Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
n
g (IJ
E
C
E)
Vo
l.
11
,
No.
1
,
Febr
uar
y
2021
, pp.
872
~
878
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v11
i
1
.
pp
872
-
878
872
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om
Developi
ng digit
al signal
clu
ster
in
g method
usin
g
l
oc
al bina
ry
pattern
histogr
am
Ra
s
had
J.
R
asr
as
1
,
Bi
lal Z
ahran
2
,
M
utaz
Rasmi
Ab
u
S
ar
a
3
,
Z
iad
AlQadi
4
1,
2,4
Depa
rtment
o
f
Com
pute
r Engi
nee
ring
,
Al
-
B
al
q
a
Appli
ed
Univ
e
rsit
y
,
Jordan
3
Depa
rtment of
Com
pute
r
Sc
ie
n
ce
,
T
ai
bah
Univ
ersity
,
Saudi
Ara
bia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Feb
25, 202
0
Re
vised
A
ug 9, 20
20
Accepte
d
Aug 17
, 202
0
In
thi
s
pape
r
we
pre
sente
d
a
n
e
w
appr
oac
h
to
m
ani
pula
t
e
a
di
git
al
sign
al
in
orde
r
to
cre
a
te
a
fea
ture
s
arr
a
y
,
which
ca
n
be
used
as
a
signat
ure
to
ret
riev
e
the
signal.
Ea
ch
digi
ta
l
signa
l
i
s
associa
te
d
wit
h
the
loc
al
bin
a
r
y
patter
n
(LBP)
histogra
m
;
thi
s
histogra
m
will
be
ca
lc
ul
at
ed
base
d
on
L
BP
oper
at
or
,
the
n
k
-
m
ea
ns
clus
te
ring
was
used
to
gene
rate
th
e
req
uire
d
f
e
at
ur
es
for
ea
ch
digi
tal
signa
l.
The
propose
d
m
et
hod
was
implemente
d
,
te
sted
and
the
obta
in
ed
e
xper
imental
result
s
were
ana
l
y
z
ed.
Th
e
result
s
show
ed
the
fle
x
ibi
l
ity
a
nd
ac
cur
a
c
y
of
the
proposed
m
et
hod.
Althou
g
diffe
ren
t
par
amete
rs
of
t
he
digi
t
al
sign
al
w
ere
ch
anged
during
implementa
ti
on
,
the
r
esult
s obt
ained
show
ed the
r
obustness of
the
proposed
m
et
ho
d.
Ke
yw
or
d
s
:
Feat
ur
e
ex
tr
act
ion
K
-
m
eans clust
erin
g
Local
bin
a
ry pa
tt
ern
(LB
P)
This
is an
open
acc
ess arti
cl
e
un
der
the
CC
B
Y
-
SA
l
ic
ense
.
Corres
pond
in
g
Aut
h
or
:
Ra
sh
ad
J. R
asr
as
,
Dep
a
rtm
ent o
f C
om
pu
te
r
E
ng
i
neer
i
ng,
Al
-
Ba
lqa
A
pp
l
ie
d
U
niv
e
rsity
,
Amm
an,
J
orda
n
.
Em
a
il
:
rash
ad.r
asras@
ba
u.
e
du.jo
1.
INTROD
U
CTION
Digital
sign
al
s
su
ch
a
s
di
gital
color
im
ages
and
dig
it
al
wav
e
si
gnal
s
are
us
e
d
us
ual
ly
in
var
io
us
com
pu
te
r
a
pp
li
cat
ion
s
s
uc
h
as
com
pu
te
r
sec
uri
ty
and
o
t
her
s
.
Be
cause o
f
th
e
big
siz
e
of
t
he
wa
ve
file
it
is
ver
y
diff
ic
ult
to
us
e
thr
w
hole
file
f
or
retriev
al
or
rec
ogniti
on
pur
poses;
here
t
he
im
po
rtan
ce
of
e
xtracti
ng
file
featur
e
s
a
pp
ea
rs
[1
]
.
Digital
wav
e
sig
nal
usual
ly
represe
nted
by
m
on
o
or
ste
re
o.
Mo
no
desc
ribes
a
s
yst
e
m
wh
e
re
al
l
the
aud
i
o
sig
nals
are
m
ixed
toge
ther
a
nd
r
ou
t
ed
th
rou
gh
a
sing
le
a
ud
i
o
c
hannel.
Ste
reo
sound
syst
e
m
s
hav
e
tw
o
i
nd
e
pe
nd
e
nt
aud
io
c
ha
nn
e
ls,
an
d
the
sig
nals
are
reprod
uced
by
tw
o
c
hannels
s
e
par
a
te
d
by
so
m
e
distance
[2
]
.
T
he
am
plitu
de
val
ues
of
each
c
olu
m
n
are
ra
nges
f
r
om
-
1
t
o
+
1
an
d
t
hey
are
t
he
res
ults
of
sam
pling
and
quanti
zat
io
n of
t
he vo
ic
e
sig
nal. F
ig
ure
1
s
h
ows s
o
m
e sa
m
ples of a
giv
e
n w
ave
file
.
Wh
il
e
Fi
gure
s
2
a
nd
3
sh
ow
t
he
wav
e
of
the
voic
e
sig
nal
.
On
e
of
the
m
os
t
us
e
d
a
pp
li
ca
ti
on
s
relat
ed
to
di
gital
wav
e
sign
al
s
proce
s
sing
is
voic
e
re
trie
val
an
d
rec
ogniti
on.
