T
E
L
KO
M
N
I
KA
T
e
lec
om
m
u
n
icat
ion
,
Com
p
u
t
i
n
g,
E
lec
t
r
on
ics
an
d
Cont
r
ol
Vol.
18
,
No.
3
,
J
une
2020
,
pp.
1216
~
1223
I
S
S
N:
1693
-
6930,
a
c
c
r
e
dit
e
d
F
ir
s
t
G
r
a
de
by
Ke
me
nr
is
tekdikti
,
De
c
r
e
e
No:
21/E
/KP
T
/2018
DO
I
:
10.
12928/
T
E
L
KO
M
NI
KA
.
v18i3.
13541
1216
Jou
r
n
al
h
omepage
:
ht
tp:
//
jour
nal.
uad
.
ac
.
id/
index
.
php/T
E
L
K
OM
N
I
K
A
Im
age
p
r
oc
e
ssi
n
g an
al
ysi
s of
s
ig
m
oi
d
al
H
ad
am
ar
d
w
av
e
le
t
w
ith
PC
A
t
o d
e
t
e
c
t
h
id
d
e
n
ob
j
e
c
t
Am
m
ar
Wis
am
Al
t
ah
e
r
1
,
S
a
b
ah
K
h
u
d
h
air
A
b
b
as
2
1
D
ep
ar
t
men
t
o
f
In
f
o
rmat
i
o
n
T
ech
n
o
l
o
g
y
,
T
ech
n
i
ca
l
Co
l
l
ag
e
o
f
Man
a
g
emen
t
,
Al
-
Fu
rat
A
l
-
A
w
s
a
t
T
ec
h
n
i
cal
U
n
i
v
ers
i
t
y
,
Iraq
2
Imam
A
l
-
K
ad
h
u
m
U
n
i
v
er
s
i
t
y
Co
l
l
e
g
e
N
a
j
af
,
Iraq
Ar
t
icle
I
n
f
o
AB
S
T
RA
CT
A
r
ti
c
le
h
is
tor
y
:
R
e
c
e
ived
J
ul
9
,
2019
R
e
vis
e
d
J
a
n
29
,
2020
Ac
c
e
pted
F
e
b
26
,
2020
In
n
o
v
a
t
i
v
e
t
ac
t
i
c
s
are
em
p
l
o
y
e
d
b
y
t
erro
r
i
s
t
s
t
o
c
o
n
cea
l
w
ea
p
o
n
s
a
n
d
ex
p
l
o
s
i
v
es
t
o
p
erp
e
t
rat
e
v
i
o
l
en
t
at
t
ack
s
,
acco
u
n
t
i
n
g
fo
r
t
h
e
d
ea
t
h
s
o
f
mi
l
l
i
o
n
s
o
f
l
i
v
es
e
v
ery
y
ear
a
n
d
c
o
n
t
ri
b
u
t
i
n
g
t
o
h
u
g
e
eco
n
o
m
i
c
l
o
s
s
es
t
o
t
h
e
g
l
o
b
al
s
o
c
i
et
y
.
A
ch
i
e
v
i
n
g
a
h
i
g
h
t
h
rea
t
d
et
ec
t
i
o
n
rat
e
d
u
r
i
n
g
an
i
n
s
p
ect
i
o
n
o
f
cro
w
d
s
t
o
reco
g
n
i
ze
an
d
d
e
t
ect
t
h
reat
e
l
emen
t
s
fro
m
a
s
ecu
re
d
i
s
t
a
n
ce
i
s
t
h
e
m
o
t
i
v
a
t
i
o
n
f
o
r
t
h
e
d
ev
e
l
o
p
men
t
o
f
i
n
t
e
l
l
i
g
e
n
t
i
ma
g
e
d
at
a
a
n
al
y
s
i
s
fro
m
a
mach
i
n
e
l
earn
i
n
g
p
ers
p
ec
t
i
v
e.
A
met
h
o
d
p
ro
p
o
s
ed
t
o
red
u
ce
t
h
e
i
ma
g
e
d
i
me
n
s
i
o
n
s
w
i
t
h
s
u
p
p
o
rt
v
ec
t
o
r,
l
i
n
eari
t
y
a
n
d
o
r
t
h
o
g
o
n
al
.
T
h
e
fu
n
ct
i
o
n
al
i
t
y
o
f
CW
D
i
s
co
n
t
i
n
g
en
t
u
p
o
n
t
h
e
p
l
en
ary
ch
arac
t
eri
za
t
i
o
n
o
f
fu
s
i
o
n
d
a
t
a
fro
m
mu
l
t
i
p
l
e
i
mag
e
s
en
s
o
r
s
.
T
h
e
p
r
o
p
o
s
e
d
met
h
o
d
co
m
b
i
n
e
s
mu
l
t
i
p
l
e
s
e
n
s
o
rs
b
y
h
y
b
ri
d
fu
s
i
o
n
o
f
s
i
g
m
o
i
d
al
H
a
d
amar
d
w
av
e
l
et
t
ran
s
f
o
rm
an
d
PC
A
b
as
i
s
fu
n
c
t
i
o
n
s
.
W
eap
o
n
r
eco
g
n
i
t
i
o
n
an
d
t
h
e
d
et
ec
t
i
o
n
s
y
s
t
em,
u
s
i
n
g
Ima
g
e
s
eg
me
n
t
a
t
i
o
n
a
n
d
K
mean
s
s
u
p
p
o
r
t
v
e
c
t
o
r
mach
i
n
e
A
cl
a
s
s
i
fi
er
i
s
a
n
au
t
o
n
o
mo
u
s
p
ro
ce
s
s
f
o
r
t
h
e
rec
o
g
n
i
t
i
o
n
o
f
t
h
rea
t
w
eap
o
n
s
re
g
ard
l
es
s
o
f
mak
e,
v
ari
e
t
y
,
s
h
ap
e,
o
r
p
o
s
i
t
i
o
n
o
n
t
h
e
s
u
s
p
ec
t
’s
b
o
d
y
d
es
p
i
t
e
co
n
ceal
me
n
t
.
K
e
y
w
o
r
d
s
:
C
onc
e
a
led
we
a
pon
de
tec
ti
on
I
R
i
mage
P
r
inciple
c
omponent
a
na
lys
is
S
igm
oidal
H
a
da
mar
d
S
uppor
t
voc
ter
mac
hine
Th
i
s
i
s
a
n
o
p
en
a
c
ces
s
a
r
t
i
c
l
e
u
n
d
e
r
t
h
e
CC
B
Y
-
SA
l
i
ce
n
s
e
.
C
or
r
e
s
pon
din
g
A
u
th
or
:
Ammar
W
is
a
m
Altahe
r
,
De
pa
r
tm
e
nt
of
I
nf
or
mat
ion
T
e
c
hnology
,
T
e
c
hnica
l
C
oll
a
ge
of
M
a
na
ge
ment,
Al
-
F
ur
a
t
Al
-
Aw
s
a
t
T
e
c
hnica
l
Unive
r
s
it
y,
I
r
a
q
.
E
mail:
d
r
.
a
mm
a
r
@a
tu
.
e
du.
iq
1.
I
NT
RODU
C
T
I
ON
R
e
mot
e
de
tec
ti
on
of
pe
r
s
on
bor
ne
c
onc
e
a
led
we
a
p
ons
f
r
om
a
s
e
c
ur
e
dis
tanc
e
is
th
e
ne
e
d
of
the
hour
but
c
ur
r
e
ntl
y,
a
va
il
a
ble
de
tec
ti
on
s
ys
tems
r
e
quir
e
phys
ica
l
e
xa
mi
na
ti
on
withi
n
a
pr
e
s
c
r
ibed
dis
tanc
e
.
De
tec
ti
ng
the
pr
e
s
e
nc
e
of
we
a
pons
s
uc
h
a
s
guns
,
knives
,
e
xplos
ives
,
e
tc.
,
via
c
ha
nne
l
c
he
c
kpoint
s
is
time
-
c
ons
umi
ng
lea
ding
to
de
lays
,
a
nd
is
a
ls
o
s
us
c
e
pti
ble
to
a
high
incide
nc
e
o
f
f
a
ls
e
a
lar
ms
.
M
a
ny
a
dva
nc
e
d
s
ur
ve
il
lanc
e
s
ys
tems
a
r
e
de
ve
loped
by
r
e
s
e
a
r
c
he
s
with
r
obus
t,
inbui
lt
s
e
ns
or
s
that
c
a
n
ope
r
a
te
f
r
om
a
s
e
c
ur
e
dis
tanc
e
to
de
tec
t
c
onc
e
a
led
w
e
a
pons
without
a
ny
phys
ica
l
c
ontac
t.
M
a
ny
va
r
iants
of
s
e
ns
or
tec
hnolo
gy
s
uc
h
a
s
ther
mal
or
in
f
r
a
r
e
d,
mi
c
r
owa
ve
,
mi
l
li
mete
r
wa
ve
s
,
or
ter
a
he
r
tz
a
r
e
a
va
il
a
ble,
but
only
non
-
inva
s
ive
a
nd
non
-
ha
z
a
r
dous
s
e
ns
or
s
a
r
e
us
e
d
in
publi
c
s
e
c
ur
it
y
s
ur
ve
il
lanc
e
s
ys
tems
.
T
o
im
pr
ove
the
s
pa
ti
a
l
r
e
s
olut
ion,
mul
ti
s
pe
c
tr
a
l
.
Da
ta
o
f
the
s
a
me
s
c
e
ne
,
take
n
f
r
o
m
mul
ti
ple
s
e
ns
or
s
a
r
e
c
ombi
ne
d
th
r
ough
a
pr
oc
e
s
s
c
a
ll
e
d
I
mage
F
us
ion.
