
m
th
e
p
ar
ticip
an
ts
,
a
n
d
with
th
e
f
ac
e
at
th
e
ce
n
ter
o
f
th
e
im
ag
e.
T
h
e
im
ag
es
h
a
v
e
b
ee
n
ca
p
tu
r
ed
with
s
u
b
jects si
ttin
g
o
n
ch
air
s
at
a
d
is
tan
ce
o
f
ab
o
u
t
2
f
ee
t
f
r
o
m
th
e
ca
m
er
as
[
2
5
]
.
T
h
e
r
ain
b
o
w
m
o
d
e
is
em
p
l
o
y
ed
f
o
r
t
h
er
m
al
im
ag
in
g
with
t
h
e
im
ag
e
o
u
tp
u
t
co
n
tain
in
g
b
o
th
th
er
m
al
an
d
R
GB
im
ag
es.
T
h
e
im
ag
es g
en
er
ated
b
y
t
h
e
ca
m
er
a
h
av
e
b
ee
n
s
av
ed
in
J
PEG
f
o
r
m
at
.
T
h
e
Flu
k
e
4
8
0
s
tan
d
s
as
a
p
o
r
tab
le
in
f
r
ar
ed
th
er
m
al
ca
m
er
a,
th
e
s
p
ec
if
icatio
n
s
o
f
th
e
ca
m
er
a
u
s
e
d
ar
e
g
iv
e
n
in
T
ab
le
2
.
T
h
e
im
ag
es
h
av
e
b
ee
n
ca
p
tu
r
e
d
at
a
co
n
s
tan
t
tem
p
er
at
u
r
e
o
f
ar
o
u
n
d
2
7
.
5
°C
.
T
h
e
ca
p
t
u
r
in
g
p
r
o
ce
s
s
h
as
b
e
en
ca
r
r
ie
d
o
u
t
f
o
r
two
c
o
n
s
ec
u
tiv
e
d
a
y
s
to
m
ain
tain
th
e
te
m
p
er
atu
r
e.
Dif
f
er
e
n
t
im
ag
e
m
o
d
e
o
p
tio
n
s
in
Flu
k
e
4
8
0
T
i
Pro
ca
n
b
e
em
p
lo
y
ed
.
Flu
k
e
4
8
0
T
i
Pro
im
ag
e
c
h
ar
a
cter
is
tics
h
av
e
b
ee
n
d
iv
id
ed
in
t
o
s
ev
en
m
o
d
es:
r
ain
b
o
w,
ir
o
n
b
o
w,
b
lu
e
-
r
ed
,
h
ig
h
co
n
tr
ast,
am
b
er
,
h
o
t
m
etal,
an
d
g
r
a
y
s
ca
le
.
T
h
e
co
lo
r
p
alette
p
r
o
ce
s
s
in
g
,
d
is
p
lay
m
ar
k
er
s
ad
j
u
s
tm
en
t,
em
is
s
iv
ity
s
ettin
g
s
,
an
d
v
a
r
i
o
u
s
p
r
e
-
p
r
o
ce
s
s
in
g
o
p
er
atio
n
s
o
n
th
e
im
ag
es c
ap
tu
r
ed
b
y
th
e
ca
m
e
r
a
h
av
e
b
ee
n
d
o
n
e
u
s
in
g
FLUK
E
Sm
ar
tView
v
er
s
io
n
4
.
3
.
3
1
1
.
0
s
o
f
twar
e
.
C
o
lo
r
p
alette
I
r
o
n
b
o
w
an
d
Gr
a
y
s
ca
le
h
av
e
b
ee
n
a
g
en
er
al
-
p
u
r
p
o
s
e
p
alette
th
at
in
s
tan
tly
d
etec
ts
th
er
m
al
an
o
m
alies
an
d
b
o
d
y
h
ea
t.
R
ain
b
o
w
h
as
t
h
e
o
p
tim
u
m
co
lo
r
p
alette
f
o
r
s
itu
atio
n
s
with
litt
le
h
ea
t
s
h
if
t
an
d
a
f
o
cu
s
o
n
a
n
a
r
ea
with
s
im
ilar
h
ea
t
i
n
ten
s
ity
.
I
n
a
b
r
o
ad
er
s
en
s
e,
t
h
e
f
ac
e
tem
p
er
at
u
r
e
d
o
es
n
o
t
c
h
an
g
e
Te
c
h
n
i
q
u
e
s
O
b
j
e
c
t
i
v
e
S
u
mm
a
r
y
S
t
r
e
n
g
t
h
Li
mi
t
a
t
i
o
n
F
A
S
T
F
e
a
t
u
r
e
mat
c
h
i
n
g
f
e
a
t
u
r
e
t
r
a
c
k
i
n
g
R
e
c
o
g
n
i
z
e
s c
o
r
n
e
r
s
b
y
c
o
n
t
r
a
s
t
i
n
g
p
i
x
e
l
b
r
i
g
h
t
n
e
ss
Ef
f
i
c
i
e
n
t
c
o
m
p
u
t
a
t
i
o
n
r
o
b
u
st
t
o
n
o
i
se
N
o
i
se
se
n
si
t
i
v
i
t
y
l
o
w
p
e
r
f
o
r
m
a
n
c
e
w
i
t
h
c
h
a
n
g
e
i
n
l
i
g
h
t
i
n
g
M
S
ER
R
e
g
i
o
n
-
b
a
s
e
d
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
i
ma
g
e
seg
m
e
n
t
a
t
i
o
n
Lo
c
a
t
e
s a
r
e
a
s w
i
t
h
c
o
n
si
st
e
n
t
i
n
t
e
n
si
t
y
f
l
u
c
t
u
a
t
i
o
n
s
a
c
r
o
ss
v
a
r
i
e
d
z
o
o
m
l
e
v
e
l
s.
R
o
b
u
st
t
o
g
e
o
me
t
r
i
c
t
r
a
n
sf
o
r
ma
t
i
o
n
s
S
e
n
s
i
t
i
v
e
t
o
r
e
g
i
o
n
s
i
z
e
H
a
r
r
i
s c
o
r
n
e
r
d
e
t
e
c
t
i
o
n
F
e
a
t
u
r
e
mat
c
h
i
n
g
f
e
a
t
u
r
e
t
r
a
c
k
i
n
g
A
p
p
l
i
e
s e
i
g
e
n
v
a
l
u
e
s
t
o
d
e
t
e
c
t
c
o
r
n
e
r
s
a
n
d
c
a
l
c
u
l
a
t
e
t
h
e
c
h
a
n
g
e
i
n
i
n
t
e
n
s
i
t
y
f
o
r
mi
n
o
r
w
i
n
d
o
w
sh
i
f
t
s.
R
o
b
u
st
t
o
l
i
g
h
t
i
n
g
c
h
a
n
g
e
Lo
w
p
e
r
f
o
r
m
a
n
c
e
w
i
t
h
c
h
a
n
g
e
s
i
n
r
o
t
a
t
i
o
n
a
n
d
sca
l
e
s
e
n
s
i
t
i
v
e
t
o
n
o
i
s
e
Ei
g
e
n
f
a
c
e
s
F
a
c
e
r
e
c
o
g
n
i
t
i
o
n
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
I
l
l
u
st
r
a
t
e
f
a
c
e
s
a
s
l
i
n
e
a
r
c
o
m
b
i
n
a
t
i
o
n
s
o
f
i
ma
g
e
s
o
b
t
a
i
n
e
d
u
si
n
g
P
C
A
O
p
t
i
mi
z
e
f
a
c
i
a
l
f
e
a
t
u
r
e
r
e
p
r
e
s
e
n
t
a
t
i
o
n
S
e
n
s
i
t
i
v
e
t
o
c
h
a
n
g
e
i
n
l
i
g
h
t
i
n
g
a
n
d
p
o
s
e
HOG
O
b
j
e
c
t
d
e
t
e
c
t
i
o
n
i
m
a
g
e
seg
m
e
n
t
a
t
i
o
n
Ex
t
r
a
c
t
s
l
o
c
a
l
i
z
e
d
sh
a
p
e
i
n
f
o
r
mat
i
o
n
b
y
a
p
p
l
y
i
n
g
g
r
a
d
i
e
n
t
o
r
i
e
n
t
a
t
i
o
n
s
C
a
p
t
u
r
e
s
e
d
g
e
a
n
d
sh
a
p
e
R
o
b
u
s
t
t
o
l
i
g
h
t
i
n
g
c
h
a
n
g
e
Lo
w
p
e
r
f
o
r
m
a
n
c
e
i
n
a
c
h
a
n
g
e
o
f
s
c
a
l
e
s
e
n
si
t
i
v
e
t
o
o
c
c
l
u
s
i
o
n
O
R
B
F
e
a
t
u
r
e
d
e
scri
p
t
i
o
n
f
e
a
t
u
r
e
d
e
t
e
c
t
i
o
n
f
e
a
t
u
r
e
m
a
t
c
h
i
n
g
I
n
t
e
g
r
a
t
e
s FA
S
T
k
e
y
p
o
i
n
t
d
e
t
e
c
t
o
r
w
i
t
h
t
h
e
B
R
I
EF d
e
s
c
r
i
p
t
o
r
t
o
i
n
c
o
r
p
o
r
a
t
e
r
o
t
a
t
i
o
n
a
l
i
n
v
a
r
i
a
n
c
e
Ef
f
i
c
i
e
n
t
c
o
m
p
u
t
a
t
i
o
n
r
o
b
u
st
t
o
l
i
g
h
t
i
n
g
c
h
a
n
g
e
Lo
w
p
e
r
f
o
r
m
a
n
c
e
i
n
t
h
e
c
h
a
n
g
e
o
f
s
c
a
l
e
s
e
n
si
t
i
v
e
t
o
n
o
i
s
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
13
,
No
.
3
,
Dec
em
b
er
20
24
:
333
-
3
4
3
336
d
r
asti
ca
lly
.
