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.
1224
~
1228
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.
14256
1224
Jou
r
n
al
h
omepage
:
ht
tp:
//
jour
nal.
uad
.
ac
.
id/
index
.
php/T
E
L
K
OM
N
I
K
A
A
m
od
ifie
d
sym
m
e
t
r
ic
l
oc
al
b
in
ar
y p
a
t
t
e
r
n
f
or
i
m
age
f
e
a
t
u
r
e
s
e
xt
r
a
c
t
io
n
M
aj
e
d
O.
Dw
airi
Facu
l
t
y
o
f
E
n
g
i
n
eeri
n
g
T
ec
h
n
o
l
o
g
y
,
Co
mmu
n
i
ca
t
i
o
n
E
n
g
i
n
eeri
n
g
D
e
p
art
me
n
t
,
Al
-
b
a
l
q
a
A
p
p
l
i
ed
U
n
i
v
er
s
i
t
y
,
J
o
r
d
an
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
Oc
t
20
,
2019
R
e
vis
e
d
F
e
b
1
2
,
2020
Ac
c
e
pted
F
e
b
23
,
2020
T
h
e
p
r
o
ces
s
o
f
i
d
e
n
t
i
fy
i
n
g
i
mag
e
s
an
d
p
a
t
t
er
n
s
i
s
o
n
e
o
f
t
h
e
mo
s
t
i
m
p
o
r
t
an
t
p
ro
ce
s
s
e
s
o
f
d
i
g
i
t
al
i
m
ag
e
p
r
o
ces
s
i
n
g
,
w
h
i
c
h
i
s
u
s
ed
i
n
m
an
y
ap
p
l
i
cat
i
o
n
s
s
u
ch
as
fi
n
g
erp
r
i
n
t
reco
g
n
i
t
i
o
n
,
face
reco
g
n
i
t
i
o
n
an
d
p
at
t
ern
reco
g
n
i
t
i
o
n
.
D
u
e
t
o
t
h
e
l
arg
e
s
i
ze
o
f
t
h
e
i
ma
g
e,
t
h
e
p
r
o
ces
s
o
f
i
d
e
n
t
i
fy
i
n
g
t
h
e
i
mag
e
re
q
u
i
res
a
g
reat
t
i
me,
w
h
i
c
h
i
n
t
u
rn
l
ead
s
u
s
t
o
e
x
t
rac
t
s
o
me
ch
aract
er
i
s
t
i
c
s
o
f
t
h
e
mag
n
i
t
u
d
e
o
f
t
h
e
v
o
l
u
me,
w
h
i
ch
can
b
e
u
s
ed
a
s
an
i
d
en
t
i
f
i
er
t
o
ret
r
i
ev
e
t
h
e
i
ma
g
e
o
r
reco
g
n
i
ze
i
t
an
d
t
h
u
s
w
e
h
a
v
e
d
ev
o
t
e
d
a
l
o
t
o
f
t
i
me
t
o
i
d
en
t
i
f
y
t
h
e
i
ma
g
e.
In
t
h
i
s
res
earch
p
a
p
er,
a
mo
d
i
f
i
ed
s
y
mmet
r
i
c
l
o
cal
b
i
n
ar
y
p
at
t
ern
(MSL
BP)
met
h
o
d
w
as
p
ro
p
o
s
ed
t
o
e
x
t
rac
t
t
e
x
t
u
re
fe
at
u
re
s
.
T
h
e
p
ro
p
o
s
ed
al
g
o
ri
t
h
m
w
a
s
i
m
p
l
eme
n
t
e
d
o
n
ma
n
y
d
i
g
i
t
al
fi
n
g
e
rp
ri
n
t
’
s
i
mag
e
s
an
d
t
h
e
l
o
cal
s
t
r
u
ct
u
re
feat
u
res
o
f
t
h
es
e
i
mag
e
s
w
ere
o
b
t
ai
n
e
d
.
Sev
eral
i
mag
e
reco
g
n
i
t
i
o
n
ex
p
er
i
men
t
s
are
co
n
d
u
c
t
ed
o
n
t
h
es
e
feat
u
re
s
an
d
co
mp
are
d
w
i
t
h
o
t
h
er
al
g
o
r
i
t
h
ms
.
T
h
e
re
s
u
l
t
s
o
f
t
h
e
p
r
o
p
o
s
e
d
al
g
o
ri
t
h
m
s
h
o
w
e
d
t
h
at
t
h
e
d
i
g
i
t
al
i
ma
g
e
w
as
rep
re
s
en
t
ed
i
n
a
v
ery
s
mal
l
s
i
ze
a
n
d
fu
rt
h
ermo
r
e
t
h
e
s
p
ee
d
an
d
accu
rac
y
o
f
i
mag
e
reco
g
n
i
t
i
o
n
b
as
e
d
o
n
t
h
e
p
ro
p
o
s
ed
met
h
o
d
w
as
i
n
creas
e
d
s
i
g
n
i
f
i
can
t
l
y
.
U
n
l
i
k
e
t
h
e
met
h
o
d
s
b
a
s
ed
o
n
L
BP,
t
h
e
p
ro
p
o
s
ed
met
h
o
d
g
i
v
es
t
h
e
s
ame
feat
u
res
o
f
t
h
e
i
ma
g
e
ev
en
i
f
t
h
e
i
mag
e
w
as
ro
t
a
t
ed
w
i
t
h
an
y
an
g
l
e.