Di
f
f
e
r
e
nt
im
a
ge
f
us
ion
tec
hni
que
s
ha
ve
be
e
n
de
ve
loped
to
take
f
u
ll
a
dva
ntage
of
the
int
e
gr
a
ti
on
o
f
c
ompl
e
menta
r
y
inf
or
mation
a
nd
ha
ve
f
ound
e
xtens
ive
a
ppli
c
a
ti
ons
in
the
f
ields
of
c
omput
e
r
vis
ion,
r
e
mot
e
s
e
ns
ing,
medic
ine,
biom
e
tr
ics
,
r
ob
oti
c
s
,
e
tc.
T
he
block
diagr
a
m
o
f
ge
ne
r
a
l
im
a
ge
f
us
ion
is
s
hown
in
F
igur
e
1
.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
I
mage
pr
oc
e
s
s
ing
analys
is
of
s
igmoidal
H
adamar
d
w
av
e
let
w
it
h
P
C
A
to
de
tec
t
.
.
.
.
(
A
mm
ar
W
is
am
A
lt
ahe
r
)
1217
F
igur
e
1
.
Ge
ne
r
a
l
i
mage
f
us
ion
s
c
he
ma
T
he
da
ta
a
c
quir
e
d
f
r
om
mul
ti
ple
s
e
ns
or
s
is
pa
s
s
e
d
thr
ough
a
pr
e
-
pr
oc
e
s
s
ing
r
e
gis
tr
a
ti
on
s
tage
,
whic
h
is
a
pixel
to
pixel
c
or
r
e
s
ponde
nc
e
to
a
void
a
ny
incons
is
tenc
ies
a
nd
ha
lo
e
f
f
e
c
t.
T
he
r
e
gis
ter
e
d
im
a
ge
s
a
r
e
then
pa
s
s
e
d
thr
ough
the
im
a
ge
f
us
ion
block
whe
r
e
in
im
a
ge
s
a
r
e
f
us
e
d
a
s
pe
r
the
c
hos
e
n
f
u
s
ion
r
ule
a
lgor
it
hm.
P
os
t
f
us
ion,
c
omput
a
ti
ons
a
r
e
pe
r
f
or
m
e
d
de
pe
nding
on
the
de
s
ir
e
d
a
ppli
c
a
ti
on,
includin
g
im
a
ge
e
nha
nc
e
ment,
f
il
te
r
ing,
f
e
a
tur
e
e
xt
r
a
c
ti
on,
c
las
s
if
ica
ti
on,
objec
t
de
tec
ti
on,
a
nd
r
e
c
ognit
ion.
T
r
a
dit
io
na
l
wa
lk
thr
ough
de
tec
tor
s
or
ha
ndhe
ld
de
vice
s
ins
talled
a
t
the
a
ir
po
r
ts
,
publi
c
tr
a
ns
por
t
s
tations
,
a
nd
oth
e
r
publi
c
ve
nue
s
ha
ve
be
e
n
known
to
p
r
oduc
e
f
a
ls
e
pos
it
iv
e
s
,
a
nd
in
o
ther
c
a
s
e
s
ha
ve
f
a
il
e
d
to
de
tec
t
a
tt
a
c
k
e
r
s
with
we
a
pons
r
e
s
ult
ing
in
c
a
tas
tr
ophic
da
mage
.
An
a
d
a
pti
ve
f
us
ion
pr
oc
e
s
s
to
tr
a
c
k
we
a
pons
us
ing
vis
u
a
l
im
a
ge
inf
or
mation
a
long
with
de
e
p
I
R
i
mage
s
is
im
ple
mente
d
in
the
pr
opos
e
d
method
to
pe
r
c
e
ptually
e
nha
nc
e
im
a
ge
s
.
T
he
a
im
is
uti
l
izing
inf
r
a
r
e
d
s
e
ns
or
s
is
t
o
e
xploi
t
the
c
a
pa
bil
it
y
of
looki
ng
th
r
ough
f
a
br
ic
or
c
lot
h
mate
r
ial
a
nd
pr
ovide
s
uf
f
icie
nt
ge
ometr
ic
r
e
s
olu
ti
on
to
identif
y
thr
e
a
t
e
leme
nts
s
uc
h
a
s
knives
,
guns
or
c
he
mi
c
a
ls
.
I
n
thi
s
s
tudy,
the
us
e
d
im
a
ge
s
a
r
e
ge
n
e
r
a
ted
f
r
om
vis
ibl
e
li
ght
c
a
mer
a
s
a
nd
long
-
wa
ve
length
I
R
c
a
mer
a
s
F
igur
e
2.
S
hows
the
f
unc
ti
ona
l
b
lock
diag
r
a
m
o
f
I
R
c
a
mer
a
,
r
e
s
olut
ion
in
a
VL
c
a
mer
a
im
a
ge
is
higher
than
a
n
I
R
i
mage
,
but
the
c
ompos
it
e
f
us
e
d
im
a
ge
ha
s
e
xpa
nde
d
inf
or
mation
c
ontent
a
nd
high
r
e
s
olut
ion.
A
H
ybr
id
im
a
ge
f
us
ion
is
a
c
hieve
d
by
c
ombi
n
ing
the
be
s
t
f
e
a
tur
e
s
of
va
r
ious
tr
a
ns
f
or
ms
to
obtain
e
xc
e
ll
e
nt
e
ne
r
gy
c
ompac
tnes
s
,
r
e
duc
e
d
dim
e
ns
ions
a
nd
e
nha
nc
e
d
f
e
a
tur
e
s
with
mi
nim
um
a
r
ti
f
a
c
ts
.
A
s
igm
oidal
H
a
da
mar
d
wa
ve
let
tr
a
ns
f
or
m
is
a
ppli
e
d
to
mul
ti
ple
s
e
ns
o
r
im
a
ge
s
to
pr
oduc
e
a
high
-
r
e
s
olut
ion
f
us
e
d
im
a
ge
.
A
f
ur
ther
goa
l
of
the
pr
opos
e
d
method
be
yond
de
tec
ti
ng
the
c
onc
e
a
led
we
a
pon
is
the
c
las
s
if
ica
ti
on
of
the
we
a
pon
us
ing
the
K
-
mea
ns
s
uppor
t
ve
c
tor
mac
hine
c
las
s
if
ier
.
T
he
pr
opos
e
d
im
pleme
ntation
of
a
n
a
utom
a
ti
c
c
onc
e
a
led
we
a
pon
de
tec
ti
on
s
y
s
tem
r
e
quir
e
s
innovative
s
e
n
s
or
tec
hnologi
c
a
l
s
olut
ions
a
nd
e
f
f
icie
nt
mac
hine
lea
r
ning
a
lgor
it
hms
in
c
onjunction
with
i
mage
pr
oc
e
s
s
ing
a
ppli
c
a
ti
ons
.
F
igur
e
2
.
T
he
r
mal
i
mage
s
e
ns
ing
c
a
mer
a
2.
RE
L
AT
E
D
WORK
T
he
pr
oblems
o
f
im
a
ge
a
na
lys
is
us
ing
to
de
tec
ti
ng
hidden
objec
ts
unde
r
ne
a
th
c
lot
he
s
it
ha
s
e
xpos
e
d
in
many
f
ields
of
de
tec
ti
ng
domain
[
1
-
3]
.
T
he
methods
of
de
tec
ti
ng
objec
t
be
hind
c
lot
he
s
us
ing
x
-
r
a
y,
t
e
r
a
he
r
tz
a
nd
M
M
W
we
r
e
e
lucida
ted
[
4
,
5
]
.
T
he
s
it
ua
ti
ona
l
p
e
r
c
e
pti
on
a
wa
r
e
ne
s
s
pr
ovided
by
im
a
ge
f
us
ion
s
uppor
ts
the
r
e
s
e
a
r
c
he
r
s
a
powe
r
in
the
medic
a
l
f
i
e
ld,
mi
l
it
a
r
y
major
,
c
omput
e
r
s
c
ienc
e
vis
ion
f
ield
,
de
f
e
ns
e
a
nd
lot
o
f
a
nother
f
ield
[
6
-
8]
.
How
e
ve
r
,
in
our
r
e
a
l
-
ti
me
s
ys
tem,
the
c
omm
e
r
c
ial
,
c
omput
a
ti
ona
l
a
nd
t
he
type
of
e
nha
nc
e
ment
it
be
c
omes
im
por
tant
in
c
omm
e
r
c
ial
s
ur
ve
il
lanc
e
s
ys
tems
.
C
on
s
e
que
ntl
y,
mul
ti
ple
-
s
e
n
s
or
im
a
ge
f
us
ion
of
vis
ua
l
a
nd
in
f
r
a
r
e
d
im
a
ge
s
ha
s
be
e
n
tes
ted
e
xtens
ively
[
9,
10]
.
E
a
r
li
e
r
r
e
s
e
a
r
c
h
s
tu
dies
a
ls
o
e
xplor
e
d
p
s
ue
doc
olor
ing
of
X
-
r
a
y
im
a
ge
s
,
c
lut
te
r
r
e
moval
with
late
nc
y
ti
me
r
e
s
pons
e
s
,
ne
ur
a
l
n
e
twor
ks
,
M
I
M
O
-
S
A
R
b
a
s
e
d
UW
B
i
maging,
f
a
s
t
L
T
R
Ana
l
ys
is
[
11
-
13]
.
T
he
f
oc
us
wa
s
diver
ted
f
r
om
the
e
xt
e
ns
ively
us
e
d
dis
c
r
e
te
wa
ve
let
tr
a
ns
f
or
m
,
pr
incipa
l
c
ompon
e
nt
a
na
lys
is
to
the
older
W
a
ls
h
Ha
da
mar
d
im
a
ge
tr
a
ns
f
or
m
a
long
with
s
igm
oid
F
unc
ti
on
whic
h
pr
ovides
e
xc
e
ll
e
nt
e
ne
r
gy
c
ompac
ti
on.
T
his
a
bil
it
y
of
H
a
da
mar
d
t
r
a
ns
f
or
m
ha
s
e
a
r
ne
d
i
ts
e
lf
the
dis
ti
nc
ti
on
of
th
e
tr
a
ns
f
o
r
m
in
I
T
U
-
T
.