Mo
st
of
these
a
ppli
cat
ion
s
us
e
t
he
natu
re
of
th
e
di
gital
wav
e
file
to
ge
ner
at
e
so
m
e
featur
es
f
or
the
f
il
e
by
m
ean
of
cal
culat
ing
som
e
par
am
e
te
rs
su
ch
a
s
crest
facto
r,
dy
nam
ic
ran
ge,
m
ean
of
t
he
norm
al
iz
ed
data
(sigm
a),
an
d
s
ta
nd
a
rd
de
viati
on
of
the
norm
al
iz
ed
data
(Mu
),
t
he
se
par
am
et
ers
can
be
easi
ly
c
al
culat
ed
an
d
us
e
d
as
a
fea
tures
for
di
gital
wav
e
si
gn
al
[1
-
3]
.
Ca
lc
ulati
ng
th
ese
sta
ti
sti
cal
par
am
et
ers
requires
unde
rsta
nd
i
ng
di
gital
vo
ic
e
cha
racteri
sti
cs
and
natu
r
e,
an
d
so
m
e
tim
e
the
y
do
no
t
giv
e
a
n
acce
pta
ble
re
cogniti
on
rati
o
if
we
us
e
the
m
to
recog
nize
the
voic
e
eve
n
if
they
giv
e
sta
ble
a
nd
fixe
d
feat
ur
e
s
f
or
each
wa
ve
file
.
Th
ese
featur
e
s
will
re
m
ai
n
the
sam
e
eve
n
if
we
c
hang
e
sam
pling
f
re
qu
ency, am
plit
ude or
ph
ase
s
hifting
a
s s
how
n
i
n
Ta
bles
1
a
nd
2.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
o
m
p
En
g
IS
S
N: 20
88
-
8708
Develo
ping
digi
tal signal cl
us
te
ring
meth
od
us
in
g
l
oca
l
binary
patt
ern
his
tog
r
am
(
Rash
ad J.
R
as
r
as
)
873
Figure
1
.
Sam
ples o
f
a stere
o wav
e
f
il
e
Figure
2.
V
oic
e w
a
ve
si
gn
al
i
n
ti
m
e d
om
ai
n
Figure
3
.
V
oic
e w
a
ve
si
gn
al
i
n fr
e
qu
e
ncy
dom
ai
n
Table
1.
Stat
ist
ic
al
f
eat
ur
es
W
av
e f
ile
Featu
res
Sig
m
a
Mu
Peak
(
cr
est) f
acto
r
(dB
)
Dy
n
a
m
i
c r
an
g
e (
d
B)
Co
w
0
.39
5
3
3
-
0
.00
0
2
6
7
8
6
8
.06
0
9
8
4
.28
8
1
Do
g
0
.16
5
6
-
6
.59
7
e
-
005
1
5
.61
8
8
4
9
.89
0
6
Du
ck
0
.07
6
1
1
8
0
.00
0
2
3
2
7
6
2
2
.37
0
3
4
9
.07
2
4
Do
lp
h
in
0
.17
1
9
7
-
0
.00
1
6
6
1
5
.29
0
5
4
6
.22
9
Ho
rse
0
.41
3
7
9
0
.00
2
6
4
2
6
7
.66
4
5
4
2
.14
3
9
Table
2.
Stat
ist
ic
al
f
eat
ur
es
fo
r
the
sam
e file
w
it
h varyin
g
s
a
m
pling
fr
e
que
ncy
Co
w wave f
ile
sa
m
p
l
in
g
f
requ
en
cy
Featu
res
Sig
m
a
Mu
Peak
(
cr
est) f
acto
r
(dB
)
Dy
n
a
m
i
c r
an
g
e (
d
B)
1
1
0
2
5
0
.39
5
3
3
-
0
.00
0
2
6
7
8
6
8
.06
0
9
8
4
.28
8
1
1
5
0
0
0
0
.39
5
3
3
-
0
.00
0
2
6
7
8
6
8
.06
0
9
8
4
.28
8
1
2
0
0
0
0
0
.39
5
3
3
-
0
.00
0
2
6
7
8
6
8
.06
0
9
8
4
.28
8
1
1000
0
.39
5
3
3
-
0
.00
0
2
6
7
8
6
8
.06
0
9
8
4
.28
8
1
2000
0
.39
5
3
3
-
0
.00
0
2
6
7
8
6
8
.06
0
9
8
4
.28
8
1
To
overc
om
e
t
he
a
bove
m
ent
ion
e
d
disad
va
nt
ages,
we
can
extract
th
e
vo
i
ce
sig
nal
feat
ures
base
d
on
local
bin
a
ry
pa
tt
ern
(LBP
).
H
ere
we
ca
n
cal
culat
e
LBP
his
togram
to
be
use
d
as
a
n
in
pu
t
data
set
to
ge
ner
at
e
the
dig
it
al
file
featu
res
.
LB
P
and
it
s
va
riants
su
c
h
as
com
plete
d
no
i
se
-
in
var
ia
nt
local
-
struct
ur
e
pa
tt
ern
(CNLP)
[
4],
a
nd
dom
inant
L
BP
(
DLBP
)
[5]
has
bee
n
favor
a
bly
ap
plied
to
a
wi
de
vari
et
y
of
a
ppli
cat
ion
s
,
su
c
h
as
te
xtur
e
cl
assifi
cat
ion
[6
-
13]
,
face
a
naly
sis
[14
-
16
]
,
sp
eech
r
eco
gn
it
io
n
[
9,
10
]
and
oth
e
rs
[
17,
19
]
.