T
h
e
o
b
jectiv
e
is
to
o
b
s
er
v
e
tem
p
e
r
atu
r
e
f
l
u
ctu
atio
n
s
in
v
a
r
io
u
s
ar
ea
s
o
f
th
e
f
ac
e
an
d
th
en
e
m
p
lo
y
th
em
f
o
r
id
e
n
tific
atio
n
p
u
r
p
o
s
es.
As
a
r
esu
lt,
b
ec
au
s
e
th
e
te
m
p
er
atu
r
e
s
h
if
t
is
s
u
b
tle,
R
ain
b
o
w
m
o
d
e
h
as
b
ee
n
well
s
u
ited
f
o
r
f
ac
e
th
e
r
m
al
i
m
ag
in
g
in
th
is
s
tu
d
y
as sh
o
wn
in
Fig
u
r
e
1
(
b
)
.
A
s
am
p
le
o
f
th
e
th
e
r
m
al
im
a
g
es
ca
p
tu
r
ed
al
o
n
g
with
t
h
eir
r
esp
ec
tiv
e
v
is
ib
le
s
p
ec
tr
u
m
im
ag
es
is
s
h
o
wn
in
Fig
u
r
e
2
.
T
h
e
r
a
w
im
ag
es
o
b
tain
ed
o
f
s
ize
1
,
080
×
1
,
4
4
0
h
a
v
e
b
ee
n
r
esized
to
4
8
0
×
4
8
0
,
f
o
r
p
er
f
o
r
m
in
g
im
a
g
e
p
r
o
ce
s
s
in
g
o
p
er
atio
n
s
f
o
r
b
etter
p
r
esen
t
atio
n
.
T
h
e
ca
m
e
r
a
h
as
b
ee
n
ca
lib
r
ated
b
y
f
ir
s
t
ad
ju
s
tin
g
th
e
tem
p
er
atu
r
e
b
et
wee
n
1
5
°C
an
d
4
0
°C
,
f
o
ll
o
wed
b
y
s
u
b
s
eq
u
e
n
t
ad
ju
s
tin
g
o
f
th
e
m
u
ltip
o
in
t
f
o
cu
s
.
T
h
e
ca
m
er
a
g
en
er
ates
a
p
air
o
f
b
o
t
h
th
er
m
al
im
ag
es
a
n
d
v
is
ib
le
s
p
ec
tr
u
m
im
ag
es.
I
t
ex
am
in
es
th
e
r
aw
im
ag
e
a
n
d
r
em
o
v
es
s
en
s
o
r
n
o
is
e.
Pre
-
p
r
o
ce
s
s
in
g
tech
n
i
q
u
es
lik
e
p
allet
an
d
em
is
s
iv
ity
h
av
e
b
ee
n
th
en
em
p
lo
y
ed
o
n
p
r
o
ce
s
s
ed
th
er
m
al
im
ag
es
to
ac
cu
s
to
m
th
e
im
ag
e
ac
co
r
d
i
n
g
ly
.
T
h
e
em
is
s
iv
ity
p
ar
am
eter
w
h
ich
is
th
e
r
atio
o
f
an
o
b
ject’
s
ac
tu
al
em
ittan
ce
to
th
e
em
ittan
c
e
o
f
a
b
lack
b
o
d
y
at
th
e
s
am
e
tem
p
er
atu
r
e
h
as
b
ee
n
s
et
b
etwe
en
0
.
6
an
d
1
.
0
.
Fo
r
th
e
s
ak
e
o
f
u
n
if
o
r
m
ity
,
th
e
em
is
s
iv
ity
p
ar
am
eter
s
h
av
e
b
ee
n
s
et
to
0
.
8
.
E
m
is
s
iv
ity
is
m
ater
ial
-
d
ep
en
d
en
t
an
d
is
an
im
p
o
r
ta
n
t
p
r
o
p
e
r
ty
wh
en
m
ea
s
u
r
in
g
te
m
p
er
at
u
r
es
with
a
th
er
m
al
ca
m
er
a.
(
a)
(
b
)
Fig
u
r
e
1
.
C
o
n
f
ig
u
r
atio
n
s
ettin
g
f
o
r
d
ataset
ac
q
u
is
itio
n
(
a)
ill
u
s
tr
ated
(
b
)
tem
p
er
atu
r
e
f
lu
ctu
atio
n
s
T
ab
le
2
.
Sp
ec
if
icatio
n
tab
le
f
o
r
f
lu
k
e
4
8
0
T
i Pr
o
F
e
a
t
u
r
e
s
S
p
e
c
i
f
i
c
a
t
i
o
n
I
n
f
r
a
r
e
d
sp
e
c
t
r
a
l
b
a
n
d
7
.
5
μm
t
o
1
4
μm
Th
e
r
m
a
l
i
m
a
g
e
r
e
s
o
l
u
t
i
o
n
6
4
0
×
4
8
0
S
c
r
e
e
n
r
e
s
o
l
u
t
i
o
n
1
2
8
0
×
9
6
0
Th
e
r
m
a
l
se
n
si
t
i
v
i
t
y
5
0
m
K
Te
mp
e
r
a
t
u
r
e
r
a
n
g
e
-
20
°
C
t
o
+
8
0
0
°C
C
o
l
o
r
p
a
l
l
e
t
s
R
a
i
n
b
o
w
,
i
r
o
n
b
o
w
,
b
l
u
e
-
r
e
d
,
h
i
g
h
c
o
n
t
r
a
st
,
a
m
b
e
r
,
h
o
t
m
e
t
a
l
,
g
r
a
y
s
c
a
l
e
Fig
u
r
e
2
.
T
h
er
m
al
an
d
co
r
r
esp
o
n
d
in
g
v
is
u
al
im
ag
es f
r
o
m
th
e
T
I
D
d
ataset
T
h
e
im
ag
e
h
as
b
ee
n
tr
an
s
f
o
r
m
ed
to
g
r
ey
s
ca
le
s
o
th
at
f
ea
tu
r
e
ex
tr
ac
tio
n
tech
n
iq
u
es
co
u
ld
p
er
f
o
r
m
o
p
er
atio
n
s
.
Fu
r
t
h
er
,
th
e
t
h
e
r
m
al
im
ag
es
h
av
e
b
ee
n
e
n
h
an
ce
d
b
y
a
p
p
ly
in
g
h
is
to
g
r
am
eq
u
aliza
tio
n
.
Data
au
g
m
en
ta
tio
n
is
ad
d
itio
n
ally
d
o
n
e
b
y
ap
p
ly
in
g
r
o
tatio
n
an
d
s
ca
lin
g
to
in
cr
ea
s
e
th
e
s
ize
o
f
th
e
d
ataset.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
Th
erma
l ima
g
in
g
-
b
a
s
ed
id
en
ti
fica
tio
n
o
f fa
cia
l fe
a
t
u
r
es in
n
o
is
y
en
viro
n
men
t
(
P
a
la
k
Ma
h
a
ja
n
)
337
Af
ter
g
en
e
r
atin
g
a
g
r
e
y
-
lev
el
eq
u
alize
d
im
a
g
e,
e
x
ter
n
al
f
ac
to
r
s
n
o
is
e
h
as
b
ee
n
ap
p
lied
to
a
n
aly
ze
h
o
w
v
a
r
io
u
s
f
ea
tu
r
e
ex
tr
ac
tio
n
tech
n
iq
u
e
s
h
av
e
b
ee
n
in
f
lated
.
Fig
u
r
e
3
d
ep
icts
th
e
f
r
am
ewo
r
k
f
o
r
ass
ess
in
g
th
e
p
er
f
o
r
m
an
ce
o
f
f
ea
tu
r
e
d
escr
i
p
to
r
s
tech
n
i
q
u
es
in
th
e
in
f
r
ar
ed
b
a
n
d
f
o
r
th
e
r
m
al
im
a
g
es.
T
h
e
p
s
eu
d
o
co
d
e
f
o
r
th
e
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
y
is
d
is
cu
s
s
ed
in
Alg
o
r
ith
m
1
.
T
h
e
o
b
jectiv
e
is
to
d
is
co
v
er
th
e
s
en
s
itiv
ity
to
war
d
s
ex
ter
n
al
f
ac
to
r
s
i.e
.
,
n
o
is
e.
T
w
o
ty
p
es
o
f
n
o
is
e:
Gau
s
s
ian
an
d
s
alt
an
d
p
e
p
p
er
n
o
is
e
h
a
v
e
b
ee
n
ad
d
e
d
as
n
o
is
e
to
th
e
o
r
ig
in
al
th
e
r
m
al
im
ag
e.
T
h
e
p
ix
el
v
alu
e
h
as
b
ee
n
ad
j
u
s
ted
b
y
s
ettin
g
th
e
n
o
is
e
d
en
s
i
ty
v
alu
e
as
0
.
0
1
f
o
r
s
alt
an
d
p
ep
p
e
r
n
o
is
e.
Simil
ar
l
y
,
g
au
s
s
ian
n
o
is
e
h
as b
ee
n
ad
d
ed
with
m
ea
n
,
=
0
.
1
.
Fig
u
r
e
3
.
B
l
o
ck
d
iag
r
am
o
f
th
e
f
r
am
ewo
r
k
f
o
r
f
ea
tu
r
e
id
en
ti
f
icatio
n
in
th
er
m
al
im
ag
es
A
l
g
or
i
th
m
1
.
F
e
a
tu
r
e
i
d
e
nt
if
c
at
i
o
n
i
n
t
he
r
m
a
l
i
m
a
ge
s
Input
: Thermal
i
mage,
I
TM
Output
: Feature
p
oints:
Inliers,
F
IN
and Outliers,
F
OUT
Start
Step I: Calibrate Thermal camera: Radiometric calibration
Set a constant environment with controlled temperature= 27.5°C.
Warm up the camera for five minutes to attain a steady temperature.