86
J
P
E
G_X
R
i
mage
c
om
pr
e
s
s
ion
s
tanda
r
d
[
14]
.
3.
M
E
T
HO
DOL
OG
Y
T
he
ne
e
d
f
or
c
onc
e
a
led
we
a
pon
de
tec
ti
on
s
ys
tems
ha
s
incr
e
a
s
e
d
tr
e
mendous
ly
due
to
a
r
is
e
in
c
r
im
e
a
nd
ter
r
o
r
is
m.
I
mage
s
c
a
ptur
e
d
f
r
om
typi
c
a
l
vis
ua
l
a
nd
inf
r
a
r
e
d
c
a
mer
a
s
ys
tems
pr
ovide
l
ow
-
c
ontr
a
s
t
im
a
ge
s
,
lac
king
unif
or
m
il
lu
mi
na
ti
on;
r
e
duc
ing
vis
ua
l
a
c
c
ur
a
c
y
[
15,
16]
.
R
e
c
e
nt
de
ve
lopm
e
nts
in
im
a
ge
f
us
ion
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
1693
-
6930
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
,
Vol.
18
,
No
.
3
,
J
une
2020:
1216
-
1223
1218
tec
hniques
r
e
s
olve
thes
e
is
s
ue
s
.
I
mage
f
us
ion
is
a
tec
hnique
whic
h
invol
ve
s
f
us
ing
c
a
ptur
e
d
vi
s
ua
l
a
nd
inf
r
a
r
e
d
im
a
ge
s
a
ll
owing
c
ontext
ua
l
e
nha
nc
e
ment,
good
im
a
ge
c
ontr
a
s
t,
a
nd
pr
ovides
a
c
c
ur
a
te
vis
ua
l
inf
or
mation,
a
nove
l
im
a
ge
f
us
ion
tec
hnique
is
pr
opos
e
d
to
e
nha
nc
e
the
vis
ua
l
inf
or
mation
of
t
he
f
us
e
d
im
a
ge
,
mi
nim
izing
r
e
dunda
nc
y
a
nd
pr
ovidi
ng
go
od
e
ne
r
gy
c
ompac
tnes
s
.
F
us
e
d
im
a
ge
s
obtai
ne
d
f
r
om
thi
s
nove
l
a
ppr
oa
c
h
e
na
ble
de
tec
ti
on
of
c
onc
e
a
led
we
a
pons
e
a
s
il
y
with
the
he
lp
o
f
c
lus
ter
ing
a
nd
c
las
s
if
ica
ti
on
tec
hniques
.
T
he
f
unc
ti
ona
l
block
diagr
a
m
of
the
pr
opos
e
d
s
ys
tem
is
s
hown
in
F
igur
e
3
.
F
igur
e
3.
F
unc
ti
ona
l
b
lock
d
iagr
a
m
of
h
ybr
id
p
r
op
os
e
d
s
ys
tem
3.
1.
P
r
op
os
e
d
m
e
t
h
od
T
r
a
ns
f
or
mation
tec
hniques
including
Ha
da
mar
d,
W
a
ve
lets
a
nd
P
C
A
methods
.
T
o
f
us
e
vis
ua
l
a
nd
inf
r
a
r
e
d
im
a
ge
s
c
a
ptur
e
d
f
r
om
thei
r
r
e
s
pe
c
ti
ve
c
a
mer
a
s
ys
tem
s
,
the
im
a
ge
s
a
r
e
s
ubjec
ted
thr
ou
gh
thr
e
e
dif
f
e
r
e
nt
tr
a
ns
f
or
mations
a
nd
a
f
us
ion
pr
oc
e
s
s
is
d
oing
to
ge
t
the
de
manding
i
mage
.
T
he
i
mage
obtai
ne
d
a
f
ter
the
f
us
ion
pr
oc
e
s
s
is
then
pr
oc
e
s
s
e
d
in
a
c
onc
e
a
led
we
a
pon
de
tec
ti
on
s
ys
tem
whic
h
de
tec
ts
the
hidden
we
a
pon
us
ing
K
-
mea
ns
c
lus
ter
ing
a
nd
S
VM
c
las
s
if
ica
ti
on
a
lgo
r
it
hms
.
T
he
pr
opos
e
d
method
is
de
s
c
r
ibed
in
the
f
o
ll
owing
s
teps
:
F
ir
s
t,
r
e
s
ize
the
r
e
gis
ter
e
d
vis
ua
l
a
nd
inf
r
a
r
e
d
im
a
ge
s
.
T
he
n
c
omput
e
the
Ha
da
mar
d
tr
a
ns
f
or
m
matr
ix
whic
h
is
us
e
d
to
tr
a
ns
f
or
m
the
i
mage
s
int
o
Ha
da
mar
d
tr
a
ns
f
or
med
i
mage
s
.
T
he
s
e
im
a
ge
s
a
r
e
then
s
c
a
led
us
ing
the
s
igm
oidal
ba
s
i
s
f
unc
ti
on.
Ne
xt
,
we
s
hould
modi
f
y
thes
e
im
a
ge
s
int
o
DW
T
2
Da
ube
c
hies
de
c
ompo
s
it
ion,
v
e
c
tor
is
e
the
de
c
ompos
e
d
a
ppr
oxim
a
te
a
nd
de
tail
c
oe
f
f
icie
nts
,
then
C
omput
e
mea
n,
c
ova
r
ianc
e
.
Af
ter
thi
s
c
omput
e
e
igenva
lues
a
nd
e
igen
pr
oc
e
s
s
ve
c
tor
s
f
or
Ha
da
mar
d
-
DW
T
,
t
he
n
f
ind
P
C
A
of
the
c
oe
f
f
icie
nts
pr
oc
e
s
s
ing
of
de
c
om
pos
ing
,
a
pply
f
us
ion
r
u
le
by
mul
t
ipl
ying
the
pr
incipa
l
c
omponents
with
the
Ha
da
mar
d
-
DW
T
de
c
ompo
s
e
d
C
oe
f
f
icie
nts
a
nd
f
us
e
them
togethe
r
,
Apply
I
nve
r
s
e
DW
T
a
nd
inver
s
e
Ha
da
mar
d
tr
a
ns
f
or
m
to
ge
t
the
f
us
e
d
im
a
ge
,
S
ubjec
t
the
F
u
s
e
d
I
mage
to
a
c
onc
e
a
led
we
a
pon
de
tec
ti
on
s
ys
tem,
p
e
r
f
or
m
c
lus
ter
ing
us
ing
K
-
mea
ns
a
lgor
it
hm
on
the
f
us
e
d
im
a
ge
,
e
xtr
a
c
t
f
e
a
tur
e
s
of
the
f
us
e
d
im
a
ge
a
nd
Us
e
S
VM
c
las
s
if
ier
to
de
tec
t
whe
ther
a
we
a
pon
is
c
onc
e
a
led
or
no
t.
3.
2.
I
m
age
f
u
s
ion
p
r
oc
e
s
s
s
igm
oid
al
Had
a
m
ar
d
wav
e
let
t
r
an
s
f
or
m
a
n
d
P
CA
A
nove
l
a
ppr
oa
c
h
whic
h
c
ombi
ne
s
the
S
igm
oidal
Ha
da
mar
d
tr
a
ns
f
or
m
a
long
with
DW
T
a
nd
P
C
A
is
us
e
d
a
s
a
n
i
mage
f
us
ion
tec
hnique
a
s
s
hown
in
F
igur
e
4
to
i
mpr
ove
the
ove
r
a
ll
pe
r
f
o
r
manc
e
.
T
h
e
e
ne
r
gy
c
ompac
tnes
s
of
the
f
us
e
d
im
a
ge
f
r
om
the
pr
op
os
e
d
a
lgor
it
hm
is
e
xc
e
ll
e
nt
a
nd
r
e
dunda
nc
y
is
mi
nim
a
l.
T
he
modi
f
ied
DW
T
c
oe
f
f
icie
nts
f
or
both
vis
ua
l
a
nd
inf
r
a
r
e
d
im
a
ge
s
obtaine
d
f
r
o
m
S
igm
oidal
H
a
da
mar
d
W
a
ve
let
tr
a
ns
f
or
m
a
r
e
f
e
d
to
the
P
C
A.
T
he
P
C
A
g
e
ne
r
a
tes
pr
incipa
l
c
omponents
to
be
us
e
d
f
or
im
a
g
e
f
us
ion.
T
he
f
us
ion
a
na
lys
is
pr
oc
e
s
s
is
doing
to
e
a
c
h
o
f
the
de
c
im
a
ted
hybr
id
Ha
da
mar
d
DW
T
c
oe
f
f
ic
ients
by
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
I
mage
pr
oc
e
s
s
ing
analys
is
of
s
igmoidal
H
adamar
d
w
av
e
let
w
it
h
P
C
A
to
de
tec
t
.
.
.
.
(
A
mm
ar
W
is
am
A
lt
ahe
r
)
1219
mul
ti
plyi
ng
them
wi
th
pr
incipa
l
c
omponents
.
Af
t
e
r
that
us
ing
the
r
e
s
ult
of
the
f
us
e
d
pr
oc
e
s
s
togethe
r
us
ing
the
f
us
ion
p
r
oc
e
s
s
to
f
o
r
m
a
f
us
e
d
im
a
ge
.
T
he
f
us
e
d
im
a
ge
obtain
e
d
f
r
om
the
s
igm
oidal
Ha
da
mar
d
w
a
ve
let
tr
a
ns
f
or
m
with
P
C
A
is
then
given
to
a
c
onc
e
a
led
we
a
pon
de
tec
ti
on
s
ys
tem
to
de
tec
t
the
hidden
we
a
pon.
F
igur
e
4
.
L
og
s
igm
oid
f
unc
ti
on
3.
3.