T
h
e
L
B
P
e
n
c
o
d
e
s
t
h
e
c
o
-
o
c
c
u
r
r
e
n
c
e
o
f
n
e
i
g
h
b
o
r
i
n
g
p
i
x
e
l
c
o
m
p
a
r
i
s
o
n
s
w
i
t
h
i
n
a
l
o
c
a
l
a
r
e
a
.
I
t
i
s
c
o
m
p
u
t
a
t
i
o
n
a
l
l
y
e
f
f
i
c
i
e
n
t
,
s
i
m
p
l
e
,
a
n
d
r
o
b
u
s
t
a
g
a
i
n
s
t
s
o
m
e
p
a
r
a
m
e
t
e
r
s
c
h
a
n
g
e
s
.
A
c
l
u
s
t
e
r
r
e
f
e
r
s
t
o
a
coll
ect
ion
of
data
p
oin
ts
com
bin
ed
toge
ther
beca
us
e
of
certai
n
sim
il
arit
ie
s.
A
ce
ntr
oid
is
the
l
ocati
on
r
ep
res
enting
the
ce
nt
er
of
the
cl
us
te
r.
K
-
m
eans
al
gorith
m
identifie
s
k
num
ber
of
c
entr
oid
s,
an
d
then
al
locat
es
ever
y
data
po
i
nt
t
o
the n
ea
rest cl
ust
er, wh
il
e
kee
ping the
cent
roi
ds
as sm
al
l as po
s
sible
[20
-
25]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
11
, No
.
1
,
Febr
uar
y
2021
:
872
-
878
874
2.
PROP
OSE
D MET
HO
D
The pr
opose
d
m
et
ho
d
ca
n be
i
m
ple
m
ented
by
ap
plyi
g
t
he f
ollow
i
ng 2 pha
ses:
Ph
ase
1
: LB
P
histo
gr
am
calc
ulati
on
.
This
ph
a
se ca
n be im
ple
m
ented per
form
ing
the foll
owin
g
s
te
ps
:
a.
Get the
dig
it
al
wav
e
f
i
le
.
b.
Re
sh
ape
the
w
ave
file
into o
ne
row ar
ray.
c.
Fo
r
eac
h value
in the r
ow cal
cula
te
LBP
oper
at
or
as
s
how
n
i
n
Fi
gure
4
Figure
4.
LBP
histo
gr
am
calc
ulati
on
d.
Add o
ne
to
the
re
petit
ion
of L
BP operat
or v
a
lue.
Figure
5
s
hows
the calc
ulate
d LB
P h
ist
ogram
of the
duc
k w
ave
file
.
Ph
ase
2
:
K
-
m
eans cl
us
te
rin
g
Cl
us
te
rin
g
m
e
ans
gro
upin
g
the
data
val
ues
in
the
input
data
file
into
cl
us
te
rs
(
gro
ups)
[20
-
25]
,
each
cl
us
te
r
wi
ll
hav
e
a
ce
nte
r
(c
entr
oid),
se
t
of
val
ues
w
hich
a
re
belo
ng
t
o
a
nd
within
a
cl
us
te
r
s
u
m
(sum
of
the
va
lues
bel
ong
to
the
cl
us
t
er)
,
one
or
m
or
e
of
these
pa
r
a
m
et
ers
can
be
us
e
d
t
o
f
orm
t
he
data
file
fea
tures.
Figure
6
s
hows
how a
data in
put set
was gr
ou
ped into
2 cl
us
t
ers:
To per
f
or
m
the clusterin
g p
ha
se w
e
h
a
ve
t
o
a
pp
ly
the
foll
ow
ing
ste
ps
:
1
)
Get the
LBP
histogram
o
f
the
dig
it
al
voice si
gn
al
.
2)
In
it
ia
li
ze the num
ber
of clu
ste
rs
a
nd the ce
nt
ro
id
of eac
h
cl
us
te
r.
3)
Wh
il
e th
ere a
r
e a ch
a
nges i
n t
he
cal
culat
ed
centr
oid
s
do th
e f
ollow
i
ng
:
a)
Ca
lc
ulate
the dist
ances b
et
we
en
eac
h data se
t value a
n
d
cl
ust
er cen
t
ro
i
d,
wh
ic
h
is e
qual
to absol
ute
value o
f
the
d
e
fer
e
nce
betwee
n
the
cente
r
a
nd the
d
at
a it
em
v
al
ue
.
b)
Sele
ct
the
valu
e n
ea
rest cluste
r,
t
he
m
ini
m
u
m
d
ist
ance the
m
ini
m
u
m
clust
er
nu
m
ber
.
c)
Find the
ne
w
c
entr
oid
s
by a
ve
rag
i
ng the
valu
es w
it
hi
n
the
clusters
.
Wor
ked exam
ple:
The
f
ollo
wing
exam
ple
sh
ows
ho
w
to
group
the
i
nput
data
into
2
c
lusters
with
th
e
fo
ll
owin
g
centr
oid
s i
niti
al
v
al
ues
Tab
le
s
3
a
nd
4
:
c1
=
16
c2
=
22
Figure
5.
LBP
histo
gr
am
o
f d
uck w
a
ve fil
e
Figure
6.