Use a temperature
-
controlled blackbody reference source.
Keep a distance of 2 feet between the subject and the camera.
Capture the thermal image,
I
TM.
Step II: Preprocess the therm
al image:
for each
I
TM.
€
TID do
Set
color_pallet
= Rainbow and
emissivity
= 0.8
Resize the image = 480x480 resolutions.
Convert the image,
I
TM
into a grayscale
image,
I
GTM
Perform histogram equalization on
I
GTM
, to create enhanced image, I
TEQ
.
Apply augmentation on
I
GTM
using rotation= 30
°
and scaling=0.1
end for
Step III: Introduce External factor noise:
for each image
I
TEQ
€
TID do
Apply Salt and pepper noise with
noise_density
=0.01 to create
I
TSP
Additionally, apply Gaussian noise with mean,
μ
= 0.1 to create
I
TG
.
end for
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
13
,
No
.
3
,
Dec
em
b
er
20
24
:
333
-
3
4
3
338
Step IV: Extract Thermal Features:
for each image
I
TSP
,
I
TG
€
TID do
Apply the following feature extraction methods:
ORB, HOG, FAST, Harris, MSER, Eigen
Map the identified features for each ROI in every
I
TSP
, and
I
TG
.
Calculate the pertinent thermal features for each ROI.
Select relevant features around the face by computing feature points:
Inliers,
F
IN
and Outliers,
F
OUT
Further applying PCA to reduce the feature space.
end f
or
A
s
m
all
f
r
ac
tio
n
o
f
n
o
is
e
h
as
b
ee
n
ad
d
ed
to
th
e
th
e
r
m
al
im
ag
e
to
s
tu
d
y
th
e
s
u
b
tle
ef
f
e
ct
o
f
n
o
is
e
wh
ile
p
r
eser
v
in
g
t
h
e
th
er
m
al
p
r
o
p
er
ties
o
f
th
e
im
ag
e
.
T
h
e
n
o
is
e
v
alu
e
h
as
b
ee
n
s
elec
ted
s
u
ch
th
at
a
f
in
e
lev
el
o
f
d
is
to
r
tio
n
is
th
er
e
i
n
th
e
im
ag
es.
Af
ter
war
d
,
th
e
f
ea
tu
r
e
ex
tr
ac
tio
n
tech
n
iq
u
es
n
a
m
ely
,
OR
B
,
FA
ST,
MSE
R
,
Har
r
is
,
HOG,
an
d
E
i
g
en
f
u
n
ctio
n
s
wer
e
ap
p
lied
t
o
th
e
d
ataset.
T
h
e
f
ea
t
u
r
es
h
av
e
b
ee
n
ex
tr
ac
ted
th
o
r
o
u
g
h
ly
b
y
ea
ch
tech
n
iq
u
e.
T
h
e
r
esu
ltan
t
f
ea
tu
r
es
ar
e
co
m
p
ar
ed
an
d
th
e
p
er
f
o
r
m
an
ce
o
f
ea
ch
f
e
atu
r
e
ex
tr
ac
to
r
tech
n
i
q
u
e
is
ev
alu
at
ed
.
A
co
m
p
a
r
is
o
n
h
as
also
b
e
en
d
o
n
e
b
etwe
en
th
e
f
ea
t
u
r
e
p
o
in
ts
id
en
tifie
d
in
th
e
im
ag
es
’
in
f
r
a
r
ed
a
n
d
v
is
ib
le
s
p
ec
tr
al
r
e
g
io
n
s
.
T
h
e
s
en
s
itiv
ity
o
f
th
ese
f
ea
tu
r
e
ex
tr
ac
tio
n
tech
n
iq
u
es
to
war
d
v
ar
io
u
s
f
ac
to
r
s
h
as
b
ee
n
an
al
y
ze
d
as
well.
T
h
e
ev
alu
atio
n
is
b
ased
o
n
th
e
f
ac
ial
f
ea
tu
r
es
id
en
tifie
d
with
in
th
e
in
n
er
an
d
o
u
te
r
f
ac
ial
d
escr
ip
t
o
r
p
o
i
n
ts
.
=
∗
(3
)
W
h
er
e
r
e
p
r
esen
ts
th
e
r
ed
u
c
ed
r
ep
r
esen
tatio
n
o
f
t
h
e
ex
tr
ac
ted
f
ea
tu
r
es
f
o
r
T
I
D
d
ataset
f
o
r
im
ag
e
o
r
ig
in
al
m
atr
ix
.
T
h
e
to
p
E
ig
en
v
ec
to
r
s
ar
e
r
e
p
r
esen
ted
b
y
s
elec
ted
f
r
o
m
E
ig
e
n
m
atr
ix
cr
ea
ted
u
s
in
g
co
v
a
r
ian
ce
co
m
p
o
n
e
n
ts
.
Sin
ce
th
e
s
ize
o
f
d
ataset
is
s
m
a
ll,
a
m
in
im
al
v
alu
e
o
f
h
as
b
e
en
s
elec
ted
u
s
in
g
h
it
-
an
d
-
tr
ial
m
eth
o
d
,
s
p
ec
if
ic
ally
=
3
th
at
c
o
n
v
e
r
g
es
th
e
d
a
ta
to
lo
wer
-
d
im
en
s
io
n
al
s
p
a
ce
with
f
ast
co
m
p
u
tatio
n
al
s
p
ee
d
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
As
d
is
cu
s
s
ed
ab
o
v
e,
in
th
is
s
t
u
d
y
a
Flu
k
e
4
8
0
T
i
Pr
o
in
f
r
a
r
ed
ca
m
er
a
h
as
b
ee
n
u
s
ed
to
c
ap
tu
r
e
th
e
im
ag
es.
T
h
e
im
p
lem
e
n
tatio
n
o
f
th
e
alg
o
r
ith
m
h
as
b
ee
n
d
o
n
e
o
n
MA
T
L
AB
2
0
1
9
b
.
T
h
e
e
x
p
er
im
en
tal
wo
r
k
is
p
er
f
o
r
m
ed
o
n
I
n
tel(
R
)
C
o
r
e
(
T
M)
i7
-
8
7
5
0
H
C
PU
with
2
0
GB
R
AM
ac
co
m
p
an
ie
d
b
y
NVI
DI
A
GeFo
r
ce
GT
X
1
0
5
0
T
i.
Als
o
f
o
r
p
r
e
-
p
r
o
ce
s
s
in
g
th
e
th
er
m
al
im
ag
es,
th
e
s
tu
d
y
h
as
em
p
lo
y
e
d
S
m
ar
tView
s
o
f
twar
e
s
p
ec
if
ically
d
esig
n
ed
f
o
r
th
er
m
al
im
ag
es a
n
d
co
m
p
atib
le
wi
th
Flu
k
e
ca
m
er
a.
3
.
1
.
Co
m
pa
riso
n bet
wee
n v
is
ua
l a
nd
infr
a
re
d s
pect
rum
im
a
g
es
As
m
en
tio
n
ed
ab
o
v
e,
t
h
is
s
tu
d
y
in
v
esti
g
ates
th
e
p
er
f
o
r
m
a
n
c
e
o
f
f
ea
tu
r
e
id
en
tific
atio
n
tech
n
iq
u
es
f
o
r
d
etec
tin
g
f
ac
ial
im
ag
es
in
t
h
e
v
is
ib
le
an
d
in
f
r
ar
ed
s
p
ec
tr
u
m
th
at
r
ep
r
esen
ts
s
p
atial
an
d
f
r
eq
u
e
n
cy
d
o
m
ain
r
esp
ec
tiv
ely
.
A
co
m
p
a
r
is
o
n
h
as
b
ee
n
m
ad
e
b
ased
o
n
r
esu
lts
g
en
er
ated
b
y
ap
p
l
y
in
g
v
ar
io
u
s
f
ea
tu
r
e
id
en
tific
atio
n
tech
n
iq
u
es.
Fig
u
r
e
4
s
h
o
ws
th
e
v
a
r
io
u
s
f
ea
tu
r
es
ex
tr
ac
ted
b
y
th
e
s
ix
f
ea
tu
r
e
d
etec
tio
n
tech
n
iq
u
es
in
r
esp
ec
tiv
e
s
p
ec
tr
u
m
s
.
T
h
e
p
r
ep
r
o
ce
s
s
in
g
d
o
n
e
in
th
e
v
is
ib
le
s
p
ec
tr
u
m
is
o
n
ly
th
e
co
lo
r
co
n
v
er
s
io
n
o
f
R
GB
im
ag
es
in
to
g
r
e
y
-
lev
el
im
ag
es.
T
h
e
in
li
er
an
d
o
u
tlier
f
ea
t
u
r
es
o
f
th
e
f
ac
ial
im
ag
es
h
a
v
e
b
ee
n
o
b
s
er
v
ed
.
W
ith
in
lier
s
b
ein
g
th
e
o
n
e
th
at
lies
co
m
p
let
ely
in
s
id
e
a
r
eg
i
o
n
o
f
th
e
f
ac
e
an
d
o
u
tlier
s
b
ein
g
th
e
o
n
e
d
etec
ted
o
u
ts
id
e
th
e
f
ac
ial
r
eg
io
n
s
.
As
ca
n
b
e
o
b
s
er
v
ed
f
r
o
m
th
e
im
a
g
es,
f
o
r
v
is
ib
le
s
p
ec
tr
u
m
im
a
g
es
f
ac
ial
f
ea
tu
r
es
as
wel
l
as
s
o
m
e
o
th
er
f
ea
tu
r
es
s
u
ch
as
tex
t
wr
itten
o
r
tex
tu
r
e
p
r
esen
t
in
th
e
ap
p
ar
el
h
as
also
b
ee
n
d
etec
ted
.