Had
am
ar
d
T
r
an
s
f
or
m
T
he
s
e
a
r
e
s
pe
c
ial
c
a
s
e
s
of
F
our
ier
T
r
a
ns
f
or
ms
whic
h
pe
r
f
o
r
m
s
ymm
e
t
r
ic,
li
ne
a
r
,
o
r
thogonal
a
nd
co
mpl
e
x
ope
r
a
ti
ons
.
T
he
tr
a
ns
f
or
m
be
ing
or
th
ogona
l
a
nd
non
-
s
inu
s
oidal
in
na
tur
e
e
f
f
e
c
ti
ve
ly
a
ll
ows
the
de
c
ompos
it
ion
o
f
a
s
ignal
int
o
s
e
ts
of
r
e
c
t
a
ngular
a
nd
or
thogonal
s
ignals
c
a
ll
e
d
W
a
ls
h
f
u
nc
ti
ons
.
T
he
s
e
W
a
ls
h
f
unc
ti
ons
a
r
e
made
up
of
only
two
va
lues
,
+
1
a
nd
-
1
[
17]
.
T
he
tr
a
ns
f
or
mation
is
not
we
ight
e
d
a
nd
pr
ovides
only
r
e
a
l
va
lues
.
W
HT
is
unit
a
r
y
,
r
e
a
l
with
no
mul
ti
pli
e
r
s
a
nd
is
s
uit
a
ble
f
or
a
ppli
c
a
ti
ons
li
ke
s
pe
e
c
h
pr
oc
e
s
s
ing,
medic
a
l
im
a
ge
a
na
lys
is
,
f
il
ter
i
ng,
non
-
medic
a
l
im
a
ge
pr
oc
e
s
s
ing
a
nd
c
omp
r
e
s
s
i
on
[
18]
.
T
he
Ha
da
mar
d
matr
ix
f
o
r
2x2
m
is
c
ompu
ted
by
(
1
)
.
H
m
=
1
√
2m
[
H
m
H
m
H
m
−
H
m
]
(
1)
H
m
−
Had
am
ar
d
M
at
r
i
x
,
1
√
2
−
nor
m
al
l
i
z
at
i
on
f
ac
t
or
−
,
W
he
n
m
=
1
,
the
Ha
da
mar
d
mat
r
ix
is
a
s
de
f
ined
by
1
=
1
√
2
[
1
1
1
−
1
]
(
2)
T
he
Ha
da
mar
d
mat
r
ix
,
whe
n
mul
ti
p
li
e
d
by
it
s
tr
a
ns
pos
e
′
gi
ve
s
a
n
iden
ti
t
y
mat
r
ix
a
s
de
f
ine
d
in
the
(
3)
,
′
=
(
3)
−
,
H
m
′
−
T
r
ans
pos
ed
Had
am
ar
d
M
at
r
i
x
,
−
T
he
v
is
ua
l
a
n
d
in
f
r
a
r
e
d
im
a
ge
s
a
r
e
in
di
vidua
ll
y
g
iven
to
Ha
da
mar
d
tr
a
ns
f
o
r
ma
ti
o
n.
T
he
t
r
a
ns
f
or
mati
on
o
f
im
a
ge
s
is
pe
r
f
or
med
ba
s
e
d
o
n
t
he
(
4
)
a
nd
(
5)
.
′
=
′
(
4)
,
′
=
′
(
5)
T
he
in
ve
r
s
e
t
r
a
ns
f
or
mat
ion
o
f
the
t
r
a
ns
f
or
med
vis
u
a
l
a
n
d
in
f
r
a
r
e
d
i
mage
s
a
r
e
c
a
lcula
ted
by
t
he
(
6
)
a
nd
(
7)
,
=
′
(
6)
=
′
(
7)
−
−
′
−
′
−
−
′
−
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
1693
-
6930
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
,
Vol.
18
,
No
.
3
,
J
une
2020:
1216
-
1223
1220
T
he
tr
a
ns
f
or
med
c
oe
f
f
icie
nts
pr
oduc
e
d
by
dis
c
r
e
te
wa
ve
le
ts
(
DW
T
)
a
nd
d
is
c
r
e
te
c
os
ine
tr
a
ns
f
or
m
(
DC
T
)
a
dd
mo
r
e
nois
e
to
the
input
im
a
ge
s
[
19]
.
N
ois
e
is
a
dde
d
in
the
mi
ddle
a
nd
lowe
r
f
r
e
que
nc
y
s
ub
-
ba
nds
dur
ing
de
c
ompos
it
ion
[
20
]
.
T
he
c
oe
f
f
icie
nts
ge
ne
r
a
ted
by
Ha
da
mar
d
T
r
a
ns
f
or
ms
a
r
e
s
im
il
a
r
to
thos
e
of
DW
T
a
nd
DC
T
c
oe
f
f
icie
nts
,
but
with
the
a
ddit
ion
of
les
s
nois
e
[
21]
.
T
he
Ha
da
mar
d
t
r
a
ns
f
or
m
ha
s
only
r
e
a
l
va
lues
,
is
s
ymm
e
tr
ic
a
nd
o
r
thogonal
in
na
tur
e
whic
h
he
l
ps
in
the
f
a
s
t
de
c
ompos
it
ion
of
im
a
ge
s
.
I
t
us
e
s
only
two
ope
r
a
ti
ons
-
a
ddit
ions
a
nd
s
ubtr
a
c
ti
ons
,
a
nd
it
pr
ovides
e
xc
e
ll
e
nt
e
ne
r
gy
c
ompac
tnes
s
.
S
c
a
li
ng
Ha
da
mar
d
mat
r
ix
by
s
igm
oidal
ba
s
is
f
unc
ti
ons
s
igm
oidal
ba
s
is
f
unc
ti
ons
a
r
e
tr
a
ns
f
e
r
f
unc
ti
ons
[
22
]
u
s
e
d
to
c
a
lc
ulate
the
output
of
the
laye
r
f
r
om
the
input
.
T
he
t
r
a
ns
f
e
r
f
un
c
ti
on
of
the
logar
it
hmi
c
s
igm
oidal
f
unc
ti
on
is
in
(
8
)
.
(
)
=
1
(
1
+
e
x
p
(
−
)
)
(
8)
T
he
logar
it
hmi
c
s
igm
oidal
tr
a
ns
f
e
r
f
unc
t
ion’
s
gr
a
ph
is
s
hown
in
F
igur
e
4
.
T
his
tr
a
ns
f
e
r
f
unc
ti
on
’
s
e
leme
nts
ha
ve
va
lue
s
be
twe
e
n
0
a
nd
1
only.
T
his
tr
a
ns
f
e
r
f
unc
ti
on
pr
oduc
e
s
a
non
-
tr
ivi
a
l,
r
e
a
l
-
va
lued,
unique,
c
onti
nuous
a
nd
dif
f
e
r
e
nti
a
ble
s
olut
ion.
T
he
s
c
a
li
ng
of
Ha
da
mar
d
tr
a
ns
f
or
med
im
a
ge
ge
ne
r
a
tes
a
n
e
nha
nc
e
d
vis
ua
l
pe
r
c
e
pti
on
of
the
im
a
ge
.
3.
4.
Dis
c
r
e
t
e
wave
let
t
r
an
s
f
or
m
(
DWT
)
Alf
r
e
d
Ha
a
r
,
the
Hunga
r
ian
mathe
maticia
n
wa
s
t
he
f
ir
s
t
to
invent
the
dis
c
r
e
te
wa
ve
let
tr
a
ns
f
or
m
(
DW
T
)
a
s
a
da
ta
tr
a
ns
f
or
mation
tool
.
T
he
dis
c
r
e
te
wa
ve
let
tr
a
ns
f
or
m
method
c
onve
r
ts
ti
me
domain
da
ta
to
f
r
e
que
nc
y
domain
da
ta
a
nd
divi
de
s
the
s
ignal
in
to
two
ba
nds
,
a
lowe
r
f
r
e
que
nc
y
a
nd
a
higher
f
r
e
que
nc
y
ba
nd
[
23]
.
T
he
de
c
ompos
e
d
s
ignal
now
c
ontains
f
our
ba
nds
,
low
-
low
(
L
L
)
,
l
ow
-
high
(
L
H)
,
high
-
l
ow
(
HL
)
a
nd
high
-
high
(
HH
)
f
r
e
que
nc
y
ba
nds
.
T
he
de
c
ompos
it
ion
is
ba
s
e
d
on
the
human
vi
s
ua
l
s
ys
t
e
m,
s
o
the
DW
T
de
c
ompos
e
s
the
im
a
ge
da
ta
int
o
a
ppr
oxi
mate
a
nd
de
tailed
c
oe
f
f
icie
nts
.
L
ow
-
f
r
e
que
nc
y
ba
nds
c
ontain
c
oa
r
s
e
da
ta
of
i
mage
c
ons
ti
tut
ing
the
a
ppr
oxim
a
te
c
oe
f
f
icie
nts
.
T
he
higher
f
r
e
que
nc
y
ba
nds
(
L
H,
HL
,
a
nd
HH
)
c
ontain
de
tailed
inf
or
mation
o
f
the
i
mage
a
nd
c
ons
ti
tut
e
de
tailed
c
oe
f
f
icie
nts
.
B
a
s
e
d
on
the
wa
ve
lets
us
e
d,
va
r
iants
of
the
DW
T
include
Da
ube
c
hies
,
S
yml
e
ts
,
Ha
a
r
,
C
oif
lets
,
bi
-
or
thogonal
,
a
nd
dua
l
-
tr
e
e
c
ompl
e
x
wa
ve
let
tr
a
ns
f
or
m
(
DC
W
T
)
.
W
a
ve
lets
a
r
e
s
im
il
a
r
to
wa
ve
s
with
os
c
il
lations
ha
ving
a
mpl
it
ude
a
r
ound
the
or
igi
n
(
z
e
r
o
)
.
T
he
mos
t
f
r
e
que
ntl
y
us
e
d
DW
T
va
r
iant
is
Da
ube
c
hies
wa
ve
lets
.