K
e
y
w
o
r
d
s
:
C
S
L
B
P
E
xtr
a
c
ti
on
ti
me
I
mage
f
e
a
tur
e
s
I
mage
r
otation
L
B
P
M
S
L
B
P
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
:
M
a
jed
O.
Dw
a
ir
i
,
F
a
c
ult
y
of
E
nginee
r
ing
T
e
c
hnology
,
C
omm
unica
ti
on
E
nginee
r
ing
De
pa
r
tm
e
nt,
Al
-
ba
lqa
Applied
Unive
r
s
it
y
,
Amman,
J
or
da
n
.
E
mail:
maje
ddw@
ba
u.
e
du.
jo
1.
I
NT
RODU
C
T
I
ON
Gr
a
y
digi
tal
im
a
ge
s
a
r
e
lar
ge
in
s
ize
a
nd
a
r
e
us
ua
ll
y
s
e
ve
r
a
l
thous
a
nd
pixels
in
s
i
z
e
,
mor
e
ove
r
it
take
s
a
gr
e
a
t
de
a
l
of
ti
me
to
pr
oc
e
s
s
the
im
a
ge
s
to
ident
if
y
them
[
1]
,
T
a
ble
1
s
hows
s
ome
dif
f
e
r
e
nt
in
s
ize
im
a
ge
s
,
a
nd
the
r
e
quir
e
d
ti
me
to
identif
y
e
a
c
h
im
a
ge
.
F
r
om
thi
s
table
we
c
a
n
s
e
e
that
the
a
ve
r
a
ge
ti
me
to
pr
oc
e
s
s
e
a
c
h
pixel
f
or
matc
hing
e
qua
l
10.
9
45
mi
c
r
o
s
e
c
onds
,
w
hich
is
c
ons
ider
e
d
a
high
ti
me.
F
igur
e
1
s
hows
that
ther
e
is
a
li
ne
a
r
r
e
lations
hip
be
twe
e
n
the
im
a
ge
s
ize
a
nd
th
e
r
e
quir
e
d
r
e
c
ognit
ion
or
identif
ying
ti
me
.
T
o
a
void
th
is
pr
oblem
,
a
nd
to
mi
nim
ize
the
r
e
c
o
gnit
ion
ti
me
we
ha
ve
to
s
e
e
k
a
n
e
f
f
icie
nt
method
c
a
pa
ble
to
r
e
pr
e
s
e
nt
the
im
a
ge
by
a
s
e
t
of
va
lues
c
a
ll
e
d
im
a
ge
f
e
a
tur
e
s
,
whic
h
c
a
n
be
us
e
d
a
s
a
n
ide
nti
f
ier
to
r
e
tr
ieve
o
r
r
e
c
ognize
the
im
a
ge
.
An
im
a
ge
f
e
a
tur
e
s
is
a
s
e
t
of
metr
ics
c
a
lcula
ted
in
im
a
ge
pr
oc
e
s
s
ing
a
nd
they
a
r
e
c
r
e
a
ted
to
qua
n
ti
f
y
the
pe
r
c
e
ived
textur
e
of
a
n
i
mage
.
I
mage
f
e
a
tur
e
s
give
us
inf
o
r
mation
a
bout
th
e
s
pa
ti
a
l
a
r
r
a
nge
ment
of
pixels
va
lues
or
int
e
ns
it
ies
in
a
n
im
a
ge
or
s
e
lec
ted
r
e
gion
of
a
n
im
a
ge
,
s
uc
h
a
s
e
nc
r
y
pti
on
a
nd
de
c
r
ypti
on
[
2
-
4]
.
T
he
e
xt
r
a
c
ted
f
e
a
tur
e
s
mus
t
f
or
m
a
ke
y
whic
h
c
a
n
be
us
e
d
a
s
a
n
im
a
ge
identif
ier
,
a
nd
he
r
e
thes
e
f
e
a
tur
e
s
mus
t
be
:
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
A
modifi
e
d
s
y
mm
e
tr
ic
local
binar
y
patt
e
r
n
for
ima
ge
featur
e
s
e
x
tr
ac
ti
on
(
M
ajed
O.
Dw
air
i)
1225
−
Unique
f
or
e
a
c
h
im
a
ge
a
nd
da
ta
[
5,
6
].
−
S
mall
s
ize
c
ompar
ing
with
the
i
mage
s
ize
[
7
].
−
C
a
pa
ble
to
r
e
duc
e
the
im
a
ge
r
e
tr
ieving
ti
me
[
8
].
−
S
im
ple
to
be
c
r
e
a
ted.
−
De
pe
nda
ble
on
the
im
a
ge
textur
e
[
5
,
9
,
1
0
].
−
Unc
ha
nge
a
ble
if
the
im
a
ge
wa
s
r
otate
d
.
T
a
ble
1
.