G
rou
ping in
put data
set
into
2 cl
us
t
ers
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
o
m
p
En
g
IS
S
N: 20
88
-
8708
Develo
ping
digi
tal signal cl
us
te
ring
meth
od
us
in
g
l
oca
l
binary
patt
ern
his
tog
r
am
(
Rash
ad J.
R
as
r
as
)
875
Table
3.
Wor
ke
d
e
xam
ple p
a
sses 1 an
d 2
Pass
1
Pass
2
Data
D1
D2
Near
est
clu
ster
New c
en
t
roid
s
Data
D1
D2
Near
est
clu
ster
New c
en
t
roid
s
15
1
7
1
1
5
.33
3
6
.25
15
0
.33
2
1
.25
1
1
8
.56
4
5
.9
15
1
7
1
15
0
.33
2
1
.25
1
16
0
6
1
16
0
.67
2
0
.25
1
19
3
3
2
19
3
.67
1
7
.25
1
19
3
3
2
19
3
.67
1
7
.25
1
20
4
2
2
20
4
.67
1
6
.25
1
20
4
2
2
20
4
.67
1
6
.25
1
21
9
1
2
21
5
.67
1
5
.25
1
22
6
0
2
22
6
.67
1
4
.25
1
28
12
6
2
28
1
2
.67
8
.25
2
35
19
13
2
35
1
9
.67
1
.25
2
40
24
18
2
40
2
4
.67
3
.75
2
41
25
19
2
41
2
5
.67
4
.75
2
42
26
20
2
42
2
6
.67
5
.75
2
43
27
21
2
43
2
7
.67
6
.75
2
44
28
22
2
44
2
8
.67
7
.75
2
60
44
38
2
60
4
4
.67
2
3
.75
2
61
45
39
2
61
4
5
.67
2
4
.75
2
65
49
43
2
65
4
9
.67
2
8
.75
2
Table
4.
Wor
ke
d
e
xam
p
le
p
a
sses 3 an
d 4
Pass
3
Pass
4
Data
D1
D2
Near
est
clu
ster
New c
en
t
roid
s
Data
D1
D2
Near
est
clu
ster
New c
en
t
roid
s
15
3
.56
3
0
.9
1
1
9
.50
4
7
.89
15
4
.50
3
2
.89
1
1
9
.50
4
7
.89
No
chan
g
es, so
sto
p
15
3
.56
3
0
.9
1
15
4
.50
3
2
.89
1
16
2
.56
2
9
.9
1
16
3
.50
3
1
.89
1
19
0
.44
2
6
.9
1
19
0
.50
2
8
.89
1
19
0
.44
2
6
.9
1
19
0
.50
2
8
.89
1
20
1
.44
2
5
.9
1
20
0
.50
2
7
.89
1
20
1
.44
2
5
.9
1
20
0
.50
2
7
.89
1
21
2
.44
2
4
.9
1
21
1
.50
2
6
.89
1
22
3
.44
2
3
.9
1
22
2
.50
2
5
.89
1
28
9
.44
1
7
.9
1
28
8
.50
1
9
.89
1
35
1
6
.44
1
0
.9
2
35
1
5
.50
1
2
.89
2
40
2
1
.44
5
.9
2
40
2
0
.50
7
.89
2
41
2
2
.44
4
.9
2
41
2
1
.50
6
.89
2
42
2
3
.44
3
.9
2
42
2
2
.50
5
.89
2
43
2
4
.44
2
.9
2
43
2
3
.50
4
.89
2
44
2
5
.44
1
.9
2
44
2
4
.50
3
.89
2
60
4
1
.44
1
4
.1
2
60
4
0
.50
1
2
.11
2
61
4
2
.44
1
5
.1
2
61
4
1
.50
13
.11
2
65
4
6
.44
1
9
.1
2
65
4
5
.50
1
7
.11
2
The
cal
c
ulate
d param
et
ers
are:
Ce
ntro
i
ds
: C
1=
19.50, C
2=
47.89
W
it
hin
cl
us
te
r
su
m
s: W
CS
1=
195,
WSC2
=
431
Nu
m
ber
of poi
nts: N
um
ber
of
points i
n
cl
us
t
er
1=10, N
um
ber
of
points i
n cl
us
te
r 2=9
3.
RESU
LT
S
A
NA
L
YS
I
S
The
pr
opos
e
d m
et
ho
d was im
plem
ented
us
in
g
va
rio
us
d
i
gital
w
ave f
il
es.
Each tim
e
a L
BP h
ist
ogram
was
cal
culat
ed
and
us
ed
for
c
lusterin
g,
the
m
ai
n
adv
a
ntag
es
of
the
pro
posed
m
et
ho
d
is
a
flexibili
ty
,
her
e
w
e
can
us
e
the
c
e
ntr
oid
s,
or
within
cl
us
te
rs
s
um
s,
or
cl
us
te
r
po
i
nts
to
c
reat
e
wa
ve
file
fea
tures,
al
so
it
is
easy
to
adjust
the
nu
m
ber
of
cl
us
te
rs
to
e
xp
a
nd
the
num
ber
of
el
e
m
ents
in
the
featur
e
s
ar
ra
y.
The
e
xp
e
ri
m
ental
resu
lt
s
s
howe
d
that
the
obta
ined
featu
res
f
or
eac
h
w
ave
file
are
uniq
ue
,
thu
s
t
hey
can
be
us
e
d
as
a
key
or
s
i
g
n
a
t
u
r
e
t
o
r
e
t
r
i
e
v
e
o
r
r
e
c
o
g
n
i
z
e
t
h
e
w
a
v
e
f
i
l
e
,
a
n
d
T
a
b
l
e
5
s
h
o
w
s
t
h
e
c
a
l
c
u
l
a
t
e
d
f
e
a
t
u
r
e
s
f
o
r
s
o
m
e
w
a
v
e
f
i
l
e
s
a
m
p
l
e
s
.