T
his
wa
s
f
or
m
ulate
d
in
1988
by
the
f
a
mous
B
e
lgi
a
n
mathe
maticia
n,
I
n
gr
id
Da
ube
c
hies
.
M
other
wa
ve
lets
s
how
pr
ope
r
ti
e
s
li
ke
s
ymm
e
tr
y,
o
r
thogonalit
y,
e
tc
.
DW
T
is
e
xtens
ively
us
e
d
in
im
a
ge
c
ompr
e
s
s
ion
[
24
,
25]
.
As
it
pr
ovides
e
xc
e
ll
e
nt
e
ne
r
gy
c
ompac
tnes
s
,
is
r
obus
t,
a
nd
a
c
hi
e
ve
s
gr
e
a
ter
c
omp
r
e
s
s
ion
.
T
he
DW
T
c
a
n
pe
r
f
or
m
N
leve
l
de
c
ompos
it
ion
de
pe
nding
on
the
a
ppli
c
a
ti
on
r
e
quir
e
ments
.
2D
-
DW
T
is
de
f
ined
by
(
9)
.
.
(
1
,
1
,
2
,
2
)
=
∑
∑
(
(
,
)
1
√
1
2
(
−
1
1
,
−
2
2
)
)
−
1
=
0
−
1
=
0
(
9
)
w
he
r
e
1
√
1
2
(
−
1
1
,
−
2
2
)
.
R
e
pr
e
s
e
n
ts
a
wa
v
e
let
t
ype
.
T
h
e
inve
r
s
e
2D
-
DW
T
is
de
f
ined
b
y
(
10
)
,
(
,
j
=
∑
∑
∑
∑
2
−
1
1
=
0
2
−
1
1
=
0
1
−
1
1
=
0
1
−
1
1
=
0
(
10
)
w
he
r
e
P
is
,
=
(
(
1
,
1
,
2
,
2
)
1
√
1
2
−
1
(
−
1
1
,
−
2
2
)
)
.
M
im
icking
the
human
vis
ua
l
s
ys
te
m,
DW
T
is
us
e
d
in
many
a
ppli
c
a
ti
ons
in
d
if
f
e
r
e
nt
f
ields
including
e
nginee
r
ing
,
s
c
ienc
e
,
c
omput
e
r
s
,
wir
e
les
s
c
omm
unica
ti
ons
,
mathe
matics
,
biom
e
dica
l
s
ignal
pr
oc
e
s
s
ing
a
nd
im
a
ge
pr
oc
e
s
s
ing.
3.
5.
P
r
in
c
ip
al
c
om
p
on
e
n
t
an
alys
is
(
P
CA
)
P
r
incipa
l
c
omponent
a
n
a
lys
is
(
P
C
A)
is
a
dim
e
ns
i
ona
li
ty
r
e
duc
ti
on
a
lgo
r
it
hm
invente
d
in
1901
by
the
E
ngli
s
h
mathe
maticia
n,
Ka
r
l
P
e
a
r
s
on.
I
n
th
e
1930s
,
the
Ame
r
ica
n
mathe
maticia
n
Ha
r
old
Hote
ll
ing
de
ve
loped
the
s
a
me
P
C
A
to
s
uit
mor
e
a
ppli
c
a
ti
ons
.
P
C
A
is
a
ls
o
known
a
s
the
‘
Hote
ll
ing
t
r
a
ns
f
or
m’
or
‘
Kos
a
mbi
-
Ka
r
hune
n
-
L
oe
v
e
tr
a
ns
f
or
m.
I
n
layma
n
’
s
ter
ms
,
P
C
A
f
inds
pa
tt
e
r
ns
in
the
da
ta
a
nd
e
xpr
e
s
s
e
s
the
da
ta
it
e
ms
of
it
s
dif
f
e
r
e
nc
e
s
a
nd
s
im
il
a
r
it
ies
.
T
his
f
e
a
tur
e
a
ll
ows
the
da
ta
to
be
c
ompr
e
s
s
e
d,
whic
h
r
e
s
ult
s
in
a
r
e
duc
ti
on
of
the
dim
e
ns
ion
of
da
ta
without
a
ny
los
s
of
inf
o
r
mation.
As
the
de
mand
f
o
r
d
igi
tal
da
ta
ha
s
incr
e
a
s
e
d
e
xpone
nti
a
ll
y
ove
r
r
e
c
e
nt
ye
a
r
s
,
the
ne
e
d
f
or
c
ompr
e
s
s
ion
tec
hniques
ha
s
s
o
incr
e
a
s
e
d.
P
C
A
c
a
ter
s
to
e
ve
r
incr
e
a
s
ing
de
mand
f
or
c
ompr
e
s
s
ion
a
lgor
it
hms
.
T
his
be
ing
a
s
tatis
ti
c
a
l
mul
ti
va
r
iate
a
na
lys
is
method,
it
c
a
n
be
pe
r
f
o
r
med
us
ing
e
it
he
r
s
ingul
a
r
va
lue
d
e
c
ompos
it
ion
of
da
ta
or
by
e
igenva
lue
de
c
ompos
it
ion
of
c
ova
r
ianc
e
of
da
ta.
T
his
tec
hnique
is
los
s
y
in
na
t
ur
e
.
T
he
f
ir
s
t
pr
incipa
l
c
omponent
s
a
r
e
known
to
c
ontain
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
I
mage
pr
oc
e
s
s
ing
analys
is
of
s
igmoidal
H
adamar
d
w
av
e
let
w
it
h
P
C
A
to
de
tec
t
.
.
.
.
(
A
mm
ar
W
is
am
A
lt
ahe
r
)
1221
90%
of
the
im
a
ge
inf
or
mation
.
Al
l
other
pr
incipa
l
c
omponents
c
ontain
va
r
ianc
e
in
f
or
mation.
T
he
e
mpi
r
ica
l
m
e
a
n
ve
c
tor
s
f
or
the
input
f
e
a
tur
e
s
da
tas
e
t
a
r
e
Xi
c
a
lcula
ted
by
the
(
11
)
.
M
e
a
n
̅
=
1
∑
=
1
(
11
)
T
he
e
ige
nve
c
t
or
s
a
nd
e
igenva
lues
a
r
e
c
o
mpu
ted
us
i
ng
(1
2
).
[
]
[
]
=
(1
2
)
T
he
r
e
s
u
lt
i
ng
e
ige
nve
c
to
r
s
a
r
e
s
or
ted
in
t
he
de
s
c
e
n
ding
o
r
de
r
W
.
T
he
n
the
p
r
i
nc
ipa
l
c
ompo
ne
n
ts
a
r
e
p
r
o
j
e
c
ted
onto
th
e
inpu
t
.
F
e
a
t
u
r
e
s
da
tas
e
t
by
us
ing
the
(
1
3
)
.
=
(1
3
)
Know
n
t
o
ge
ne
r
a
te
s
om
e
los
s
a
s
it
i
de
nt
if
ies
pa
tt
e
r
ns
a
nd
ba
s
e
d
on
the
m
e
xp
r
e
s
s
e
s
the
da
t
a
a
s
d
if
f
e
r
e
nc
e
s
o
r
s
im
il
a
r
pa
t
ter
ns
,
th
e
pe
r
f
o
r
manc
e
o
f
ob
jec
t
-
ide
nti
f
ic
a
ti
on
s
ys
te
ms
w
it
h
P
C
A
de
c
r
e
a
s
e
s
.
3.
6.
Conce
aled
we
ap
on
d
e
t
e
c
t
io
n
T
he
f
us
e
d
im
a
ge
obtaine
d
f
r
om
t
he
p
r
opos
e
d
S
ig
moi
da
l
Ha
da
mar
d
W
a
ve
let
with
P
C
A
tec
hnique
is
given
to
the
c
onc
e
a
led
we
a
pon
de
tec
ti
on
(
C
W
D)
s
ys
tem.
T
he
C
W
D
s
ys
tem
is
c
ompos
e
d
o
f
a
C
lus
ter
ing
block
a
nd
a
c
las
s
if
ier
block.
T
he
f
us
e
d
im
a
ge
i
s
f
ir
s
t
e
nha
nc
e
d
to
r
e
move
a
ny
nois
e
a
nd
c
lu
s
t
e
r
ing
is
pe
r
f
or
med
on
the
e
nha
nc
e
d
f
us
e
d
im
a
ge
.
K
-
mea
ns
c
lus
ter
ing
is
pe
r
f
or
med
on
the
f
us
e
d
im
a
ge
,
with
the
be
s
t
c
lus
ter
c
hos
e
n
f
r
om
s
e
ve
r
a
l
c
lus
ter
s
a
nd
the
r
e
gi
on
of
int
e
r
e
s
t
is
e
xtr
a
c
ted
f
r
om
i
t.
T
his
is
then
f
e
d
to
a
s
uppor
t
ve
c
tor
mac
hine
(
S
VM
)
c
las
s
if
ier
to
c
las
s
if
y
whe
ther
the
s
e
lec
ted
c
lus
ter
e
d
im
a
ge
c
ont
a
ins
a
we
a
pon
or
not.
4.
RE
S
UL
T
AN
D
DI
S
CU
S
I
ON
E
xpe
r
im
e
nts
we
r
e
c
onduc
ted
on
a
s
e
t
of
vis
ua
l
a
nd
inf
r
a
r
e
d
im
a
ge
da
tas
e
ts
to
tes
t
the
e
f
f
icie
nc
y
of
the
pr
opos
e
d
hybr
id
a
lgor
i
thm
ove
r
t
r
a
dit
ional
im
a
ge
f
us
ion
tec
hniques
to
de
tec
t
the
c
onc
e
a
led
we
a
pon
f
r
om
a
s
e
c
ur
e
dis
tanc
e
.