I
mage
s
ize
a
nd
identi
f
ying
ti
me
I
ma
ge
N
umbe
r
of
r
ow
s
N
umbe
r
of
c
ol
umns
S
iz
e
(
pi
xe
l)
M
a
tc
hi
ng t
im
e
(
S
e
c
onds
)
1
368
267
98256
0.021000
2
265
570
151050
0.031000
3
283
534
151122
0.034000
4
225
675
151875
0.036000
5
600
385
231000
0.038000
6
500
1065
532500
0.051000
7
1079
1950
2104050
0.205000
8
1300
3027
3935100
0.388000
A
ve
r
a
ge
919370
0.1005
T
im
e
f
or
e
a
c
h pi
xe
l
10.945 mi
c
r
os
e
c
onds
F
igur
e
1
.
R
e
lations
hip
be
twe
e
n
im
a
ge
s
ize
a
nd
ide
nti
f
ying
ti
me
2.
RE
L
AT
E
D
WORKS
S
e
ve
r
a
l
a
lgor
it
hms
we
r
e
c
onduc
ted
to
e
xt
r
a
c
t
im
a
ge
f
e
a
tur
e
s
,
mos
t
o
f
them
a
r
e
ba
s
e
d
on
c
a
lcula
ti
ng
loca
l
binar
y
pa
tt
e
r
n
(
L
B
P
)
f
o
r
e
a
c
h
pixel
,
then
the
r
e
pe
ti
ti
on
o
f
e
a
c
h
L
B
P
va
lue
is
to
be
f
ind,
thes
e
r
e
pe
ti
ti
ons
will
f
or
m
the
im
a
ge
f
e
a
tur
e
s
[
8
,
9
].
I
n
[
6
]
,
the
do
mi
na
te
L
B
P
ope
r
a
tor
wa
s
pr
opos
e
d,
thi
s
ope
r
a
tor
is
to
be
c
a
lcula
ted
f
or
e
a
c
h
pixel
,
a
nd
the
r
e
pe
ti
ti
ons
of
t
he
ope
r
a
tor
va
lues
f
or
m
the
im
a
ge
f
e
a
tur
e
s
,
the
pr
opos
e
d
method
he
r
e
is
ve
r
y
s
im
ple
a
nd
f
ull
y
ba
s
e
d
on
L
B
P
method
[
9
]
.
I
n
[
1
0
]
,
a
window
method
f
o
r
a
n
e
nha
nc
e
d
im
a
ge
[
1
1
-
17
]
f
e
a
tur
e
s
e
xtr
a
c
ti
on
wa
s
pr
opos
e
d,
th
is
method
is
ve
r
y
s
im
ple
a
nd
e
f
f
icie
nt
but
if
the
im
a
ge
wa
s
r
otate
d
the
f
e
a
tur
e
s
will
c
ha
nge
,
whic
h
will
c
os
t
e
xt
r
a
wor
k
a
nd
ti
me
to
de
a
l
with
pr
oc
e
s
s
of
id
e
nti
f
ying
the
im
a
ge
.
I
n
[
18
-
23
]
de
f
e
r
e
nt
va
r
iants
of
a
lgor
it
h
m
we
r
e
pr
opos
e
d,
a
ll
o
f
them
a
r
e
ba
s
e
d
on
L
B
P
a
nd
c
e
ntr
a
l
s
ymm
e
tr
ic
L
B
P
(
C
S
L
B
P
)
ope
r
a
tor
s
,
thes
e
methods
c
r
e
a
te
a
unique
f
e
a
tur
e
f
or
e
a
c
h
im
a
ge
,
but
th
e
y
a
r
e
ve
r
y
s
e
ns
it
ive
to
the
im
a
ge
r
otation
.
F
igur
e
2
(
a
)
s
hows
how
to
c
a
lcula
te
L
B
P
ope
r
a
tor
f
o
r
e
a
c
h
pixel,
while
F
igur
e
2
(
b
)
s
hows
how
to
c
a
lcula
t
e
C
S
L
B
P
ope
r
a
t
or
f
or
e
a
c
h
pixel
.
C
S
L
B
P
methods
c
r
e
a
tes
f
o
r
e
a
c
h
im
a
ge
a
unique
f
e
a
tur
e
s
a
r
r
a
y
o
f
16
va
lue
s
a
s
s
hown
in
T
a
ble
2,
but
thes
e
f
e
a
tur
e
s
a
r
e
ve
r
y
s
e
ns
it
ive
to
th
e
im
a
ge
pos
it
ion
a
nd
if
the
i
mage
wa
s
r
otate
d
a
t
lea
s
t
f
or
1
de
gr
e
e
the
f
e
a
tur
e
s
a
r
r
a
y
wi
ll
be
c
ha
nge
d
a
c
c
or
dingl
y
a
s
s
hown
in
T
a
ble
3
,
a
nd
thi
s
is
the
majo
r
dis
a
dva
ntage
of
thi
s
method.
0
0
.
0
5
0
.
1
0
.
1
5
0
.
2
0
.
2
5
0
.
3
0
.
3
5
0
.
4
0
0
.
5
1
1
.
5
2
2
.
5
3
3
.
5
4
x
1
0
6
T
i
m
e
(
S
e
c
o
n
d
s
)
S
i
z
e
(
P
i
x
e
l
s
)
I
m
a
g
e
m
a
t
c
h
i
n
g
t
i
m
e
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:
1224
-
1228
1226
F
igur
e
2
.