The
pro
posed
m
et
ho
d
was
te
ste
d
us
in
g
t
he
sam
e
wav
e
file
but
wit
h
diff
e
ren
t
sam
pling
fr
e
qu
e
ncies,
T
able
6
sh
ows
that
the
fe
at
ur
es
f
or
t
he
wa
ve
file
re
m
ai
n
the
sam
e.
Also
the
pro
po
se
d
m
et
h
od
was
te
ste
d
us
ing
the
s
a
m
e
w
a
v
e
f
i
l
e
b
u
t
w
i
t
h
d
i
f
f
e
r
e
n
t
a
m
p
l
i
t
u
d
e
s
,
T
a
b
l
e
7
s
h
o
w
s
t
h
a
t
t
h
e
f
e
a
t
u
r
e
s
f
o
r
t
h
e
w
a
v
e
f
i
l
e
r
e
m
a
i
n
t
h
e
s
a
m
e
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
11
, No
.
1
,
Febr
uar
y
2021
:
872
-
878
876
Table
5.
Wa
ve fil
e featu
res
W
av
e f
ile
Featu
res (
C
en
troid
s)
C1
C2
C3
C4
Co
w
243
183
93
8
Do
g
227
164
102
60
Du
ck
231
176
112
71
Do
lp
h
in
218
158
73
32
Ho
rse
237
167
79
7
Do
n
k
ey
229
177
121
66
Eleph
an
t
255
249
114
15
Sp
o
ck
234
131
66
10
Table
6.
Wa
ve fil
e featu
res w
hen v
a
ryi
ng sa
m
pl
ing
fr
e
quen
cy
W
av
e f
ile cow
sa
m
p
lin
g
f
requ
en
cy
Featu
res (
C
en
troid
s)
C1
C2
C3
C4
1000
243
183
93
8
1500
243
183
93
8
2000
243
183
93
8
2500
243
183
93
8
3000
243
183
93
8
1
0
0
0
0
243
183
93
8
1
2
0
0
0
243
183
93
8
1
4
0
0
0
243
183
93
8
Table
7.
Wa
ve fil
e featu
res w
h
en
v
a
ryi
ng a
m
pl
it
ud
e
W
av
e f
ile cow
A
m
p
litu
d
e
Featu
res (
C
en
troid
s)
C1
C2
C3
C4
Origin
al
243
183
93
8
Ad
d
ed
by
0
.03
243
183
93
8
Su
b
tracted by 0.01
243
183
93
8
Multip
lied
by
1
.2
243
183
93
8
Multip
lied
by
0
.2
243
183
93
8
Div
id
ed
by
1
.2
243
183
93
8
Div
id
ed
by
0
.
8
243
183
93
8
4.
CONCL
US
I
O
N
A
flexi
ble,
fi
xed,
an
d
acc
ur
at
e
m
et
ho
d
of
wa
ve
fil
e
featur
es
e
xt
racti
o
n
was
pro
posed
a
nd
i
m
el
e
m
ented.
The
pro
posed
m
et
ho
d
reli
es
on
LBP
hist
ogram
.
More
than
one
par
am
eter
can
be
use
d
to
fo
rm
the
file
featu
re
s,
an
d
num
ber
of
data
it
e
m
s
in
the
feat
ur
e
arr
ay
can
be
e
asi
ly
adj
us
ta
ble.
It
was
s
how
n
that
the
gen
e
rated
f
eat
ur
es
f
or
a
ny
wav
e
file
are
un
i
qu
e
,
an
d
th
ey
can
be
us
e
d
as
a
sign
at
ur
e
to
recog
nize
the
file
.
The
si
gn
at
ur
e
is robust a
gaini
st t
he
c
hange
of sam
pling
fr
e
quency a
nd the
fi
le
a
m
plit
ud
e.
REFERE
NCE
S
[1]
I.
Gu
y
on,
et al
.
,
“
Feat
ure
Ext
ra
ctio
n,
Founda
ti
ons
and
Appl
icati
on
s
,”
Springer
,
20
06.
[2]
Ross
ing,
Thomas,
F.
Ric
h
ard
Moore,
an
d
Paul
A.
W
hee
le
r
,
“
T
he
Scie
n
ce
of
S
ound,
”
3rd
ed.
S
an
Franc
isco
,
C
A:
Addi
son
-
We
sle
y
Dev
el
op
ers P
res
s
,
2002.
[3]
A.
Al
-
Qaisi,
S.
A.
Khawat
reh
,
A.
Shara
dqah
,
Z
.
A.
Alqadi,
"W
ave
Fil
e
Feat
ur
e
s
Ext
racti
on
usi
ng
Reduc
ed
LB
P,"
Inte
rnational
Jo
urnal
of El
e
ct
ri
c
al
and
Comput
er
Engi
n
ee
ring
(
IJE
CE)
,
vol
.
8
,
no
.
5,
pp.
2780
-
278
7,
2018
.
[4]
N.
Shrivasta
va
a
nd
V.
T
y
agi,
“
Noise
-
inva
ri
ant
struct
ure
pa
tt
ern
for
image
te
xt
ure
cl
assifi
ca
t
io
n
and
ret
rie
v
al,
”
Mult
imedi
a
Tool
s
and
Ap
pl
ic
at
io
ns
,
vol. 75, no. 1
8,
pp
.