T
he
s
e
lec
ted
im
a
ge
da
tas
e
ts
[
26]
a
r
e
r
e
gis
ter
e
d
with
two
c
a
mer
a
s
viz
vis
ibl
e
li
ght
c
a
mer
a
a
nd
inf
r
a
r
e
d
c
a
mer
a
with
the
tes
t
ob
jec
t
hidden
un
de
r
plain
c
lot
hing
a
s
s
hown
in
the
s
a
mpl
e
im
a
ge
d
a
tas
e
t
in
F
igur
e
5.
T
he
di
f
f
e
r
e
nt
f
us
ion
a
ppr
oa
c
he
s
c
ompar
e
d
we
r
e
P
ixel
a
ve
r
a
ge
f
us
ion
,
P
ixel
M
a
xim
um
f
us
io
n,
DW
T
a
ve
r
a
ge
f
us
ion,
DW
T
with
P
C
A
f
us
ion
a
nd
the
pr
opos
e
d
hybr
id
f
us
ion.
T
he
il
lus
tr
a
ti
ve
c
ompar
is
on
of
dif
f
e
r
e
nt
f
us
ion
a
ppr
oa
c
he
s
a
r
e
s
hown
in
F
igur
e
6
a
nd
F
igur
e
7
.
I
t
is
vivi
dly
c
lea
r
the
f
us
e
d
im
a
ge
is
s
ha
r
pe
r
wit
h
e
nha
nc
e
d
c
ontr
a
s
t,
s
howing
a
hidden
we
a
pon.
T
hough
ther
e
is
s
ome
a
ddit
ive
nois
e
due
to
the
t
r
a
ns
f
or
mation,
thi
s
a
r
ti
f
a
c
t
doe
s
not
r
e
s
ult
in
I
mage
los
s
.
T
he
f
oll
owing
metr
ics
P
S
NR
,
M
S
E
,
e
nt
r
opy,
s
tanda
r
d
de
viation,
a
nd
mut
ua
l
inf
o
r
mation
a
r
e
mea
s
ur
e
d
to
a
ppr
e
c
iate
the
i
mage
qua
li
ty
of
the
pr
opos
e
d
im
a
ge
f
us
ion
method
ve
r
s
us
other
f
us
ion
a
l
gor
it
hms
.
T
he
pe
r
f
or
manc
e
met
r
ics
f
o
r
i
mage
s
e
ts
1
a
nd
2
a
r
e
tabula
ted
in
T
a
ble
1
a
nd
T
a
ble
2
s
hows
that
the
pr
opos
e
d
im
a
ge
f
us
ion
method
outper
f
o
r
ms
the
tr
a
dit
ional
f
us
ion
tec
hniques
.
T
he
f
us
e
d
im
a
ge
is
then
pr
oc
e
s
s
e
d
by
the
c
onc
e
a
led
we
a
pon
de
tec
ti
on
s
y
s
tem
whic
h
c
on
s
is
ts
of
a
c
lus
ter
ing
block
a
nd
a
c
las
s
if
ier
block.
T
he
f
us
e
d
im
a
ge
is
e
nha
nc
e
d
a
nd
K
-
mea
ns
c
lus
ter
ing
i
s
pe
r
f
or
med
with
the
be
s
t
c
lus
ter
pr
ovided
to
the
s
u
pe
r
vis
ing
c
las
s
if
ier
.
T
he
e
va
luation
of
the
c
las
s
if
ier
pe
r
f
o
r
m
a
nc
e
is
done
by
met
r
ics
s
uc
h
a
s
de
tec
ti
on
r
a
te
(
t
r
ue
pos
it
ive
r
a
te)
a
nd
f
a
ls
e
incide
nc
e
r
a
te.
T
he
c
las
s
if
ier
pe
r
f
or
manc
e
metr
ics
we
r
e
a
na
lyze
d
f
or
the
dif
f
e
r
e
n
t
f
us
ion
tec
hniques
a
nd
it
is
noted
that
the
c
ombi
ne
d
de
tec
ti
on
r
a
tes
a
nd
f
a
ls
e
incide
nc
e
r
a
tes
outper
f
or
m
a
ll
tr
a
dit
ional
f
us
ion
methods
.
T
he
c
las
s
if
ier
metr
ics
tabula
ted
in
T
a
ble
3
ve
r
y
c
lea
r
ly
s
howc
a
s
e
that
the
f
a
ls
e
pos
it
ive
r
a
te
of
th
e
pr
opos
e
d
s
ys
tem
is
z
e
r
o
a
nd
the
de
tec
ti
on
r
a
te
is
s
upe
r
io
r
to
tr
a
dit
ional
methods
.
(
a
)
(
b)
F
igur
e
5.
(
a
)
I
mage
da
tas
e
ts
1
(
man
s
tanding
vis
ua
l
a
nd
I
R
)
,
(
b)
I
mage
da
tas
e
ts
2
(
book
s
he
lf
vis
ua
l
a
nd
I
R
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
1693
-
6930
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
,
Vol.
18
,
No
.
3
,
J
une
2020:
1216
-
1223
1222
(
a
)
(
b
)
(
c
)
(
a
)
(
b
)
(
c
)
(
d
)
(
e
)
(
d)
(
e
)
F
igur
e
6.
I
l
lus
tr
a
ti
ve
c
ompar
is
on
of
im
a
ge
f
us
ion
tec
hniques
(
L
-
R
)
,
(
a
)
P
ixAvg,
(
b)
P
ixM
a
x,
(
c
)
DW
T
,
(
d
)
DW
T
+
P
C
A,
(
e
)
p
r
opos
e
d
method
–
im
a
ge
da
tas
e
t2
F
igur
e
7.
I
l
lus
tr
a
ti
ve
c
ompar
is
on
of
im
a
ge
f
us
ion
tec
hniques
(
L
-
R
)
,
(
a
)
P
ixAvg
,
(
b)
P
ixM
a
x,
(
c
)
DW
T
,
(
d
)
DW
T
+
P
C
A,
(
e
)
p
r
opos
e
d
method
–
im
a
ge
da
tas
e
t2
T
a
ble
1
.
I
mage
f
us
ion
p
e
r
f
o
r
manc
e
m
e
tr
ics
of
dif
f
e
r
e
nt
im
a
ge
f
us
ion
methods
f
or
im
a
ge
s
e
t1
M
e
th
odol
ogy
S
td
D
e
v
E
nt
r
opy
M
ut
ua
l
I
nf
o
M
S
E
P
S
N
R
P
ix
e
l
A
vg
2.5973e
-
16
0
0.1441
6.7454e
-
32
35.8407
P
ix
e
l
M
a
x
2.5973e
-
16
0
0.1441
6.7454e
-
32
36.7641
D
W
T
0
0.1390
0.1441
0.01
68.1308
D
W
T
+
P
C
A
0.3687
1.1867
-
1.0426
0.1702
55.8220
H
a
da
ma
r
d+
D
W
T
+
P
C
A
0.0664
0.0429
0.1592
0.0050
71.1406
T
a
ble
2
.
I
mage
f
us
ion
p
e
r
f
o
r
manc
e
m
e
tr
ics
of
dif
f
e
r
e
nt
im
a
ge
f
us
ion
methods
f
or
i
mage
s
e
t2
M
e
th
odol
ogy
S
td
D
e
v
E
nt
r
opy
M
ut
ua
l
I
nf
o
M
S
E
P
S
N
R
P
ix
e
l
A
vg
2.1635e
-
16
0
0.0311
6.7454e
-
32
37.1407
P
ix
e
l
M
a
x
2.1635e
-
16
0
0.0311
6.7454e
-
32
38.7194
D
W
T
0
0.0764
0.0311
0.02
65.1205
D
W
T
+
P
C
A
0.2443
0.0871
-
0.056
0.1020
58.0448
H
a
da
ma
r
d+
D
W
T
+
P
C
A
0.0664
0.0410
0.0980
0.0041
72.003
T
a
ble
3
.
Obje
c
t
de
tec
ti
on
c
las
s
if
ier
m
e
t
r
ics
f
or
i
ma
ge
s
e
t1
M
e
th
odol
ogy
S
td
D
e
v
E
nt
r
opy
M
ut
ua
l
I
nf
o
M
S
E
P
S
N
R
P
ix
e
l
A
vg
2.1635e
-
16
0
0.0311
6.7454e
-
32
37.1407
P
ix
e
l
M
a
x
2.1635e
-
16
0
0.0311
6.7454e
-
32
38.7194
D
W
T
0
0.0764
0.0311
0.02
65.1205
D
W
T
+
P
C
A
0.2443
0.0871
-
0.056
0.1020
58.0448
H
a
da
ma
r
d+
D
W
T
+
P
C
A
0.0664
0.0410
0.0980
0.0041
72.003
5.
CONC
L
USI
ON
T
his
pa
pe
r
c
onc
ludes
that
the
pr
opos
e
d
hybr
id
im
a
ge
f
us
ion
a
lgor
it
hm
ba
s
e
d
on
a
c
ombi
na
ti
on
of
s
igm
oidal
Ha
da
mar
d
wa
ve
let
t
r
a
ns
f
or
ms
a
nd
P
C
A
with
mo
r
phologi
c
a
l
p
r
oc
e
s
s
ing
will
he
a
vil
y
im
pr
ove
the
im
a
ge
f
us
ion
qua
li
ty
a
nd
c
a
n
be
the
f
utu
r
e
lo
de
s
tar
of
r
e
s
e
a
r
c
h
a
nd
de
ve
lopm
e
nt
in
da
ta
f
us
io
n
f
ields
.
P
r
e
s
e
nt
c
omm
e
r
c
ially
a
va
il
a
ble
e
a
r
ly
we
a
pon
de
te
c
ti
on
s
ys
tems
r
e
quir
e
manua
l
s
c
r
e
e
ning,
but
the
p
r
opos
e
d
c
onc
e
a
led
we
a
pon
de
tec
ti
on
s
ys
tem
c
a
n
be
ope
r
a
ted
a
c
c
ur
a
tely
a
t
a
s
e
c
ur
e
dis
tanc
e
.