(
a
)
C
a
lcula
ti
ng
L
B
P
ope
r
a
tor
,
(
b)
C
a
lcula
ti
ng
C
S
L
B
P
ope
r
a
tor
T
a
ble
2.
I
mage
s
f
e
a
tur
e
s
us
ing
C
S
L
B
P
method
T
a
ble
3
.
I
mage
f
e
a
tur
e
s
be
f
or
e
a
nd
a
f
ter
r
otation
I
ma
ge
f
e
a
tu
r
e
s
I
ma
ge
1
I
ma
ge
2
I
ma
ge
3
I
ma
ge
4
I
ma
ge
5
20622
208737
8579
13704
11358
14974
134462
7909
11075
6918
7113
68040
2245
4273
3513
13009
100008
7475
8238
7227
8030
108792
2225
4090
4851
4305
38518
1354
2879
2562
4289
43447
1460
3314
2583
13024
158969
6432
9804
11850
13964
108208
5696
10205
12066
4290
52415
1424
3204
2625
4779
38166
1360
2609
2274
6085
100799
2487
3688
4116
12934
124558
7843
8525
7905
5526
83046
2429
3669
3396
14550
110970
8001
10245
6447
81540
2447315
30071
49862
59801
I
ma
ge
f
e
a
tu
r
e
s
I
ma
ge
1
R
ot
a
te
d i
ma
ge
1
20622
21331
14974
16060
7113
7346
13009
13418
8030
6186
4305
3922
4289
4457
13024
12646
13964
13282
4290
5081
4779
4526
6085
5506
12934
13146
5526
5815
14550
14611
81540
81701
3.
T
HE
P
ROP
OS
E
D
M
S
L
B
P
M
E
T
HO
D
T
he
pr
opos
e
d
mod
if
ied
s
ymm
e
tr
ic
L
B
P
method
c
a
lcula
tes
f
or
e
a
c
h
pixel
us
ing
the
pixel
ne
ighbor
s
with
de
pth
e
qua
l
1
a
nd
de
pth
e
qua
l
2
a
s
s
hown
in
F
igur
e
3.
He
r
e
if
r
otate
the
im
a
ge
the
pixel
ne
ighbor
s
did
not
c
ha
nge
,
s
o
the
f
e
a
tu
r
e
s
r
e
main
the
s
a
me
a
f
ter
a
ny
r
otation
of
the
im
a
ge
.
M
S
L
B
P
method
c
a
n
be
im
pl
e
mente
d
a
pplyi
ng
the
f
ol
lowing
s
teps
(
f
o
r
e
a
c
h
pixel)
:
a.
I
nit
ialize
the
4
e
leme
nts
f
e
a
tur
e
s
a
r
r
a
y
to
z
e
r
os
.
b.
F
ind
the
a
ve
r
a
ge
o
f
the
ne
ighbor
s
with
de
pth
=
1
(
a
v0)
.
c.
F
ind
the
a
ve
r
a
ge
o
f
the
ne
ighbor
s
with
de
pth
=
2
(
a
v1)
.
d.
I
f
a
v0
gr
e
a
ter
o
r
e
qua
l
pixel
va
lue
make
a
0
=
1,
e
ls
e
make
a
0
=
0.
e.
I
f
a
v1
gr
e
a
ter
o
r
e
qua
l
pixel
va
lue
make
a
1
=
1,
e
ls
e
make
a
1
=
0.
f.
F
ind
the
index
of
the
f
e
a
tur
e
s
a
r
r
a
y
(
I
=
a0
+
2*a
1)
.
g.
Add
1
to
the
f
e
a
tur
e
s
a
r
r
a
y
with
index
=
I.
F
igur
e
4
s
hows
a
n
e
xa
mpl
e
o
f
how
to
f
ind
the
f
e
a
t
ur
e
s
a
r
r
a
y
index
f
or
one
pixel
.
F
igur
e
3.
C
a
lcula
ti
ng
M
S
L
B
P
ope
r
a
tor
F
igur
e
4.
C
a
lcula
ti
ng
f
e
a
tur
e
s
a
r
r
a
y
index
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
A
modifi
e
d
s
y
mm
e
tr
ic
local
binar
y
patt
e
r
n
for
ima
ge
featur
e
s
e
x
tr
ac
ti
on
(
M
ajed
O.
Dw
air
i)
1227
4.
I
M
P
L
E
M
E
NT
AT
I
ON
AN
D
E
XP
E
RI
M
E
NT
A
L
RE
S
UL
T
S
T
he
pr
opos
e
d
M
S
L
B
P
method
wa
s
im
pleme
nted
u
s
ing
va
r
ious
f
inger
p
r
int
i
mage
s
with
va
r
ious
s
ize
s
,
a
nd
f
or
e
a
c
h
im
a
ge
the
im
a
ge
f
e
a
tur
e
s
a
r
r
a
y
wa
s
a
unique
f
or
e
a
c
h
im
a
ge
.
E
a
c
h
im
a
ge
wa
s
r
otate
d
f
o
r
va
r
ious
de
gr
e
e
s
,
a
nd
the
r
e
s
ult
ing
f
e
a
tur
e
s
r
e
main
the
s
a
me
without
a
ny
c
ha
nge
.