10887
-
10
906,
2016
.
[5]
S.
Li
ao,
M.
W
.
K.
La
w,
and
A.
C.
S.
Chung,
“Dom
ina
nt
loc
al
bina
r
y
patter
ns
for
te
xtu
re
cl
ass
ifi
c
at
ion
,
”
I
E
EE
Tr
ansacti
ons on
Image
Proc
ess
ing
,
vol
.
18
,
no
.
5
,
pp.
1107
-
1118
,
2009.
[6]
T.
Oj
al
a
,
M.
Pie
ti
käi
n
en,
and
T.
Mäe
npää,
“
Mult
ire
soluti
on
gra
y
-
sca
le
and
ro
ta
t
io
n
inva
ri
ant
te
x
tu
re
c
la
ss
ifi
c
at
ion
with
local
bin
ar
y
pa
tterns,”
I
EEE
Tr
ansacti
o
ns
on
Pat
te
rn
Analysis
and
Mac
hi
ne
Intelli
g
ence
,
vol.
24,
no
.
7,
pp.
971
-
987
,
20
02.
[7]
Z.
Guo,
L. Z
h
an
g,
and
D
.
Zh
ang, “A
complet
ed m
odel
ing
of
lo
cal
bina
r
y
patter
n
oper
at
or
for
t
exture
class
ifi
c
at
ion
,
”
IEE
E
Tr
ansacti
o
ns on
Image Proce
ss
ing
,
vo
l. 19,
no.
6
,
pp
.
1657
-
1663,
2010
.
[8]
Z.
Guo,
L.
Zha
n
g,
and
D.
Zha
ng
,
“
Rota
ti
on
inv
ariant
t
ext
ure
class
ifi
cation
using
L
BP
var
ia
nc
e
(LB
PV
)
with
global
m
at
chi
ng,
”
Pat
t
e
rn R
ec
ogn
it
ion
,
vol.
43
,
no
.
3
,
pp
.
706
-
719
,
2010
.
[9]
G.
Zha
o
and
M.
Piet
ikäinen
,
“
D
y
namic
t
ext
ure
re
cogni
ti
o
n
using
l
oca
l
b
ina
r
y
patte
rns
with
an
appl
i
ca
t
ion
to
f
acia
l
expr
essions,”
IE
EE
Tr
ansacti
ons
on
Pa
tt
ern
Analysis and
Ma
chi
n
e
Int
el
l
ige
n
ce
,
v
ol.
29
,
no
.
6
,
pp
.
915
-
928,
2007
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
o
m
p
En
g
IS
S
N: 20
88
-
8708
Develo
ping
digi
tal signal cl
us
te
ring
meth
od
us
in
g
l
oca
l
binary
patt
ern
his
tog
r
am
(
Rash
ad J.
R
as
r
as
)
877
[10]
G.
Zha
o,
M.
Ba
rna
rd,
and
M.
Pi
et
ik
äi
nen
,
“
Li
pr
ea
ding
with
lo
cal
spati
ot
emporal
desc
ript
ors,
”
IE
EE
Tr
ansacti
ons
on
Multimedia
,
vol.
11
,
no
.
7
,
pp
.
1254
-
1265
,
20
09.
[11]
G.
Zha
o
,
T
.
Ahonen,
J.
Ma
ta
s,
and
M.
Pietikä
i
nen,
“
Rotation
-
i
nvar
ia
n
t
image
and
vide
o
d
esc
r
ipt
ion
with
loca
l
bina
r
y
p
at
t
ern
f
e
at
ure
s,
”
I
EE
E
Tr
ansacti
ons on
I
mage
Proce
ss
in
g
,
vo
l
.
21
,
no
.
4
,
pp.
1465
-
1477
,
2012.
[12]
T.
Oj
al
a
,
M
.
Pie
ti
käi
n
en,
and
D.
Harwood,
“
A
co
m
par
at
ive
stud
y
of
te
x
ture
m
e
asure
s
with
cl
assifi
ca
t
ion
base
d
o
n
fea
tur
ed
d
istri
bu
ti
ons,”
Pattern
R
ec
ogni
ti
on
,
vol
.
29,
no
.
1
,
pp
.
51
-
59,
1996
.
[13]
G.
K
y
lbe
rg
and
I.
M.
Sintorn
,
“
Eva
lua
t
ion
of
noise
robustness
for
lo
cal
bin
ar
y
p
atter
n
desc
ri
ptors
in
te
xtur
e
cl
assifi
ca
t
ion,”
EURA
SIP
Journ
al
on
Image
and
Vi
deo
Proce
ss
in
g
,
vol
.
2013
,
no
.
17.
2013
.
[14]
X.
Ta
n
and
B.
T
riggs,
“
Enha
nc
e
d
loc
a
l
te
x
ture
f
ea
tur
e
sets
for
f
a
ce
re
cognition
u
nder
d
ifficult
li
g
hti
ng
cond
it
ions,
”
IEE
E
Tr
ansacti
o
ns on
Image Proce
ss
ing
,
vo
l. 19,
no.
6
,
pp
.
1635
-
1650,
2010
.
[15]
H.
Ta
ng
,
B
.
Yin,
Y.
Sun,
and
Y.