Alm
o
s
t
all
f
ea
tu
r
e
ex
tr
ac
tio
n
tech
n
iq
u
es
ex
ce
p
t
OR
B
h
av
e
d
etec
ted
o
n
ly
o
u
tlier
s
in
v
is
ib
le
s
p
ec
tr
u
m
im
ag
es
as
s
h
o
wn
in
Fig
u
r
e
4
(
a)
.
On
th
e
o
th
er
h
a
n
d
,
in
th
e
ca
s
e
o
f
th
er
m
al
im
ag
es
in
th
e
in
f
r
ar
ed
s
p
ec
tr
u
m
as
illu
s
tr
ated
b
y
Fig
u
r
e
4
(
b
)
,
th
ese
o
u
tlier
f
ea
tu
r
es
h
av
e
co
m
p
letely
b
ee
n
m
as
k
ed
an
d
m
a
x
im
u
m
in
lier
f
ea
tu
r
es
s
u
ch
a
s
r
eg
io
n
s
n
ea
r
th
e
ey
es
an
d
ch
ee
k
s
h
av
e
b
ee
n
o
b
s
er
v
ed
.
W
ith
O
R
B
f
ea
tu
r
e
ex
tr
ac
to
r
id
en
tifie
s
th
e
m
a
x
im
u
m
f
ac
ial
f
ea
tu
r
es
ar
o
u
n
d
th
e
in
n
e
r
cir
c
le
o
f
t
h
e
f
ac
e
f
o
llo
we
d
b
y
th
e
H
ar
r
is
id
en
tify
i
n
g
th
e
in
lier
f
ea
tu
r
es.
Ov
er
all
f
o
r
alm
o
s
t
all
2
7
im
ag
es
in
th
e
T
I
D
d
ataset,
HOG
id
en
tifie
d
th
e
least
f
ea
tu
r
es.
Als
o
,
it
ca
n
b
e
t
h
er
ef
o
r
e
co
n
clu
d
ed
th
at
th
er
m
al
im
a
g
er
y
in
th
e
in
f
r
ar
ed
s
p
ec
tr
u
m
is
le
s
s
s
en
s
it
iv
e
to
th
e
v
ar
iatio
n
s
in
ex
ter
io
r
f
ea
tu
r
es
p
r
esen
t
in
th
e
im
ag
es
co
m
p
a
r
ed
to
im
ag
es
in
th
e
v
is
ib
le
s
p
ec
tr
u
m
.
Hen
ce
f
o
r
f
ac
e
id
en
tific
atio
n
p
u
r
p
o
s
es
u
s
in
g
th
er
m
al
im
ag
in
g
,
th
e
f
r
e
q
u
en
cy
d
o
m
ain
p
er
f
o
r
m
ed
b
e
tter
in
co
m
p
ar
is
o
n
with
a
s
p
atial
d
o
m
ain
.
3
.
2
.
I
dentif
y
ing
t
he
ef
f
ec
t
o
f
dis
t
o
rt
io
ns
in t
he
infr
a
re
d s
p
ec
t
rum
T
h
e
in
v
esti
g
atio
n
en
tails
ass
ess
in
g
th
e
th
er
m
al
im
a
g
e
’
s
s
u
s
ce
p
tib
ilit
y
to
th
e
n
o
is
y
e
n
v
ir
o
n
m
e
n
t.
Gau
s
s
ian
an
d
s
alt
an
d
p
ep
p
er
n
o
is
e
h
av
e
b
ee
n
a
d
d
ed
as
n
o
is
e
to
th
e
o
r
ig
in
al
th
er
m
al
im
ag
e.
T
h
e
p
ix
el’
s
v
alu
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
Th
erma
l ima
g
in
g
-
b
a
s
ed
id
en
ti
fica
tio
n
o
f fa
cia
l fe
a
t
u
r
es in
n
o
is
y
en
viro
n
men
t
(
P
a
la
k
Ma
h
a
ja
n
)
339
h
as
b
ee
n
ad
ju
s
ted
b
y
s
ettin
g
th
e
n
o
is
e
d
en
s
ity
v
alu
e
=
0
.
0
1
f
o
r
s
alt
an
d
p
ep
p
er
n
o
is
e.
Similar
ly
,
g
au
s
s
ia
n
n
o
is
e
h
as
b
ee
n
a
d
d
ed
with
=
0
.
1
.
A
s
am
p
le
o
f
tex
t
u
al
f
ac
ial
f
ea
tu
r
es
id
en
tifie
d
b
y
v
ar
io
u
s
f
ea
tu
r
e
ex
tr
ac
tio
n
tech
n
iq
u
es
u
n
d
er
t
h
e
in
f
lu
e
n
ce
o
f
n
o
is
e
is
s
h
o
wn
in
Fig
u
r
e
5
.
Her
e,
Fig
u
r
e
5
(
a)
r
ep
r
esen
ts
th
e
s
u
s
p
ec
tib
ilit
y
o
f
f
ea
tu
r
e
ex
tr
a
ctio
n
m
eth
o
d
s
u
n
d
er
s
alt
an
d
p
ep
p
er
n
o
is
e
at
n
o
is
e
d
en
s
ity
v
alu
e
=
0
.
0
1
a
n
d
Fig
u
r
e
5
(
b
)
illu
s
tr
atin
g
th
e
e
f
f
ec
t
o
f
g
a
u
s
ian
n
o
is
e
(
m
ea
n
=
0
.
1
)
o
n
th
ese
ex
tr
ac
tio
n
m
eth
o
d
s
with
OR
B
d
etec
tin
g
th
e
m
ax
im
u
m
f
ea
tu
r
es
alo
n
g
with
th
e
o
u
tlier
s
,
r
esp
ec
tiv
ely
.
Acc
o
r
d
in
g
l
y
th
e
co
m
p
u
ted
f
ea
t
u
r
e
p
o
in
ts
i
n
lier
s
,
F
IN
,
an
d
o
u
tlier
s
,
F
OUT
f
o
r
th
e
d
ataset
h
av
e
b
ee
n
f
u
r
th
e
r
r
ep
r
esen
te
d
in
T
ab
le
3
alo
n
g
with
th
ei
r
co
r
r
esp
o
n
d
in
g
PC
A
co
m
p
o
n
e
n
ts
.
As
d
ep
icted
b
y
th
e
T
a
b
le
3
(
a)
,
OR
B
f
ea
tu
r
e
ex
tr
ac
to
r
is
ab
le
to
co
m
p
u
te
m
ax
im
u
m
n
u
m
b
e
r
o
f
f
ea
tu
r
e
p
o
in
ts
with
F
IN
id
en
tify
in
g
th
e
m
ax
im
u
m
c
o
v
er
ed
f
ac
ial
r
eg
i
o
n
.
Fo
r
all
2
7
im
ag
es
i
n
T
I
D,
u
n
it
v
ar
ian
ce
s
ca
lin
g
,
a
n
d
p
r
in
cip
al
co
m
p
o
n
en
ts
h
av
e
b
ee
n
ca
lcu
lated
u
s
in
g
s
in
g
u
lar
v
alu
e
d
ec
o
m
p
o
s
itio
n
(
SVD)
with
im
p
u
tatio
n
.
I
n
th
e
f
ir
s
t
two
m
ajo
r
p
r
i
n
cip
al
co
m
p
o
n
en
ts
,
PC
A
ac
co
u
n
t
f
o
r
8
1
.
6
%
an
d
7
.
8
%
o
f
th
e
to
tal
v
ar
ia
n
ce
,
r
esp
ec
tiv
ely
f
o
r
all
s
ix
f
ea
tu
r
e
d
ete
ctio
n
tech
n
iq
u
es
in
th
e
ca
s
e
o
f
in
lier
s
as
well
as
o
u
tlier
s
.
T
h
is
in
d
icate
s
th
at
P
C
A
ef
f
ec
tiv
ely
co
n
d
e
n
s
es
th
e
s
ig
n
if
ican
t
in
f
o
r
m
atio
n
f
r
o
m
h
i
g
h
-
d
im
en
s
io
n
al
im
ag
e
c
o
m
p
o
n
en
t
to
a
s
m
aller
s
et
o
f
v
a
r
iab
le
s
p
ac
e.
T
h
is
s
p
ac
e
r
ed
u
ctio
n
f
u
r
th
er
ac
ce
ler
ates
th
e
ef
f
icen
cy
o
f
f
ea
tu
r
e
d
etec
to
r
s
b
y
o
n
l
y
f
o
c
u
s
in
g
o
n
m
o
s
t
s
ig
n
if
ican
t
f
ea
tu
r
es
as
in
d
i
cted
b
y
T
ab
le
3
(
b
)
,
r
ep
ec
tiv
el
y
.
Fu
r
th
er
clu
s
ter
in
g
is
p
er
f
o
r
m
ed
u
s
in
g
co
r
r
elatio
n
d
is
tan
ce
an
d
av
er
ag
e
lin
k
ag
e.
T
h
e
o
u
tp
u
t
o
f
clu
s
ter
in
g
g
en
er
ates h
ea
tm
ap
s
f
o
r
b
o
t
h
f
ea
tu
r
e
p
o
in
ts
in
lier
s
,
F
IN
,
an
d
o
u
tlier
s
,
F
OUT
is
r
ep
r
esen
ted
in
Fig
u
r
e
6
,
r
esp
ec
tiv
ely
.
T
h
e
h
ea
tm
a
p
in
Fig
u
r
e
6
(
a)
d
em
o
n
s
tr
ates
h
ig
h
co
r
r
elatio
n
with
d
ar
k
r
e
d
co
lo
r
r
ep
r
esen
tin
g
s
tr
o
n
g
s
im
ilar
ity
clu
s
ter
s
in
f
ac
ial
f
ea
tu
r
es
p
o
in
ts
with
o
r
a
n
g
e
co
lo
u
r
r
ep
r
esen
tin
g
ap
p
r
o
x
im
ately
less
s
im
ilar
ity
,
an
d
lo
w
u
n
co
m
m
o
n
d
is
tin
ct
f
ea
tu
r
es
h
av
e
b
ee
n
ilu
u
s
tr
ated
b
y
b
lu
e
co
lo
u
r
.