F
igur
e
5
s
hows
a
n
or
igi
na
l
f
in
ge
r
pr
int
im
a
ge
,
a
nd
the
r
otate
d
f
or
90
de
gr
e
e
s
im
a
ge
,
while
T
a
ble
4
s
hows
the
f
e
a
tur
e
s
a
r
r
a
y
f
or
the
im
a
ge
be
f
or
e
a
nd
a
f
ter
r
otation
.
T
he
or
igi
na
l
f
inger
pr
in
t
im
a
ge
wa
s
take
n,
the
f
e
a
tur
e
s
a
r
r
a
y
wa
s
c
a
lcula
ted,
a
nd
the
s
a
me
thi
ng
wa
s
done
f
or
di
f
f
e
r
e
nt
va
r
iants
of
r
otate
d
f
inger
p
r
int
im
a
ge
,
the
r
e
s
ult
s
of
im
p
leme
ntation
is
s
hown
in
T
a
bl
e
5.
F
r
om
T
a
ble
5
we
c
a
n
s
e
e
that
the
f
e
a
tu
r
e
s
a
r
r
a
y
of
the
im
a
ge
doe
s
not
c
ha
nge
due
to
im
a
ge
r
otation
,
thi
s
gives
the
pr
opos
e
d
a
lgor
it
hm
a
big
a
dva
ntag
e
ove
r
the
other
us
e
d
method,
a
nd
r
e
ga
r
dles
s
to
the
im
a
ge
p
os
it
ion
the
f
e
a
tur
e
s
r
e
main
the
s
a
me.
T
he
f
e
a
tur
e
s
a
r
r
a
y
e
xtr
a
c
ti
on
(
c
a
lcula
ti
on)
ti
mes
we
r
e
c
a
lcu
late
d
f
or
im
a
ge
s
with
di
f
f
e
r
e
nt
s
ize
s
us
ing
both
C
S
L
B
P
a
nd
M
S
L
B
P
methods
,
the
r
e
s
ult
s
of
c
a
lcula
ti
ons
a
r
e
s
hown
in
T
a
ble
6
.
F
r
om
the
r
e
s
ult
s
s
hown
in
T
a
ble
6
we
c
a
n
s
e
e
that
the
e
xtr
a
c
ti
on
ti
me
to
c
r
e
a
te
a
n
i
mage
f
e
a
tu
r
e
s
a
r
r
a
y
us
ing
M
S
L
B
P
method
is
s
mall
a
nd
it
is
a
c
c
e
ptabl
e
,
but
C
S
L
B
P
method
is
mor
e
e
f
f
icie
nt
f
or
thi
s
c
a
s
e
.
T
his
dis
a
dva
ntage
c
a
n
be
ignor
e
d
taking
the
f
oll
owing
f
a
c
ts
int
o
c
ons
ider
a
ti
on:
a.
T
he
f
e
a
tur
e
s
a
r
r
a
y
us
ing
M
S
L
B
P
method
ha
s
only
4
e
leme
nts
,
while
the
f
e
a
tu
r
e
s
a
r
r
a
y
us
ing
C
S
L
B
P
method
ha
s
16
e
leme
nts
.
b.
T
he
da
taba
s
e
whic
h
c
a
n
be
us
e
d
to
s
tor
e
the
f
e
a
tur
e
s
a
r
r
a
ys
(
ke
ys
)
us
ing
M
S
L
P
B
method
r
e
qui
r
e
s
a
s
m
a
ll
e
r
memor
y
s
ize
.
c.
I
f
we
us
e
a
r
ti
f
icia
l
n
e
u
r
a
l
ne
twor
k
(
AN
N)
[
2
4
]
a
s
a
tool
to
identif
y
the
im
a
ge
us
ing
M
S
L
B
P
,
to
thi
s
AN
N
a
r
c
hit
e
c
tur
e
will
be
s
im
pler
.
d.
T
a
king
3
in
to
c
ons
ider
a
ti
on
AN
N
tr
a
ini
ng
t
im
e
wil
l
be
s
maller
.
e.
T
a
king
3
int
o
c
ons
ider
a
ti
on
im
a
ge
r
e
t
r
ieving
ti
me
u
s
ing
AN
N
will
be
a
ls
o
s
maller
,
a
nd
thi
s
wil
l
c
ompe
ns
a
te
the
bigger
e
xtr
a
c
ti
on
ti
me.
F
igur
e
5
.
Or
igi
na
l
a
nd
r
o
tate
d
f
inger
p
r
int
im
a
ge
s
T
a
ble
4.
F
e
a
tur
e
s
be
f
or
e
a
nd
a
f
ter
r
otation
I
ma
ge
I
ma
ge
f
e
a
tu
r
e
s
O
r
ig
in
a
l
168067
6098
3169
48766
R
ot
a
te
d
168067
6098
3169
48766
T
a
ble
5.
F
e
a
tur
e
s
f
or
the
im
a
ge
with
va
r
ious
pos
it
i
ons
T
a
ble
6.