Hu,
“
3D
fac
e
re
cogni
ti
on
using
l
oca
l
b
ina
r
y
patte
rns,”
Signa
l
Pro
ce
ss
ing
,
vo
l.
9
3
,
no.
8
,
pp
.
2190
-
2198,
2013
.
[16]
T.
Aho
nen,
A.
Hadid,
and
M.
Piet
ikäinen
,
“
Face
Descri
p
ti
on
with
Lo
ca
l
Bina
r
y
Pattern
s
:
Applic
ation
to
Face
R
e
c
o
g
n
i
t
i
o
n
,”
i
n
I
E
E
E
T
r
a
n
s
a
c
t
i
o
n
s
o
n
P
a
t
t
e
r
n
A
n
a
l
y
s
i
s
a
n
d
M
a
c
h
i
n
e
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
2
8
,
n
o
.
1
2
,
p
p
.
2
0
3
7
-
2
0
4
1
,
2006.
[17]
Saji
da
P.
,
Nade
e
m
N.
and
Jherna
D.,
"Revi
ew
on
Loc
a
l
Bina
r
y
P
at
t
ern
(LBP)
Text
ure
Descri
tor
and
Its
Vari
an
ts,
"
Inte
rnational
Jo
urnal
of
Adv
an
c
ed
R
ese
arch
(
IJ
AR
)
,
vol
.
5
,
no
.
5,
pp
.
708
-
717
,
2017.
[18]
Y.
Y
in,
X.
W
an
g,
D.
Xu,
F
.
Li
u
,
Y.
W
ang,
and
W
.
W
u,
“
Robust
visu
al
d
et
e
ct
ion
-
le
arn
ing
-
tracki
n
g
fra
m
ework
fo
r
aut
onom
ous
ae
ri
al
r
efu
eling
of
UA
Vs
,
”
IEE
E
Tr
ansacti
ons
on
I
nstrum
ent
ati
on
and
Me
asur
eme
nt
,
vo
l.
65,
no
.
3,
pp.
510
-
521
,
20
16.
[19]
B.
Z
ahr
an
,
J.
Al
-
Azz
eh
,
Z
.
Alq
a
d
i
and
Mohd
-
As
hra
f
Al
Zoghoul,
"
A
Modifie
d
L
bp
Method
To
Ext
ra
ct
Fe
at
ur
es
F
r
o
m
C
o
l
o
r
I
m
a
g
e
s
,
"
J
o
u
r
n
a
l
o
f
T
h
e
o
r
e
t
i
c
a
l
a
n
d
A
p
p
l
i
e
d
I
n
f
o
r
m
a
t
i
o
n
T
e
c
h
n
o
l
o
g
y
, v
o
l
.
9
6
,
n
o
.
1
0
,
p
p
.
3
0
1
4
-
3
0
2
4
,
2
0
1
8
.
[20]
Madhuri
A.
Tayal
,
M.
M.
Raghu
wanshi,
“
Revie
w
on
Vari
ous
Cluste
r
ing
Method
s
for
the
I
m
age
Data
,
”
Journal
of
Eme
rging Trend
s in
Computing
and
Informat
ion Sci
en
ce
s
,
v
ol. 2
,
pp.
34
-
38
,
2011
.
[21]
H.
Ta
r
iq,
S.
Bu
rne
y
,
"K
-
Mea
ns
Cluste
r
Ana
l
y
s
is
for
Im
age
Se
gm
ent
at
ion
,
”
In
t
ernati
onal
J
ournal
of
Comput
e
r
Appl
ic
a
ti
ons
,
vo
l.
96
,
no
.
4
,
pp
.
1
-
8,
2014
.
[22]
D.
Sa'adi
l
la
h
Ma
y
l
awa
t
i,
T
.
Pria
t
na,
H.
Sugil
ar,
M.
Ali
Ramdhan
i,
"
Data
sc
ie
n
ce
for
digi
tal
cu
lt
ur
e
improvem
ent
i
n
highe
r
educat
ion
using
K
-
m
ea
ns
cl
uster
ing
and
text
ana
l
y
t
ic
s,"
Int
ernati
onal
Jour
nal
of
El
e
ct
rica
l
and
Computer
Engi
ne
ering
(
IJ
ECE
)
,
vol
.
10
,
n
o.
5,
pp
.
4569
-
4
580,
2020
.
[23]
A.
Abdul
-
huss
ia
n
Hass
an,
W
.
M
d
Shah,
M.
Fairu
z
Iskanda
r
Oth
m
an,
and
H.
Hass
an,
"
Eva
lu
at
e
t
he
per
form
anc
e
of
K
-
Mea
ns
and
the
fuz
z
y
C
-
Me
ans
al
gorit
hm
s
to
form
at
ion
b
al
an
ce
d
c
luste
rs
in
wire
le
ss
sensor
net
works
,
"
Inte
rnational
Jo
urnal
of El
e
ct
ri
c
al
and
Comput
er
Engi
n
ee
ring
(
IJE
CE)
,
vol
.
10
,
n
o.
2
,
pp
.
1515
-
1
523
,
2020
.
[24]
E.
Che
rra
t
,
R.
Alaoui,
and
H
.
B
ouza
hir
,
"Im
proving
of
Fingerpr
int
Segm
ent
a
ti
o
n
Im
age
s
Based
on
K
-
m
ea
ns
an
d
DBS
CAN
Clust
eri
ng,
”
Int
ernational
Journal
o
f
El
e
ct
ri
cal
and
Computer
Eng
ine
ering
(
IJE
C
E
)
,
v
ol
.