On
th
e
o
th
er
h
an
d
,
Fig
u
r
e
6
(
b
)
h
ig
h
lig
h
ts
th
e
o
u
tlier
’
s
h
ea
tm
ap
in
d
icatin
g
d
iv
er
s
io
n
s
am
o
n
g
i
n
co
n
s
is
ten
t
f
ea
tu
r
e
p
o
in
ts
i.e
.
clu
s
ter
s
f
o
r
m
ed
o
u
ts
id
e
f
ac
ial
r
eg
io
n
s
in
d
icate
d
b
y
th
e
c
o
lo
u
r
p
allets r
ed
,
o
r
a
n
g
e,
a
n
d
b
l
u
e
r
esp
ec
tiv
ely
.
T
h
ese
h
ea
tm
ap
v
is
u
aliza
tio
n
h
elp
s
in
id
en
tify
in
g
co
n
s
is
tan
t
f
ea
tu
r
e
p
o
in
t
clu
s
ter
s
alo
n
g
with
illu
s
tr
atin
g
ir
r
eg
u
lar
ities
,
th
er
eb
y
en
h
an
ci
n
g
th
e
r
o
b
u
s
tn
ess
.
T
h
r
o
u
g
h
o
u
t
th
e
en
tire
T
I
D
d
ataset,
all
s
ix
f
ea
tu
r
e
d
etec
tio
n
tech
n
i
q
u
es
h
av
e
b
ee
n
ab
le
to
id
en
tif
y
i
n
lier
f
ea
tu
r
es
to
a
m
ax
im
u
m
ex
ten
t
with
d
etec
tio
n
ar
o
u
n
d
th
e
n
o
s
e,
ey
es,
an
d
m
o
u
th
.
W
ith
OR
B
d
etec
tin
g
th
e
s
tr
o
n
g
est
f
ea
tu
r
es
am
o
n
g
s
t
th
e
lo
t.
Fig
u
r
e
5
s
h
o
wed
ex
te
r
n
al
p
a
r
am
eter
s
lik
e
n
o
is
e
p
la
y
s
a
s
ig
n
if
ican
t
r
o
le
in
d
etec
tin
g
f
ac
ial
f
ea
tu
r
e
in
th
er
m
al
im
ag
es.
L
ik
e
in
th
e
ca
s
e
o
f
Ga
u
s
s
ian
n
o
is
e,
t
h
e
n
o
is
e
elem
en
ts
h
av
e
co
v
er
ed
all
th
e
r
e
g
io
n
s
th
r
o
u
g
h
o
u
t
t
h
e
im
a
g
e
ca
u
s
in
g
th
e
f
e
atu
r
e
ex
t
r
ac
to
r
to
d
etec
t
m
o
r
e
o
u
tlier
s
r
ath
er
th
an
in
lier
s
.
OR
B
an
d
MSE
R
,
h
o
wev
er
,
s
h
o
w
t
h
e
wo
r
s
t
p
e
r
f
o
r
m
an
ce
in
th
e
p
r
esen
ce
o
f
n
o
is
e
as
g
iv
en
in
T
ab
le
4
.
Fu
r
th
er
,
as
o
b
s
er
v
ed
f
r
o
m
th
e
Fig
u
r
e,
f
ea
tu
r
e
d
etec
tio
n
in
th
er
m
al
im
ag
es
is
in
v
ar
ia
n
t
to
th
e
ef
f
ec
t
o
f
n
o
is
e
as
th
e
f
ea
tu
r
e
e
x
tr
ac
tio
n
tech
n
iq
u
es we
r
e
ab
le
to
d
etec
t th
e
f
ac
ial
f
ea
tu
r
es.
(
a)
(
b
)
Fig
u
r
e
4.
Featu
r
e
ex
tr
ac
tio
n
(
a
)
v
i
s
ib
le
s
p
ec
tr
u
m
an
d
(
b
)
in
f
r
ar
ed
s
p
ec
tr
u
m
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
13
,
No
.
3
,
Dec
em
b
er
20
24
:
333
-
3
4
3
340
(
a)
(
b
)
Fig
u
r
e
5
.
Featu
r
e
ex
tr
ac
tio
n
i
n
th
e
p
r
esen
ce
o
f
(
a)
s
alt
an
d
p
e
p
p
er
n
o
is
e
at
n
o
is
e
d
en
s
ity
=0
.
0
1
an
d
(
b
)
g
a
u
s
s
ian
n
o
is
e
at
m
ea
n
=0
.
1
f
o
r
t
h
e
T
I
D
d
ataset
T
ab
l
e
3
.
Featu
r
e
e
x
tr
ac
ted
wit
h
in
R
OI
,
F
IN
,
an
d
F
OUT
,
an
d
th
eir
r
esp
ec
tiv
e
PC
A
co
m
p
o
n
e
n
ts
in
T
I
D
(
a)
I
mag
e
F
A
S
T
[
1
8
]
H
a
r
r
i
s
[
22
]
O
R
B
[
2
0
]
M
S
ER
[
2
1]
H
O
G
[
19
]
EI
G
EN
[
2
3
]
F
IN
F
OUT
F
IN
F
OUT
F
IN
F
OUT
F
IN
F
OUT
F
IN
F
OUT
F
IN
F
OUT
I1
49
20
35
15
80
35
53
36
41
6
36
22
I2
55
21
36
14
81
36
69
35
40
14
38
26
I3
62
15
32
18
75
39
60
48
49
12
39
24
I4
43
18
40
20
77
41
62
42
35
7
45
36
I5
45
30
39
22
76
25
63
32
39
12
42
16
I6
52
34
41
25
69
39
71
16
38
15
40
19
I7
69
25
25
7
82
45
55
35
40
9
32
22
I8
75
22
28
19
75
41
54
32
45
7
36
27
I9
31
29
41
18
76
32
56
26
47
8
52
29
I
1
0
52
31
20
12
73
33
66
27
48
8
45
35
I
1
1
56
37
22
14
82
48
62
45
52
9
41
34
I
1
2
51
10
23
9
56
47
60
33
38
16
32
13
I
1
3
49
9
31
13
69
35
58
32
68
20
39
14
I
1
4
60
15
36
15
70
25
59
19
41
5
33
17
I
1
5
59
32
39
22
78
37
55
28
45
8
39
42
I
1
6
58
18
37
21
75
36
45
29
48
7
47
33
I
1
7
69
17
41
25
69
38
58
24
45
5
21
37
I
1
8
45
16
12
35
77
26
62
22
38
13
30
22
I
1
9
44
22
50
14
74
52
57
39
57
11
38
25
I
2
0
42
28
45
19
72
45
58
24
55
15
54
24
I
2
1
48
25
39
22
73
41
71
18
48
17
35
26
I
2
2
50
26
38
26
74
39
66
27
40
7
37
28
I
2
3
55
31
34
14
86
55
65
33
42
4
39
27
I
2
4
56
35
35
7
76
28
62
41
49
15
28
24
I
2
5
54
12
22
16
85
38
68
49
32
10
33
23
I
2
6
53
19
20
18
59
41
59
40
39
11
48
30
I
2
7
55
17
35
22
80
38
60
35
48
14
41
21
(
b
)
P
C
A
c
o
m
p
o
n
e
n
t
s
P
C
1
P
C
2
P
C
3
P
C
4
P
C
5
P
C
6
F
IN
F
OUT
F
IN
F
OUT
F
IN
F
OUT
F
IN
F
OUT
F
IN
F
OUT
F
IN
F
OUT
F
A
S
T
[
1
8
]
-
0
.
6
2
-
0
.
7
5
-
2
.
8
9
1
.
4
7
0
.
0
5
2
.
4
0
0
.
0
7
-
0
.
3
5
0
.
3
9
0
.
8
4
0
.
0
0
0
.
0
0
H
a
r
r
i
s
[
22
]
5
.
3
4
-
2
.
8
7
0
.
4
9
1
.
1
1
-
0
.
5
5
-
2
.
0
3
-
1
.
5
8
0
.
5
1
0
.
2
2
1
.
1
6
0
.
0
0
0
.
0
0
O
R
B
[
2
0
]
-
7
.
3
9
6
.
0
3
1
.
0
2
0
.
5
1
0
.
5
3
-
0
.
8
2
-
0
.
5
1
-
1
.
7
2
0
.
7
1
-
0
.
3
4
-
0
.
0
0
-
0
.
0
0
M
S
ER
[
2
1]
-
2
.
9
9
3
.
2
0
0
.
2
4
-
2
.
8
2
-
1
.
4
7
0
.
3
3
0
.
4
0
1
.
0
5
-
1
.
0
8
0
.
6
1
-
0
.
0
0
-
0
.
0
0
H
O
G
[
19
]
1
.
9
0
-
6
.
2
6
0
.
2
2
-
1
.
5
1
1
.
5
9
0
.
0
3
0
.
1
3
-
1
.
1
4
-
1
.
1
0
-
0
.
8
0
-
0
.
0
0
0
.
0
0
EI
G
EN
[
2
3
]
3
.
7
6
0
.
6
5
0
.
9
2
1
.
2
5
-
0
.
1
5
0
.
0
9
1
.
4
9
1
.
6
5
0
.
8
5
-
1
.
4
8
-
0
.
0
0
-
0
.
0
0
T
ab
le
4
.
Av
e
r
ag
e
n
u
m
b
er
o
f
f
ea
tu
r
e
p
o
in
ts
d
etec
ted
F
e
a
t
u
r
e
e
x
t
r
a
c
t
o
r
I
n
l
i
e
r
s
O
u
t
l
i
e
r
s
F
A
S
T
[
1
8
]
5
3
.
2
2
2
2
.
7
4
H
a
r
r
i
s [
22
]
3
3
.
1
8
1
7
.