F
e
a
tur
e
s
e
xtr
a
c
ti
on
ti
me
I
ma
ge
F
e
a
tu
r
e
s
O
r
ig
in
a
l
168067
6098
3169
48766
R
ot
a
te
d 1 de
gr
e
e
s
168067
6098
3169
48766
R
ot
a
te
d 5de
gr
e
e
s
168067
6098
3169
48766
R
ot
a
te
d 7 de
gr
e
e
s
168067
6098
3169
48766
R
ot
a
te
d 10 de
gr
e
e
s
168067
6098
3169
48766
R
ot
a
te
d 20 de
gr
e
e
s
168067
6098
3169
48766
R
ot
a
te
d 48 de
gr
e
e
s
168067
6098
3169
48766
R
ot
a
te
d 73 de
gr
e
e
s
168067
6098
3169
48766
R
ot
a
te
d 90 de
gr
e
e
s
168067
6098
3169
48766
I
ma
ge
S
iz
e
(
pi
xe
ls
)
C
S
L
B
P
f
e
a
tu
r
e
s
e
xt
r
a
c
ti
on
time
(
s
e
c
ond)
M
S
L
B
P
f
e
a
tu
r
e
s
e
xt
r
a
c
ti
on
ti
me
(
s
e
c
ond)
1
600x
385
0.034000
0.174000
2
1300x
3027
0.248000
1.001000
3
368x
267
0.007000
0.080000
4
265x 570
0.009000
0.040000
5
283x
534
0.010000
0.043000
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:
1224
-
1228
1228
5.
CONC
L
USI
ON
A
s
im
ple
a
nd
highl
y
e
f
f
icie
nt
M
S
L
B
P
method
f
or
i
mage
f
e
a
tur
e
s
e
xt
r
a
c
ti
on
wa
s
,
pr
opos
e
d
tes
ted
a
nd
im
pleme
nted.
T
he
pr
opos
e
d
method
c
a
n
s
uit
a
ny
a
p
pli
c
a
ti
on
in
the
f
ield
of
im
a
ge
r
e
c
ognit
ion
or
im
a
ge
r
e
tr
ieva
l.
T
he
p
r
opos
e
d
method
pr
ovides
s
ome
a
c
hieve
ments
s
uc
h
a
s
:
s
mall
f
e
a
tur
e
a
r
r
a
y
s
ize
,
s
mall
f
e
a
tur
e
s
e
xtr
a
c
ti
on
time
,
a
nd
t
he
f
e
a
tur
e
s
a
r
r
a
y
va
lues
a
r
e
not
s
e
ns
it
ive
to
the
im
a
ge
pos
it
ion,
r
otating
the
im
a
ge
doe
s
no
t
lea
d
to
a
ny
c
ha
nge
s
in
the
im
a
ge
f
e
a
tu
r
e
s
a
r
r
a
y.
RE
F
E
RE
NC
E
S
[1
]
B
.
Z
ah
ran
,
et
al
.
,
“A
Mo
d
i
fi
e
d
L
b
p
Me
t
h
o
d
t
o
E
x
t
ra
ct
Feat
u
re
s
fro
m
Co
l
o
r
Im
ag
e
s
,
”
Jo
u
r
n
a
l
o
f
T
h
eo
r
et
i
ca
l
a
n
d
A
p
p
l
i
ed
In
f
o
r
m
a
t
i
o
n
Tech
n
o
l
o
g
y
, v
o
l
.
9
6
,
n
o.
10
,
p
p
.
3
0
1
4
-
3
0
2
4
,
May
2
0
1
8
.
[2
]
M
.
A
.
Zu
g
h
o
u
l
,
et
al
.
,
“E
ffi
ci
en
t
Met
h
o
d
s
u
s
e
d
t
o
E
x
t
rac
t
Co
l
o
r
Imag
e
Feat
u
res
,
”
In
t
e
r
n
a
t
i
o
n
a
l
Jo
u
r
n
a
l
o
f
Co
m
p
u
t
e
r
S
ci
e
n
ce
a
n
d
M
o
b
i
l
e
Co
m
p
u
t
i
n
g
,
v
o
l
.
6
,
n
o.
1
2
,
p
p
.
7
-
1
4
,
D
ec
2
0
1
7
.
[3
]
A
.
Y
.
H
en
d
i
,
et
al
.
,
“A
N
o
v
e
l
Si
mp
l
e
an
d
H
i
g
h
l
y
Sec
u
re
Met
h
o
d
f
o
r
D
at
a
E
n
cr
y
p
t
i
o
n
-
D
ecry
p
t
i
o
n
,
”
In
t
e
r
n
a
t
i
o
n
a
l
Jo
u
r
n
a
l
o
f
Co
m
m
u
n
i
ca
t
i
o
n
Ne
t
wo
r
ks
a
n
d
I
n
f
o
r
m
a
t
i
o
n
S
ec
u
r
i
t
y
,
v
ol.
1
1
,
n
o
.
1
,
p
p
.
2
3
2
-
2
3
7
,
J
a
n
2
0
1
9
.
[4
]
M
.
O
.
Al
-
D
w
a
i
ri
,
et
al
.
,
“A
n
E
ffi
c
i
en
t
an
d
H
i
g
h
l
y
Secu
re
T
ech
n
i
q
u
e
t
o
E
n
cry
p
t
an
d
D
ecry
p
t
Co
l
o
r
Ima
g
es
,
”
E
n
g
i
n
eer
i
n
g
,
Tech
n
o
l
o
g
y
&
A
p
p
l
i
e
d
S
ci
e
n
ce
R
e
s
ea
r
ch
,
v
o
l
.