S
ubjec
ti
v
e
vis
ua
l
a
na
lys
is
of
output
im
a
ge
s
f
r
om
t
r
a
dit
ional
f
us
ion
a
ppr
oa
c
he
s
a
nd
the
im
a
ge
qua
li
ty
metr
ics
c
onc
l
ude
that
the
pr
opos
e
d
f
us
ion
a
lgor
it
hm
outper
f
o
r
ms
e
xis
ti
ng
a
lgor
it
hms
.
T
he
methodology
is
not
c
omput
a
ti
ona
ll
y
c
ompl
e
x
or
e
xpe
ns
ive;
he
nc
e
the
hybr
id
im
a
ge
f
us
ion
th
e
a
ppr
oa
c
h
c
a
n
be
us
e
d
in
a
ppli
c
a
ti
ons
whe
r
e
im
a
ge
da
ta
c
ompr
e
s
s
ion,
high
qua
li
ty
,
a
nd
p
r
e
c
is
ion
a
r
e
not
of
pa
r
a
moun
t
im
po
r
tanc
e
.
T
he
pr
opos
e
d
s
ys
tem
c
a
n
be
c
ombi
ne
d
with
human
e
mot
ional
e
xpr
e
s
s
ion,
be
h
a
vior
a
nd
ga
it
a
na
lys
is
a
s
the
s
us
pe
c
t
c
a
n
ha
ve
a
dyna
mi
c
r
a
nge
of
c
a
pr
icious
bio
logi
c
a
l
indi
c
a
tor
s
.
AC
KNOWL
E
DGE
M
E
NT
S
T
his
wor
k
wa
s
s
uppor
ted
in
pa
r
t
by
the
De
pa
r
tm
e
nt
of
I
nf
o
r
mation
tec
hnology,
Al
-
F
ur
a
t
Al
-
Aw
s
a
t
T
e
c
hnica
l
Unive
r
s
it
y,
T
e
c
hnica
l
C
oll
a
ge
of
M
a
na
ge
ment,
I
r
a
q.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
I
mage
pr
oc
e
s
s
ing
analys
is
of
s
igmoidal
H
adamar
d
w
av
e
let
w
it
h
P
C
A
to
de
tec
t
.
.
.
.
(
A
mm
ar
W
is
am
A
lt
ahe
r
)
1223
RE
F
E
RE
NC
E
S
[1
]
A
.
A
g
u
rt
o
,
et
a
l
,
“
A
rev
i
ew
o
f
co
n
cea
l
ed
w
eap
o
n
d
et
ec
t
i
o
n
a
n
d
re
s
earch
i
n
p
ers
p
ect
i
v
e
,”
2
0
0
7
I
E
E
E
In
t
er
n
a
t
i
o
n
a
l
Co
n
f
er
e
n
ce
o
n
Ne
t
wo
r
ki
n
g
,
S
e
n
s
i
n
g
a
n
d
Co
n
t
r
o
l
,
p
p
.
4
4
3
-
4
4
8
,
2
0
0
7
.
[2
]
Z
.
X
u
e,
et
al
.
,
“
Fu
s
i
o
n
o
f
v
i
s
u
al
an
d
IR
i
mag
e
s
fo
r
co
n
ceal
ed
w
ea
p
o
n
d
et
ec
t
i
o
n
,”
P
r
o
ced
i
n
g
s
o
f
t
h
e
F
i
f
t
h
In
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
er
e
n
ce
o
n
In
f
o
r
m
a
t
i
o
n
F
u
s
i
o
n
,
v
ol
.
2
,
p
p
.
1
1
9
8
-
1
2
0
5
,
2
0
0
2
.
[3
]
S.
Y
eo
m,
et
al
.
,
“
Real
-
t
i
me
o
u
t
d
o
o
r
co
n
cea
l
ed
-
o
b
j
ect
d
et
ec
t
i
o
n
w
i
t
h
p
as
s
i
v
e
mi
l
l
i
me
t
er
w
a
v
e
i
mag
i
n
g
,”
O
p
t
i
c
s
exp
r
e
s
s
,
v
o
l
.
19
,
n
o
.
3
,
p
p
.
2
5
3
0
-
2
5
3
6
,
2
0
1
1
.
[4
]
X
.
Sh
en
,
et
al
.
,
“
D
et
ect
i
o
n
an
d
s
eg
me
n
t
a
t
i
o
n
o
f
co
n
cea
l
e
d
o
b
j
ec
t
s
i
n
t
erah
er
t
z
i
mag
e
s
,”
IE
E
E
t
r
a
n
s
a
c
t
i
o
n
s
o
n
Im
a
g
e
P
r
o
ces
s
i
n
g
,
v
o
l
.
17
,
n
o
.
12
,
p
p
.
2
4
6
5
-
2
4
7
5
,
D
ec
2
0
0
8
.
[5
]
L
.
T
ao
,
et
a
l
.
,
“
A
m
u
l
t
i
s
en
s
o
r
i
mag
e
f
u
s
i
o
n
a
n
d
e
n
h
an
ceme
n
t
s
y
s
t
em
f
o
r
a
s
s
i
s
t
i
n
g
d
ri
v
ers
i
n
p
o
o
r
l
i
g
h
t
i
n
g
co
n
d
i
t
i
o
ns
,”
34
th
A
p
p
l
i
ed
I
m
a
g
er
y
a
n
d
P
a
t
t
e
r
n
R
eco
g
n
i
t
i
o
n
W
o
r
k
s
h
o
p
(
A
IP
R
'
0
5
)
,
p
p
.
6
,
1
1
3
,
2
0
0
5
.
[6
]
J
.
T
a
n
g
,
e
t
al
.
,
“
A
d
i
rect
i
ma
g
e
co
n
t
ra
s
t
en
h
an
ce
men
t
a
l
g
o
ri
t
h
m
i
n
t
h
e
w
av
e
l
et
d
o
ma
i
n
fo
r
s
creen
i
n
g
mammo
g
ram
s
,”
IE
E
E
Jo
u
r
n
a
l
o
f
S
e
l
ect
e
d
To
p
i
c
s
i
n
S
i
g
n
a
l
P
r
o
ces
s
i
n
g
,
v
o
l
.
3
,
n
o
.
1
, p
p
.
7
4
-
80
,
Feb
2
0
0
9
.
[7
]
Y
.
Z
h
en
g
a
n
d
P.
Z
h
en
g
,
“
Mu
l
t
i
s
en
s
o
r
i
mag
e
f
u
s
i
o
n
u
s
i
n
g
fu
zz
y
l
o
g
i
c
fo
r
s
u
r
v
ei
l
l
a
n
ce
s
y
s
t
ems
,”
2
0
1
0
S
eve
n
t
h
In
t
e
r
n
a
t
i
o
n
a
l
Co
n
f
er
e
n
ce
o
n
F
u
z
z
y
S
ys
t
em
s
a
n
d
Kn
o
wl
e
d
g
e
D
i
s
c
o
ver
y
(
F
S
K
D
)
,
p
p
.
5
8
8
-
5
9
2
,
2
0
1
0
.
[8
]
H
.
Wu
an
d
Y
.
X
i
n
g
,
“
Pi
x
e
l
-
b
a
s
ed
i
ma
g
e
fu
s
i
o
n
u
s
i
n
g
w
av
el
e
t
t
ran
s
fo
rm
fo
r
SPO
T
an
d
E
T
M+
i
mag
e
,”
2
0
1
0
IE
E
E
In
t
e
r
n
a
t
i
o
n
a
l
Co
n
f
er
e
n
ce
o
n
P
r
o
g
r
es
s
i
n
In
f
o
r
m
a
t
i
c
s
a
n
d
Co
m
p
u
t
i
n
g
(
P
IC)
,
p
p
.
9
3
6
-
9
4
0
,
2
0
1
0
.
[9
]
Y
.
Z
h
o
u
an
d
M.
O
mar,
“
Pi
x
el
-
L
ev
e
l
fu
s
i
o
n
fo
r
i
n
fr
ared
an
d
v
i
s
i
b
l
e
acq
u
i
s
i
t
i
o
n
s
,”
In
t
er
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
O
p
t
o
m
ech
a
t
r
o
n
i
c
s
,
v
o
l
.
3
n
o
.
1
,
p
p
.
4
1
-
53
,
2
0
0
9
.
[1
0
]
A
.
Mi
s
h
ra
an
d
S.
Rak
s
h
i
t
,
“
Fu
s
i
o
n
o
f
n
o
i
s
y
mu
l
t
i
-
s
e
n
s
o
r
i
mag
er
y
,”
D
e
f
en
ce
S
ci
e
n
ce
Jo
u
r
n
a
l
,
v
o
l
.
58
,
n
o
.
1
,
p
p
.
1
3
6
-
1
4
6
,
J
an
2
0
0
8
.
[1
1
]
J
.
J
.
McCo
mb
e,
et
al
.
,
“
Cl
u
t
t
er
remo
v
a
l
i
n
t
h
e
au
t
o
mat
i
c
d
et
ect
i
o
n
o
f
co
n
ceal
e
d
w
eap
o
n
s
w
i
t
h
l
a
t
e
t
i
me
res
p
o
n
s
e
s
,”
2
0
1
3
E
u
r
o
p
e
a
n
I
E
E
E
R
a
d
a
r
Co
n
f
e
r
en
ce
(E
u
R
A
D
),
p
p
.
5
3
-
5
6
,
2
0
1
3
.
[1
2
]
D
.
O
'Rei
l
l
y
,
et
a
l
.
,
“
Si
g
n
a
l
p
ro
ces
s
i
n
g
t
ec
h
n
i
q
u
es
fo
r
co
n
cea
l
ed
w
ea
p
o
n
d
e
t
ect
i
o
n
b
y
u
s
e
o
f
n
eu
ra
l
n
et
w
o
r
k
s
,”
IE
E
E
2
7
th
Co
n
ven
t
i
o
n
o
f
E
l
ect
r
i
c
a
l
&
E
l
ect
r
o
n
i
c
s
E
n
g
i
n
eer
s
i
n
Is
r
a
e
l
(IE
E
E
I
)
,
p
p
.