8
5
O
R
B
[
2
0
]
7
4
.
7
8
3
8
.
3
3
M
S
ER
[
2
1]
6
0
.
5
2
3
2
.
1
1
H
O
G
[
19
]
4
4
.
7
0
1
0
.
5
6
EI
G
EN
[
2
3
]
3
8
.
5
1
2
5
.
7
8
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
Th
erma
l ima
g
in
g
-
b
a
s
ed
id
en
ti
fica
tio
n
o
f fa
cia
l fe
a
t
u
r
es in
n
o
is
y
en
viro
n
men
t
(
P
a
la
k
Ma
h
a
ja
n
)
341
(
a)
(
b
)
Fig
u
r
e
6
.
Hea
tm
ap
s
h
o
win
g
cl
u
s
ter
s
f
o
r
v
ar
io
u
s
im
ag
es in
th
e
T
I
D
d
ataset
f
o
r
v
ar
io
u
s
f
ea
t
u
r
e
ex
tr
ac
tio
n
tech
n
iq
u
es f
o
r
th
e
d
ataset
(
a)
i
n
lier
s
h
ea
tm
ap
in
d
icatin
g
an
d
(
b
)
o
u
tlier
s
h
ea
tm
ap
in
d
icatin
g
3
.
3
.
St
a
s
t
ica
l
a
na
ly
s
is
As
p
er
T
ab
le
4
,
OR
B
alg
o
r
ith
m
h
as
d
etec
ted
th
e
s
tr
o
n
g
est
an
d
th
e
m
ax
im
u
m
f
ea
tu
r
es
f
o
r
in
lier
s
id
en
tific
atio
n
,
t
o
v
alid
ate
th
e
p
er
f
o
r
m
an
ce
o
f
OR
B
we
ca
n
m
ea
s
u
r
e
th
e
s
tatis
tical
d
if
f
er
en
ce
b
etwe
en
all
th
e
alg
o
r
ith
m
s
.
T
o
test
th
e
p
er
f
o
r
m
an
ce
o
f
th
ese
f
ea
tu
r
e
ex
tr
ac
tio
n
tech
n
i
q
u
es
,
let’
s
h
y
p
o
th
esized
th
at
all
alg
o
r
ith
m
s
h
av
e
s
am
e
p
er
f
o
r
m
an
ce
m
ea
s
u
r
e
i
n
d
etec
tin
g
f
ac
ial
f
ea
tu
r
es
f
o
r
th
er
m
al
im
a
g
es
i.e
.
,
H
0
:
t
h
er
e
is
n
o
d
if
f
er
e
n
ce
in
f
ea
tu
r
e
ex
t
r
ac
tio
n
p
e
r
f
o
r
m
an
ce
o
f
all
s
ix
al
g
o
r
ith
m
s
,
H
1
:
s
o
m
e
f
ea
t
u
r
e
ex
tr
ac
tio
n
alg
o
r
ith
m
s
p
er
f
o
r
m
b
etter
th
an
t
h
e
o
th
e
r
s
.
Kr
u
s
k
al
-
W
allis
test
,
a
v
ar
ian
t
o
f
o
n
e
-
way
ANOV
A
test
ca
n
b
e
c
o
n
s
id
er
ed
f
o
r
m
ea
s
u
r
in
g
t
h
e
r
a
n
k
in
g
b
etwe
en
th
e
alg
o
r
ith
m
s
.
T
h
e
f
ea
tu
r
e
p
o
in
ts
o
u
tlier
s
,
FOUT
v
alu
es
p
r
ed
icted
b
y
th
e
m
o
d
el
h
av
e
b
ee
n
u
s
ed
i
n
Kr
u
s
k
al
-
W
allis
te
s
t
to
ac
ce
s
s
th
e
r
an
k
in
g
o
f
th
ese
alg
o
r
ith
m
s
.
T
ab
le
5
r
ep
r
esen
ts
th
e
p
er
f
o
r
m
an
ce
o
f
f
ea
tu
r
e
p
o
i
n
t d
etec
tio
n
alg
o
r
ith
m
s
u
s
in
g
Kr
u
s
k
al
-
W
alli
s
H
test
r
esp
ec
tiv
ely
.
Acc
o
r
d
in
g
ly
,
Kr
u
s
k
al
-
W
allis
H
test
in
d
icate
d
th
at
th
er
e
is
a
s
ig
n
if
ican
t
d
if
f
er
en
ce
i
n
th
e
d
ep
en
d
e
n
t
v
ar
iab
le
b
etwe
en
th
e
d
if
f
er
en
t
g
r
o
u
p
s
,
χ
2
(
5
)
=
1
0
3
.
8
4
,
p
<
.
0
0
1
,
with
a
m
ea
n
r
an
k
s
co
r
e
o
f
7
5
.
6
3
f
o
r
FAST,
5
3
.
2
8
f
o
r
H
ar
r
is
,
1
3
5
.
6
5
f
o
r
O
R
B
,
1
1
3
.
5
4
f
o
r
MSE
R
,
2
1
.
4
8
f
o
r
HOG,
8
9
.
4
3
f
o
r
E
I
GAN.
Fu
r
th
er
b
y
a
p
p
ly
in
g
a
B
o
n
f
er
r
o
n
i c
o
r
r
ec
ted
al
p
h
a
o
f
0
.
0
0
3
3
,
th
e
m
ea
n
r
a
n
k
s
o
f
th
e
f
o
llo
win
g
p
air
s
ar
e
s
i
g
n
if
ica
n
tly
d
if
f
er
e
n
t: x
1
-
x
2
x
1
-
x
3
x
1
-
x
5
x
2
-
x
3
x
2
-
x
4
x
3
-
x
5
x
3
-
x
6
x
4
-
x
5
x
4
-
x
6
wh
er
e
ea
ch
x
i
in
d
icate
s
f
ea
tu
r
e
ex
tr
ac
tio
n
alg
o
r
ith
m
r
esp
ec
tiv
ely
.
Sin
ce
th
e
p
-
v
alu
e
<
α
,
H0
is
r
ejec
ted
i.e
.
,
th
e
r
e
is
a
d
if
f
e
r
en
ce
b
etwe
en
th
e
m
ea
n
r
an
k
s
o
f
s
o
m
e
f
ea
tu
r
e
ex
tr
ac
tio
n
alg
o
r
ith
m
s
is
b
ig
en
o
u
g
h
to
b
e
s
tatis
ticall
y
s
ig
n
if
ican
t
as
s
h
o
wn
b
y
Fig
u
r
e
7
.
Fu
r
th
er
,
p
-
v
al
u
e
eq
u
als
to
0
.
He
n
ce
,
we
ca
n
co
n
clu
d
e
th
at
th
e
alg
o
r
ith
m
s
b
eh
av
e
d
if
f
er
e
n
tly
with
s
i
g
n
i
f
ican
t
p
er
f
o
r
m
a
n
ce
d
if
f
er
en
ce
.
Fig
u
r
e
7
.
Statis
tically
d
is
tr
ib
u
ted
v
alu
es f
o
r
Kr
u
s
k
al
-
W
allis
H
test
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
13
,
No
.
3
,
Dec
em
b
er
20
24
:
333
-
3
4
3
342
T
ab
le
5
.
Kr
u
s
k
al
-
W
allis
H
tes
t
p
er
f
o
r
m
an
ce
f
o
r
v
ar
io
u
s
f
ea
t
u
r
e
ex
tr
ac
tio
n
tec
h
n
iq
u
es
G
r
o
u
p
H
a
r
r
i
s
O
R
B
M
S
ER
HOG
EI
G
A
N
F
A
S
T
5
5
.
7
-
5
8
.
1
5
-
3
0
.
5
4
4
9
.
4
6
2
6
.
6
3
H
a
r
r
i
S
0
-
1
1
3
.
8
5
-
8
6
.
2
4
-
6
.
2
4
-
2
9
.
0
7
O
R
B
-
1
1
3
.
8
5
0
2
7
.
6
1
1
0
7
.
6
1
8
4
.
7
8
M
S
ER
-
8
6
.
2
4
2
7
.
6
1
0
80
5
7
.
1
7
HOG
-
6
.
2
4
1
0
7
.
6
1
80
0
-
2
2
.
8
3
4.
CO
NCLU
SI
O
N
Am
o
n
g
s
t
th
e
m
o
s
t
p
o
p
u
lar
a
n
d
ef
f
ec
tiv
e
ap
p
licatio
n
s
o
f
b
io
m
etr
ic
tech
n
o
lo
g
y
is
f
ac
e
r
ec
o
g
n
itio
n
.
T
h
e
asp
ec
t o
f
th
er
m
al
im
a
g
es in
th
e
in
f
r
ar
e
d
b
an
d
h
as b
ee
n
ex
p
lo
r
ed
f
o
r
th
e
f
ac
e
r
ec
o
g
n
itio
n
p
r
o
ce
s
s
.
A
n
o
v
el
f
ac
e
d
ataset
h
as
b
ee
n
cr
ea
ted
,
co
m
p
r
is
in
g
2
7
th
er
m
al
im
a
g
es
alo
n
g
with
th
eir
co
r
r
esp
o
n
d
in
g
v
is
u
al
b
an
d
im
ag
es.
T
h
e
f
ea
tu
r
e
ex
tr
ac
tio
n
tech
n
iq
u
es
FAST,
OR
B
,
M
SER,
Har
r
is
,
etc.
h
av
e
b
ee
n
a
p
p
lied
f
o
r
lo
ca
tin
g
f
ac
ial
f
ea
tu
r
es
lik
e
ey
es,
m
o
u
t
h
,
an
d
n
o
s
e.
T
h
e
r
esu
lts
s
h
o
wed
th
e
s
en
s
itiv
ity
o
f
v
is
ib
le
s
p
ec
tr
u
m
im
ag
es
f
o
r
tex
tu
r
al
p
atter
n
s
p
r
esen
t
in
th
e
ap
p
ar
el.