9
,
n
o.
3
,
p
p
.
4
1
6
5
-
4
1
6
8
,
J
u
n
2
0
1
9
.
[5
]
M.
H
ei
k
k
i
l
a
an
d
M.
Pi
e
t
i
k
ai
n
en
,
“
A
t
ex
t
u
re
Ba
s
ed
Me
t
h
o
d
fo
r
M
o
d
e
l
i
n
g
t
h
e
Back
g
ro
u
n
d
an
d
D
e
t
ect
i
n
g
Mo
v
i
n
g
O
b
j
ect
s
,
”
IE
E
E
Tr
a
n
s
a
ct
i
o
n
o
n
P
a
t
t
e
r
n
A
n
a
l
y
s
i
s
a
n
d
M
a
ch
i
n
e
In
t
e
l
l
i
g
e
n
ce
,
v
o
l
.
2
8
,
n
o
.
4
,
p
p
.
6
5
7
-
6
6
2
,
A
p
r
2
0
0
6
.
[6
]
M.
H
ei
k
k
i
l
a
,
et
a
l
.
,
“D
e
s
cri
p
t
i
o
n
o
f
i
n
t
ere
s
t
re
g
i
o
n
s
w
i
t
h
l
o
ca
l
b
i
n
ar
y
p
at
t
ern
s
,
”
P
a
t
t
e
r
n
R
ec
o
g
n
i
t
i
o
n
,
v
o
l
.
4
2
,
n
o
.
3
,
p
p
.
4
2
5
-
4
3
6
,
Mar
2
0
0
9
.
[7
]
Y
.
L
ee,
et
al
.
,
“H
i
erarch
i
cal
o
n
-
l
i
n
e
b
o
o
s
t
i
n
g
-
b
as
e
d
b
ack
g
ro
u
n
d
s
u
b
t
rac
t
i
o
n
,
”
2
0
1
1
1
7
t
h
Ko
r
ea
-
J
a
p
a
n
Jo
i
n
t
W
o
r
k
s
h
o
p
o
n
F
r
o
n
t
i
er
s
o
f
C
o
m
p
u
t
er
V
i
s
i
o
n
(F
C
V
)
,
U
l
s
an
,
p
p
.
1
-
5
,
2
0
1
1
.
[8
]
S
.
L
i
ao
,
et
al
.
,
“D
o
mi
n
an
t
l
o
ca
l
b
i
n
ar
y
p
at
t
er
n
s
fo
r
t
ex
t
u
re
Cl
as
s
i
ca
t
i
o
n
,
”
IE
E
E
T
r
a
n
s
a
ct
i
o
n
s
o
n
Im
a
g
e
P
r
o
ce
s
s
i
n
g
,
v
o
l
.
1
8
,
n
o
.
5
,
p
p
.
1
1
0
7
-
1
1
1
8
,
May
2
0
0
9
.
[9
]
S
.
L
i
ao
,
et
al
.
,
“Mo
d
el
i
n
g
p
i
x
e
l
p
ro
ce
s
s
w
i
t
h
s
cal
e
i
n
v
a
ri
an
t
l
o
cal
p
at
t
ern
s
fo
r
b
ac
k
g
r
o
u
n
d
s
u
b
t
rac
t
i
o
n
i
n
c
o
mp
l
ex
s
cen
e
s
,
”
IE
E
E
Co
m
p
u
t
e
r
S
o
c
i
et
y
Co
n
f
e
r
en
ce
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
ec
o
g
n
i
t
i
o
n
,
San
Fran
ci
s
co
,
CA
,
pp.
1
3
0
1
-
1
3
0
6
,
2
0
1
0
.
[1
0
]
Z
.
A
.
Al
q
a
d
i
an
d
H
.
M.
E
l
s
a
y
y
ed
,
“W
i
n
d
o
w
A
v
era
g
i
n
g
Met
h
o
d
t
o
Crea
t
e
a
Fea
t
u
re
V
i
c
t
o
r
fo
r
RG
B
Co
l
o
r
Imag
e
,
”
In
t
e
r
n
a
t
i
o
n
a
l
Jo
u
r
n
a
l
o
f
Co
m
p
u
t
e
r
S
c
i
en
ce
a
n
d
M
o
b
i
l
e
Co
m
p
u
t
i
n
g
,
v
o
l
.
6
,
n
o
.
2
,
p
p
.
6
0
-
6
6
,
Feb
2
0
1
7
.
[1
1
]
J
.
N
a
d
er,
et
a
l
.
,
“A
n
al
y
s
i
s
o
f
C
o
l
o
r
Imag
e
Fi
l
t
er
i
n
g
M
et
h
o
d
s
,
”
In
t
er
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
Co
m
p
u
t
e
r
A
p
p
l
i
ca
t
i
o
n
s
,
v
ol.
1
7
4
,
n
o.
8
,
p
p
.
1
2
-
1
7
,
Sep
2
0
1
7
.
[1
2
]
Z
.
A
.
A
l
q
ad
i
,
et
al
.