1
-
4
,
N
o
v
2
0
1
2
.
[1
3
]
X
.
Z
h
u
g
e
a
n
d
A
.
Y
aro
v
o
y
,
“
MIMO
-
SA
R
b
as
e
d
U
W
B
i
ma
g
i
n
g
f
o
r
co
n
ceal
e
d
w
ea
p
o
n
d
e
t
ect
i
o
n
,”
2
0
1
0
8
th
E
u
r
o
p
e
a
n
C
o
n
f
er
e
n
ce
o
n
S
yn
t
h
e
t
i
c
A
p
e
r
t
u
r
e
R
a
d
a
r
(
E
U
S
A
R
)
,
p
p
.
1
-
4
,
J
an
2
0
1
0
.
[1
4
]
V
.
Men
o
n
,
et
a
l
.
,
“
Fas
t
SV
D
w
i
t
h
Ran
d
o
m
H
a
d
amard
Pro
j
ec
t
i
o
n
f
o
r
H
y
p
er
s
p
e
ct
ra
l
D
i
men
s
i
o
n
a
l
i
t
y
Re
d
u
c
t
i
o
n
,”
IE
E
E
G
eo
s
ci
e
n
ce
a
n
d
R
e
m
o
t
e
S
e
n
s
i
n
g
Let
t
er
s
,
v
o
l
.
13
,
n
o
.
9
,
p
p
.
1
2
7
5
-
1
2
7
9
,
Sep
2
0
1
6
.
[1
5
]
M.
K
.
Peh
l
i̇
v
an
o
ğ
l
u
a
n
d
N
.
D
u
r
u
,
“
E
n
cry
p
t
i
o
n
o
f
W
al
s
h
H
ad
am
ard
T
r
an
s
fo
rm
ap
p
l
i
ed
i
mag
e
s
w
i
t
h
t
h
e
A
E
S
en
cry
p
t
i
o
n
al
g
o
r
i
t
h
m
,
”
24
th
S
i
g
n
a
l
P
r
o
ce
s
s
i
n
g
a
n
d
C
o
m
m
u
n
i
ca
t
i
o
n
A
p
p
l
i
ca
t
i
o
n
C
o
n
f
er
e
n
ce
(S
IU
)
,
p
p
.
3
0
1
-
3
0
4
,
2
0
1
6
.
[1
6
]
T
.
Xu
an
d
Q
.
M.
J
.
W
u
,
“
Mu
l
t
i
s
en
s
o
r
c
o
n
cea
l
ed
w
eap
o
n
d
et
ec
t
i
o
n
u
s
i
n
g
t
h
e
i
ma
g
e
fu
s
i
o
n
ap
p
r
o
a
ch
,”
6
th
In
t
er
n
a
t
i
o
n
a
l
C
o
n
f
er
e
n
ce
o
n
Im
a
g
i
n
g
f
o
r
Cr
i
m
e
P
r
ev
en
t
i
o
n
a
n
d
D
e
t
ect
i
o
n
(ICD
P
-
15)
,
p
p
.
1
-
7
,
2
0
1
5
.
[1
7
]
E
.
M.
U
p
ad
h
y
a
y
an
d
N
.
K
.
Ran
a
,
“
E
x
p
o
s
u
re
fu
s
i
o
n
fo
r
co
n
cea
l
ed
w
ea
p
o
n
d
et
ec
t
i
o
n
,”
2
0
1
4
2
nd
In
t
er
n
a
t
i
o
n
a
l
Co
n
f
er
e
n
ce
o
n
I
E
E
E
D
evi
ce
s
,
Ci
r
c
u
i
t
s
a
n
d
S
ys
t
em
s
(IC
D
C
S)
,
p
p
.
1
-
6
,
2
0
1
4
.
[1
8
]
Fars
i
H
.
an
d
Mo
h
ama
d
zad
e
h
S.
,
“
Co
mb
i
n
i
n
g
H
a
d
amard
mat
ri
x
,
d
i
s
cre
t
e
w
av
e
l
et
t
ra
n
s
f
o
rm
an
d
D
CT
feat
u
res
b
a
s
ed
o
n
PC
A
an
d
K
N
N
fo
r
i
mag
e
re
t
ri
e
v
a
l
,”
M
a
j
l
e
s
i
Jo
u
r
n
a
l
o
f
E
l
ec
t
r
i
ca
l
E
n
g
i
n
eer
i
n
g
,
v
o
l
.
7
,
n
o
.
1
,
p
p
.
9
-
1
5
,
Mar
201
3
.
[1
9
]
H
.
B.
K
ek
re,
et
al
.
,
“
Imag
e
fu
s
i
o
n
u
s
i
n
g
K
ek
re'
s
h
y
b
ri
d
w
av
el
e
t
t
ran
s
f
o
rm
,”
in
In
t
er
n
a
t
i
o
n
a
l
Co
n
f
e
r
en
ce
o
n
IE
E
E
Co
m
m
u
n
i
c
a
t
i
o
n
,
In
f
o
r
m
a
t
i
o
n
&
Co
m
p
u
t
i
n
g
Tec
h
n
o
l
o
g
y
(ICCICT)
,
p
p
.
1
-
6,
2
0
1
2
.
[2
0
]
H.
B.
K
ek
re,
et
al
.
,
“
Imag
e
ret
ri
ev
al
u
s
i
n
g
t
ex
t
u
re
p
a
t
t
ern
s
g
e
n
erat
e
d
fro
m
w
al
s
h
-
h
ad
amar
d
t
ran
s
fo
rm
mat
r
i
x
an
d
i
mag
e
b
i
t
ma
p
s
,
”
Tech
n
o
l
o
g
y
S
y
s
t
e
m
s
a
n
d
M
a
n
a
g
em
e
n
t
,
p
p
.
9
9
-
1
0
6
,
2
0
1
1
.
[2
1
]
V
.
San
t
h
i
an
d
P.
A
ru
l
m
o
zh
i
v
arma
n
,
“
H
ad
amar
d
t
ran
s
f
o
rm
b
as
e
d
ad
ap
t
i
v
e
v
i
s
i
b
l
e/
i
n
v
i
s
i
b
l
e
w
a
t
erma
rk
i
n
g
s
ch
eme
fo
r
d
i
g
i
t
al
i
mag
e
s
,”
Jo
u
r
n
a
l
o
f
In
f
o
r
m
a
t
i
o
n
S
ecu
r
i
t
y
a
n
d
A
p
p
l
i
ca
t
i
o
n
s
,
v
o
l
.
18
,
n
o
.
4
,
p
p
.
1
6
7
-
1
7
9
,
D
ec
2
0
1
3
.
[2
2
]
U
.
Q
i
d
w
a
i
an
d
C.
H
.
Ch
e
n
,
“
D
i
g
i
t
al
i
ma
g
e
p
r
o
ces
s
i
n
g
:
an
a
l
g
o
ri
t
h
m
i
c
ap
p
ro
ac
h
w
i
t
h
M
A
T
L
A
B
,”
CR
C
p
r
es
s
,
20
17
.
[2
3
]
H
.
H
.
K
an
g
,
et
a
l
.
,
“
Co
mp
res
s
i
o
n
s
ch
eme
o
f
s
u
b
-
i
ma
g
es
u
s
i
n
g
K
arh
u
n
e
n
-
L
o
ev
e
t
ran
s
fo
rm
i
n
t
h
r
ee
-
d
i
me
n
s
i
o
n
a
l
i
n
t
eg
ra
l
i
mag
i
n
g
,”
O
p
t
i
c
s
Co
m
m
u
n
i
ca
t
i
o
n
s
,
v
o
l
.
2
8
1
,
n
o
.
14
,
p
p
.
3
6
4
0
-
3
6
4
7
,
J
u
l
2
0
0
8
.
[2
4
]
Y
.
W
an
g
an
d
M.
L
u
,
"
Imag
e
Fu
s
i
o
n
Bas
e
d
Co
n
ceal
e
d
W
eap
o
n
D
et
ec
t
i
o
n
,
"
2
0
0
9
In
t
er
n
a
t
i
o
n
a
l
Co
n
fere
n
ce
o
n
Co
mp
u
t
a
t
i
o
n
a
l
In
t
e
l
l
i
g
e
n
ce
an
d
So
ft
w
are
E
n
g
i
n
eer
i
n
g
,
W
u
h
an
,
2
0
0
9
,
p
p
.
1
-
4.
d
o
i
:
1
0
.
1
1
0
9
/
CIS
E
.
2
0
0
9
.
5
3
6
4
5
8
1
[2
5
]
K
.
Fi
rd
a
u
s
y
,
et
al
.
,
“
Imag
e
E
n
h
an
ceme
n
t
u
s
i
n
g
C
o
n
t
ras
t
St
ret
c
h
i
n
g
o
n
RG
B
an
d
IH
S
D
i
g
i
t
al
Ima
g
e
,”
TE
LKO
M
NIK
A
Tel
ec
o
m
m
u
n
i
c
a
t
i
o
n
Co
m
p
u
t
i
n
g
E
l
ec
t
r
o
n
i
c
s
a
n
d
Co
n
t
r
o
l
,
v
o
l
.
5
,
n
o
.
1
,
p
p
.
45
-
50
,
A
p
r
2
0
0
7
.
[2
6
]
R.
Mai
n
i
a
n
d
H
.
A
g
g
ar
w
al
,
“
A
C
o
mp
re
h
e
n
s
i
v
e
R
ev
i
e
w
o
f
Ima
g
e
E
n
h
a
n
cemen
t
T
ec
h
n
i
q
u
es
,”
J
o
u
r
n
a
l
o
f
Co
m
p
u
t
i
n
g
,
v
o
l
.
2
,
n
o
.
3
,
p
p
.
8
-
1
3
,
Mar
2
0
1
0
.
Evaluation Warning : The document was created with Spire.PDF for Python.