Alm
o
s
t
all
tech
n
iq
u
es
ex
ce
p
t
OR
B
d
etec
ted
o
n
ly
o
u
tlier
s
in
v
is
ib
le
s
p
ec
tr
u
m
im
ag
es.
On
th
e
o
t
h
er
h
an
d
,
in
th
e
ca
s
e
o
f
th
e
r
m
al
im
ag
es
in
th
e
in
f
r
ar
e
d
s
p
e
ctr
u
m
,
th
ese
o
u
tlier
f
ea
tu
r
es
h
av
e
b
ee
n
co
m
p
letely
m
ask
ed
,
an
d
m
ax
im
u
m
in
l
ier
f
ea
tu
r
es
s
u
ch
as
r
eg
io
n
s
n
ea
r
th
e
ey
es
an
d
ch
ee
k
s
h
av
e
b
e
e
n
o
b
s
er
v
e
d
.
Als
o
,
th
eir
ef
f
icac
y
h
as
b
ee
n
s
u
b
s
tan
tiated
in
d
etec
tin
g
f
a
cial
f
ea
tu
r
es
in
th
e
p
r
esen
ce
o
f
e
x
ter
n
al
n
o
is
e.
T
h
e
r
esu
lts
s
h
o
wed
th
at
ex
ter
n
al
n
o
is
e
p
lay
s
a
s
ig
n
if
ican
t
r
o
le
in
d
etec
tin
g
f
ac
ial
f
ea
tu
r
es
in
th
e
r
m
al
im
ag
es.
S
alt
an
d
p
e
p
p
er
n
o
is
e
s
h
o
ws
le
s
s
s
en
s
it
iv
ity
as
co
m
p
ar
ed
to
Gau
s
s
ian
n
o
is
e,
as
m
o
r
e
o
u
tlier
s
th
an
in
lier
s
h
av
e
b
ee
n
d
etec
ted
f
o
r
im
ag
es
with
Gau
s
s
ian
n
o
is
e.
Fu
r
th
er
,
th
e
wo
r
k
ca
n
b
e
ex
ten
d
ed
b
y
in
teg
r
atin
g
GAN
s
an
d
R
NNs
to
s
y
n
th
esize
an
d
au
g
m
en
t
th
e
r
m
al
an
d
v
is
u
al
s
p
ec
tr
u
m
im
ag
e
s
f
o
r
ex
ten
s
iv
e
f
ea
tu
r
e
e
x
tr
ac
tio
n
a
n
d
id
en
tific
atio
n
p
r
o
ce
s
s
.
RE
F
E
R
E
NC
E
S
[
1
]
G
.
T
a
n
d
a
,
“
T
h
e
u
se
o
f
i
n
f
r
a
r
e
d
t
h
e
r
mo
g
r
a
p
h
y
t
o
d
e
t
e
c
t
t
h
e
s
k
i
n
t
e
m
p
e
r
a
t
u
r
e
r
e
s
p
o
n
s
e
t
o
p
h
y
si
c
a
l
a
c
t
i
v
i
t
y
,
”
J
o
u
r
n
a
l
o
f
P
h
y
s
i
c
s
:
C
o
n
f
e
re
n
c
e
S
e
r
i
e
s
,
v
o
l
.
6
5
5
,
n
o
.
1
,
p
.
0
1
2
0
6
2
,
N
o
v
.
2
0
1
5
,
d
o
i
:
1
0
.
1
0
8
8
/
1
7
4
2
-
6
5
9
6
/
6
5
5
/
1
/
0
1
2
0
6
2
.
[
2
]
P
.
M
a
h
a
j
a
n
,
P
.
A
b
r
o
l
,
a
n
d
P
.
K
.
L
e
h
a
n
a
,
“
S
c
e
n
e
b
a
se
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
a
e
r
i
a
l
i
ma
g
e
s
u
si
n
g
c
o
n
v
o
l
u
t
i
o
n
n
e
u
r
a
l
n
e
t
w
o
r
k
s,
”
J
o
u
rn
a
l
o
f
S
c
i
e
n
t
i
f
i
c
& I
n
d
u
st
ri
a
l
Re
s
e
a
r
c
h
,
v
o
l
.
7
9
,
n
o
.
1
2
,
p
p
.
1
0
8
7
–
1
0
9
4
,
D
e
c
.
2
0
2
0
,
d
o
i
:
1
0
.
5
6
0
4
2
/
j
si
r
.
v
7
9
i
1
2
.
3
6
6
7
1
.
[
3
]
P
.
M
a
h
a
j
a
n
,
V
.
J
a
k
h
e
t
i
y
a
,
P
.
A
b
r
o
l
,
P
.
K
.
Le
h
a
n
a
,
B
.
N
.
S
u
b
u
d
h
i
,
a
n
d
S
.
C
.
G
u
n
t
u
k
u
,
“
P
e
r
c
e
p
t
u
a
l
q
u
a
l
i
t
y
e
v
a
l
u
a
t
i
o
n
o
f
h
a
z
y
n
a
t
u
r
a
l
i
ma
g
e
s,
”
I
E
EE
T
r
a
n
sa
c
t
i
o
n
s
o
n
I
n
d
u
s
t
ri
a
l
I
n
f
o
rm
a
t
i
c
s
,
v
o
l
.
1
7
,
n
o
.
1
2
,
p
p
.
8
0
4
6
–
8
0
5
6
,
D
e
c
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
TI
I
.
2
0
2
1
.
3
0
6
5
4
3
9
.
[
4
]
D
.
G
h
o
s
e
,
S
.
M
.
D
e
sai
,
S
.
B
h
a
t
t
a
c
h
a
r
y
a
,
D
.
C
h
a
k
r
a
b
o
r
t
y
,
M
.
F
i
t
e
r
a
u
,
a
n
d
T
.
R
a
h
ma
n
,
“
P
e
d
e
s
t
r
i
a
n
d
e
t
e
c
t
i
o
n
i
n
t
h
e
r
mal
i
ma
g
e
s
u
si
n
g
sal
i
e
n
c
y
m
a
p
s
,
”
I
EEE
C
o
m
p
u
t
e
r
S
o
c
i
e
t
y
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
V
i
si
o
n
a
n
d
P
a
t
t
e
r
n
Re
c
o
g
n
i
t
i
o
n
Wo
r
k
sh
o
p
s
,
v
o
l
.
2
0
1
9
-
Ju
n
e
,
p
p
.
9
8
8
–
9
9
7
,
2
0
1
9
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
W
.
2
0
1
9
.
0
0
1
3
0
.
[
5
]
M
.
K
.
B
h
o
w
mi
k
e
t
a
l
.
,
“
T
h
e
r
m
a
l
i
n
f
r
a
r
e
d
f
a
c
e
r
e
c
o
g
n
i
t
i
o
n
–
a
b
i
o
me
t
r
i
c
i
d
e
n
t
i
f
i
c
a
t
i
o
n
t
e
c
h
n
i
q
u
e
f
o
r
r
o
b
u
s
t
sec
u
r
i
t
y
sy
s
t
e
m
,
”
i
n
Re
v
i
e
w
s,
Re
f
i
n
e
m
e
n
t
s
a
n
d
N
e
w
I
d
e
a
s
i
n
Fa
c
e
Re
c
o
g
n
i
t
i
o
n
,
I
n
Te
c
h
,
2
0
1
1
.
[
6
]
D
.
A
.
S
o
c
o
l
i
n
sk
y
,
A
.
S
e
l
i
n
g
e
r
,
a
n
d
J
.
D
.
N
e
u
h
e
i
se
l
,
“
F
a
c
e
r
e
c
o
g
n
i
t
i
o
n
w
i
t
h
v
i
s
i
b
l
e
a
n
d
t
h
e
r
mal
i
n
f
r
a
r
e
d
i
ma
g
e
r
y
,
”
C
o
m
p
u
t
e
r
Vi
si
o
n
a
n
d
I
m
a
g
e
U
n
d
e
rst
a
n
d
i
n
g
,
v
o
l
.
9
1
,
n
o
.
1
–
2
,
p
p
.
7
2
–
1
1
4
,
J
u
l
.
2
0
0
3
,
d
o
i
:
1
0
.
1
0
1
6
/
S
1
0
7
7
-
3
1
4
2
(
0
3
)
0
0
0
7
5
-
4.
[
7
]
A
.
K
u
z
d
e
u
o
v
,
D
.
A
u
b
a
k
i
r
o
v
a
,
D
.
K
o
i
s
h
i
g
a
r
i
n
a
,
a
n
d
H
.
A
.
V
a
r
o
l
,
“
TFW
:
a
n
n
o
t
a
t
e
d
t
h
e
r
ma
l
f
a
c
e
s
i
n
t
h
e
w
i
l
d
d
a
t
a
se
t
,
”
I
EEE
T
r
a
n
s
a
c
t
i
o
n
s
o
n
I
n
f
o
rm
a
t
i
o
n
Fo
ren
s
i
c
s
a
n
d
S
e
c
u
ri
t
y
,
v
o
l
.
1
7
,
p
p
.
2
0
8
4
–
2
0
9
4
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
TI
F
S
.
2
0
2
2
.
3
1
7
7
9
4
9
.
[
8
]
B
.
M
a
r
t
i
n
e
z
,
X
.
B
i
n
e
f
a
,
a
n
d
M
.
P
a
n
t
i
c
,
“
F
a
c
i
a
l
c
o
m
p
o
n
e
n
t
d
e
t
e
c
t
i
o
n
i
n
t
h
e
r
m
a
l
i
m
a
g
e
r
y
,
”
i
n
2
0
1
0
I
E
E
E
C
o
m
p
u
t
e
r
S
o
c
i
e
t
y
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
e
r
V
i
s
i
o
n
a
n
d
P
a
t
t
e
r
n
R
e
c
o
g
n
i
t
i
o
n
-
W
o
r
k
s
h
o
p
s
,
J
u
n
.