9
,
n
o.
4,
pp.
2425
-
2432
,
2019
.
[25]
I.
Qa
y
s
Abduljaleel,
and
A.
Ha
m
ee
d
Khaleel
,
"H
idi
ng
te
x
t
in
s
pee
ch
sign
al
usi
ng
K
-
m
ea
ns,
L
SB
te
chn
ique
s
an
d
cha
ot
ic
m
a
p
s
,
"
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
E
l
e
c
t
r
i
c
a
l
a
n
d
C
o
m
p
u
t
e
r
E
n
g
i
n
e
e
r
i
n
g
(
I
J
E
C
E
)
,
v
o
l
.
1
0
,
n
o
.
6
,
p
p
.
5
7
2
6
-
5
7
3
5
,
2
0
2
0
.
BIOGR
AP
H
I
ES
OF
A
UTH
ORS
Ras
ha
d
J.
Rasr
as
rec
ei
v
ed
th
e
PhD
degr
ee
from
Nati
onal
Te
chn
ic
a
l
Univ
ersity
(Kharkov
Pol
y
t
ec
hni
c
Insti
tut
e)
in
2001,
wi
th
rese
arc
h
in
au
tomate
d
int
e
ll
ig
e
nt
cont
rol
s
y
st
e
m
s.
His
rese
arc
h
int
er
est
in
cl
ude
s
image
proc
essing,
m
ac
hin
e
le
arn
ing,
and
a
dvanc
ed
computer
a
rch
i
te
c
ture.
He
works
as
an
associa
ti
v
e
profe
ss
or
at
Com
p
ute
r
Engi
n
ee
rin
g
depa
rtment,
Al
-
Bal
qa
Appli
ed
Univer
sit
y
.
Bil
al
Z
ahran
r
ec
e
ive
d
the
B
.
Sc
degr
e
e
in
E
le
c
tri
c
al
&
E
lectr
oni
c
Eng
.
fr
om
Middle
Ea
s
t
Te
chn
ic
a
l
Unive
rsit
y
,
Turk
e
y
,
in
1996
,
the
M.Sc
degr
ee
in
Com
m
unic
a
ti
ons
Eng
.
from
Univer
sit
y
of
Jordan,
Jorda
n,
in
1999
,
and the
PhD
degr
e
e
i
n
Com
pute
r
Info
rm
at
ion
S
y
st
em (CIS) fr
om
Arab
Aca
dem
y
for
B
anki
ng
and
Fina
nci
a
l
Scie
n
ce
s,
Jordan,
in
2009
.
He
is
cur
r
ent
l
y
working
as
an
As
socia
te
Pr
ofe
ss
or
at
depa
rtmen
t
of
Com
pute
r
Engi
nee
r
ing,
Facu
lty
of
Engi
ne
ering
Te
chnol
o
g
y
,
Al
-
Bal
qa
Appl
i
ed
Univer
sit
y
,
Jordan.
His
re
sea
rch
in
te
r
ests
inc
lud
e
artifi
c
ia
l
in
telli
g
enc
e,
opti
m
iz
ation
and
digi
t
al signa
l
pr
oce
ss
ing
fi
el
ds.
Email
:
za
h
ran
b
@bau.
edu
.
jo
,
/
z
ahr
anb@
y
ahoo
.
c
om
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
11
, No
.
1
,
Febr
uar
y
2021
:
872
-
878
878
Muta
z
Ra
smi
Ab
u
Sara
rec
eived
the
B.
Sc
from
Saint
Pete
rsburg
El
ec
tr
otechn
ic
a
l
Univer
sit
y
i
n
2004.
He
r
ece
ive
d
th
e
Mast
er
of
Sc
ie
nc
e
(Data
b
ase
S
ystems
)
from
S
ai
nt
Pe
te
rsburg
El
e
ct
rot
ec
hni
cal
Univer
sit
y
in
2
006.
After
work
ing
as
progra
m
m
er
at
BiSoft
Com
pan
y
in
Sain
t
Pete
rsburg.
He
r
ec
e
ive
d
th
e
Phd
in
Saint
Pete
rsb
urg
El
ectrot
ec
hn
ic
a
l
Univer
sit
y
f
rom
2007)
with
Resea
rch
and
Deve
lopment
of
Inte
gr
at
ed
Dat
aba
se
C
irc
ui
t
C
om
ponent
s
for
CAD
Schemati
c
.
His
rese
arc
h
i
nte
rest
inc
lud
es
Data
bas
e
S
y
s
te
m
s,
Alg
orit
h
m
s
and
Data
Struct
ure
s
an
d
Optimiza
ti
o
n.
He
works
as
assistant
prof
essor fro
m
2011
ti
l
l
now
at
Ta
ib
ah
Univ
er
sit
y
.
Z
iad
AlQadi
is
cur
ren
t
l
y
workin
g
as
a
Profess
or
at
Com
pute
r
En
gine
er
ing
Depa
rt
m
ent
,
Facult
y
o
f
Engi
ne
eri
ng
Tec
hnolog
y
,
Al
-
Balqa
Applie
d
Uni
ver
sit
y
.
He
is
th
e
Hea
d
of
Comput
er
engi
n
ee
r
in
g
depa
rtment
.
His
rese
arc
h
in
te
rest
s inc
lud
e
Signa
l proce
ss
ing,
par
a
l
le
l
proc
essing
,
i
m
age
proc
essing
Evaluation Warning : The document was created with Spire.PDF for Python.