2
0
1
0
,
p
p
.
4
8
–
5
4
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
W
.
2
0
1
0
.
5
5
4
3
6
0
5
.
[
9
]
M
.
K
.
B
h
o
w
mi
k
,
S
.
S
h
i
l
,
a
n
d
P
.
S
a
h
a
,
“
F
e
a
t
u
r
e
p
o
i
n
t
s
e
x
t
r
a
c
t
i
o
n
o
f
t
h
e
r
mal
f
a
c
e
u
s
i
n
g
H
a
r
r
i
s
i
n
t
e
r
e
s
t
p
o
i
n
t
d
e
t
e
c
t
i
o
n
,
”
Pr
o
c
e
d
i
a
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
1
0
,
p
p
.
7
2
4
–
7
3
0
,
2
0
1
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
r
o
t
c
y
.
2
0
1
3
.
1
2
.
4
1
5
.
[
1
0
]
C
.
N
.
F
o
n
d
j
e
,
S
.
H
u
,
a
n
d
B
.
S
.
R
i
g
g
a
n
,
“
L
e
a
r
n
i
n
g
d
o
m
a
i
n
a
n
d
p
o
s
e
i
n
v
a
r
i
a
n
c
e
f
o
r
t
h
e
r
m
a
l
-
to
-
v
i
s
i
b
l
e
f
a
c
e
r
e
c
o
g
n
i
t
i
o
n
,
”
I
E
E
E
T
r
a
n
s
a
c
t
i
o
n
s
o
n
B
i
o
m
e
t
r
i
c
s
,
B
e
h
a
v
i
o
r
,
a
n
d
I
d
e
n
t
i
t
y
S
c
i
e
n
c
e
,
v
o
l
.
5
,
n
o
.
1
,
p
p
.
1
5
–
2
8
,
J
a
n
.
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
T
B
I
O
M
.
2
0
2
2
.
3
2
2
3
0
5
5
.
[
1
1
]
A
.
M
.
G
u
z
ma
n
e
t
a
l
.
,
“
T
h
e
r
m
a
l
i
ma
g
i
n
g
a
s
a
b
i
o
me
t
r
i
c
s
a
p
p
r
o
a
c
h
t
o
f
a
c
i
a
l
si
g
n
a
t
u
r
e
a
u
t
h
e
n
t
i
c
a
t
i
o
n
,
”
I
EEE
J
o
u
r
n
a
l
o
f
Bi
o
m
e
d
i
c
a
l
a
n
d
H
e
a
l
t
h
I
n
f
o
rm
a
t
i
c
s
,
v
o
l
.
1
7
,
n
o
.
1
,
p
p
.
2
1
4
–
2
2
2
,
2
0
1
3
.
[
1
2
]
P
.
R
i
c
a
u
r
t
e
,
C
.
C
h
i
l
á
n
,
C
.
A
g
u
i
l
e
r
a
-
C
a
r
r
a
sc
o
,
B
.
V
i
n
t
i
mi
l
l
a
,
a
n
d
A
.
S
a
p
p
a
,
“
F
e
a
t
u
r
e
p
o
i
n
t
d
e
scri
p
t
o
r
s
:
i
n
f
r
a
r
e
d
a
n
d
v
i
s
i
b
l
e
sp
e
c
t
r
a
,
”
S
e
n
so
rs
,
v
o
l
.
1
4
,
n
o
.
2
,
p
p
.
3
6
9
0
–
3
7
0
1
,
F
e
b
.
2
0
1
4
,
d
o
i
:
1
0
.
3
3
9
0
/
s1
4
0
2
0
3
6
9
0
.
[
1
3
]
I
.
S
i
p
i
r
a
n
a
n
d
B
.
B
u
s
t
o
s,
“
A
r
o
b
u
s
t
3
D
i
n
t
e
r
e
st
p
o
i
n
t
s
d
e
t
e
c
t
o
r
b
a
se
d
o
n
h
a
r
r
i
s
o
p
e
r
a
t
o
r
,
”
E
u
r
o
g
r
a
p
h
i
c
s
Wo
r
k
sh
o
p
o
n
3
D
O
b
j
e
c
t
Re
t
r
i
e
v
a
l
,
EG
3
D
O
R
,
p
p
.
7
–
1
4
,
2
0
1
0
,
d
o
i
:
1
0
.
2
3
1
2
/
3
D
O
R
/
3
D
O
R
1
0
/
0
0
7
-
0
1
4
.
[
1
4
]
O
.
Jo
s
h
u
a
,
J.
S
.
O
w
o
t
o
g
b
e
,
T.
S
.
I
b
i
y
e
mi
,
a
n
d
B
.
A
.
A
d
u
,
“
A
c
o
mp
r
e
h
e
n
s
i
v
e
r
e
v
i
e
w
o
n
v
a
r
i
o
u
s
t
y
p
e
s
o
f
n
o
i
se
i
n
i
mag
e
p
r
o
c
e
ssi
n
g
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
S
c
i
e
n
t
i
f
i
c
a
n
d
En
g
i
n
e
e
r
i
n
g
Re
s
e
a
r
c
h
,
v
o
l
.
1
0
,
p
p
.
3
8
8
–
3
9
3
,
2
0
1
9
,
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
:
/
/
w
w
w
.
i
j
s
e
r
.
o
r
g
.
[
1
5
]
P
.
B
.
D
a
s
g
u
p
t
a
,
“
A
n
a
l
y
t
i
c
a
l
c
o
m
p
a
r
i
so
n
o
f
n
o
i
se
r
e
d
u
c
t
i
o
n
f
i
l
t
e
r
s
f
o
r
i
mag
e
r
e
s
t
o
r
a
t
i
o
n
u
si
n
g
S
N
R
e
s
t
i
m
a
t
i
o
n
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
C
o
m
p
u
t
e
r Tre
n
d
s
a
n
d
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
1
7
,
n
o
.
3
,
p
p
.
1
2
1
–
1
2
4
,
N
o
v
.
2
0
1
4
,
d
o
i
:
1
0
.
1
4
4
4
5
/
2
2
3
1
2
8
0
3
/
I
JC
TT
-
V
1
7
P
1
2
3
.
[
1
6
]
P
.
M
a
h
a
j
a
n
,
P
.
A
b
r
o
l
,
a
n
d
P
.
K
.
L
e
h
a
n
a
,
“
Ef
f
e
c
t
o
f
b
l
u
r
r
i
n
g
o
n
i
d
e
n
t
i
f
i
c
a
t
i
o
n
o
f
a
e
r
i
a
l
i
ma
g
e
s
u
si
n
g
c
o
n
v
o
l
u
t
i
o
n
n
e
u
r
a
l
n
e
t
w
o
r
k
s,
”
i
n
L
e
c
t
u
r
e
N
o
t
e
s
i
n
E
l
e
c
t
r
i
c
a
l
En
g
i
n
e
e
ri
n
g
,
v
o
l
.
5
9
7
,
2
0
2
0
,
p
p
.
4
6
9
–
4
8
4
.
[
1
7
]
P
.
Lu
o
,
X
.
Zh
a
n
g
,
Z.
C
h
a
n
g
,
a
n
d
W
.
Li
u
,
“
R
e
s
e
a
r
c
h
o
n
s
a
l
t
a
n
d
p
e
p
p
e
r
n
o
i
s
e
r
e
m
o
v
a
l
m
e
t
h
o
d
b
a
s
e
d
o
n
a
d
a
p
t
i
v
e
f
u
z
z
y
me
d
i
a
n
f
i
l
t
e
r
,
”
i
n
2
0
2
1
I
EE
E
5
t
h
Ad
v
a
n
c
e
d
I
n
f
o
rm
a
t
i
o
n
T
e
c
h
n
o
l
o
g
y
,
El
e
c
t
ro
n
i
c
a
n
d
A
u
t
o
m
a
t
i
o
n
C
o
n
t
r
o
l
C
o
n
f
e
re
n
c
e
(
I
AEAC
)
,
M
a
r
.
2
0
2
1
,
p
p
.
3
8
7
–
392
,
d
o
i
:
1
0
.
1
1
0
9
/
I
A
EA
C
5
0
8
5
6
.
2
0
2
1
.
9
3
9
0
9
2
3
.
[
1
8
]
E.
R
o
s
t
e
n
a
n
d
T
.
D
r
u
mm
o
n
d
,
“
F
u
s
i
n
g
p
o
i
n
t
s
a
n
d
l
i
n
e
s
f
o
r
h
i
g
h
p
e
r
f
o
r
m
a
n
c
e
t
r
a
c
k
i
n
g
,
”
i
n
T
e
n
t
h
I
EEE
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
s
i
o
n
(
I
C
C
V’
0
5
)
V
o
l
u
m
e
1
,
2
0
0
5
,
v
o
l
.
I
I
,
p
p
.
1
5
0
8
-
1
5
1
5
V
o
l
.
2
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
V
.
2
0
0
5
.
1
0
4
.
[
1
9
]
N
.
D
a
l
a
l
a
n
d
B
.
Tr
i
g
g
s
,
“
H
i
s
t
o
g
r
a
m
s
o
f
o
r
i
e
n
t
e
d
g
r
a
d
i
e
n
t
s
f
o
r
h
u
m
a
n
d
e
t
e
c
t
i
o
n
,
”
i
n
2
0
0
5
I
EEE
C
o
m
p
u
t
e
r
S
o
c
i
e
t
y
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
s
i
o
n
a
n
d
P
a
t
t
e
r
n
Re
c
o
g
n
i
t
i
o
n
(
C
VP
R’
0
5
)
,
2
0
0
5
,
v
o
l
.
1
,
n
o
.
7
,
p
p
.
8
8
6
–
8
9
3
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
.
2
0
0
5
.
1
7
7
.
Evaluation Warning : The document was created with Spire.PDF for Python.