,
“T
ru
e
Co
l
o
r
Imag
e
E
n
h
a
n
cemen
t
U
s
i
n
g
M
o
r
p
h
o
l
o
g
i
cal
O
p
erat
i
o
n
s
,
”
In
t
e
r
n
a
t
i
o
n
a
l
R
evi
e
w
o
n
Co
m
p
u
t
er
s
&
S
o
f
t
wa
r
e
,
v
o
l
.
4
,
n
o
.
5
,
p
p
.
5
5
7
-
5
6
2
,
Sep
2
0
0
9
.
[1
3
]
Z
.
A
.
A
l
q
ad
i
,
et
a
l
.
,
“Z
J
ICD
al
g
o
ri
t
h
m
f
o
r
J
P
E
G
i
mag
e
c
o
mp
re
s
s
i
o
n
/
d
eco
m
p
res
s
i
o
n
,
”
D
i
g
i
t
a
l
P
r
o
ces
s
i
n
g
,
E
l
i
xi
r
D
i
g
i
t
a
l
P
r
o
ce
s
s
i
n
g
,
v
o
l
.
9
4
,
p
p
.
4
0
3
6
8
-
4
0
3
7
4
,
2
0
1
6
.
[1
4
]
M
.
O
.
Al
-
D
w
ai
r
i
,
et
al
.
,
“O
p
t
i
mi
ze
d
T
ru
e
-
C
o
l
o
r
Imag
e
Pro
ces
s
i
n
g
,
”
W
o
r
l
d
A
p
p
l
i
e
d
S
ci
e
n
ces
Jo
u
r
n
a
l
,
v
o
l
.
8
,
n
o
.
1
0
,
p
p
.
1
1
7
5
-
1
1
8
2
,
2
0
1
0
.
[1
5
]
Z
.
A
.
A
l
q
ad
i
,
et
al
.
,
“T
ru
e
Co
l
o
r
Imag
e
E
n
h
a
n
cemen
t
U
s
i
n
g
Mo
r
p
h
o
l
o
g
i
cal
O
p
erat
i
o
n
s
,
”
In
t
e
r
n
a
t
i
o
n
a
l
R
evi
e
w
o
n
Co
m
p
u
t
er
s
&
S
o
f
t
wa
r
e
,
v
o
l
.
4
,
n
o
.
5
,
p
p
.
5
5
7
-
5
6
2
,
Sep
2
0
0
9
.
[1
6
]
H
.
A
l
-
O
t
u
m
a
n
d
M.
A
l
-
D
w
a
i
ri
,
“Ima
g
e
C
o
mp
re
s
s
i
o
n
Ba
s
ed
o
n
C
o
l
o
r
M
o
rp
h
o
l
o
g
i
ca
l
Py
ram
i
d
a
l
D
eco
m
p
o
s
i
t
i
o
n
,
”
9
th
In
t
e
r
n
a
t
i
o
n
a
l
S
ym
p
o
s
i
u
m
o
n
S
i
g
n
a
l
P
r
o
ce
s
s
i
n
g
a
n
d
It
s
A
p
p
l
i
c
a
t
i
o
n
s
,
Sh
ar
j
ah
,
p
p
.
1
-
4
,
2
0
0
7
.
[1
7
]
R
.
S.
A
.
Z
n
ei
t
,
e
t
a
l
.
,
“A
Me
t
h
o
d
o
l
o
g
y
t
o
Creat
e
a
Fi
n
g
er
p
ri
n
t
f
o
r
RG
B
Co
l
o
r
Ima
g
e
,
”
In
t
er
n
a
t
i
o
n
a
l
Jo
u
r
n
a
l
o
f
Co
m
p
u
t
er
S
c
i
en
ce
a
n
d
M
o
b
i
l
e
Co
m
p
u
t
i
n
g
,
v
o
l
.
6
,
n
o
.
1
,
p
p
.
2
0
5
-
2
1
2
,
J
an
2
0
1
7
.
[1
8
]
Saj
i
d
a
P
.
,
et
al
.
,
“Rev
i
ew
o
n
L
o
ca
l
Bi
n
ar
y
Pat
t
er
n
(L
BP)
T
ex
t
u
re
D
es
cr
i
t
o
r
an
d
It
s
V
a
ri
a
n
t
s
,
”
In
t
er
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
A
d
va
n
ced
R
es
e
a
r
c
h
,
v
o
l
.
5
,
n
o
.
5
,
p
p
.
7
0
8
-
7
1
7
,
May
2
0
1
7
.
[1
9
]
M
.
O
.
Al
-
D
w
ai
r
i
,
et
al
.
,
“
A
n
ew
met
h
o
d
fo
r
v
o
i
ce
s
i
g
n
a
l
feat
u
res
creat
i
o
n
,
”
In
t
er
n
a
t
i
o
n
a
l
Jo
u
r
n
a
l
o
f
E
l
ec
t
r
i
ca
l
a
n
d
Co
m
p
u
t
er
E
n
g
i
n
ee
r
i
n
g
(IJ
E
CE
)
,
v
o
l
.
9
,
n
o
.
5
,
p
p
.
4
0
9
2
-
4
0
9
8
,
O
c
t
2
0
1
9
.
[2
0
]
A
.
Fern
an
d
ez,
et
al
.
,
“E
v
al
u
at
i
o
n
o
f
r
o
b
u
s
t
n
e
s
s
ag
a
i
n
s
t
ro
t
a
t
i
o
n
o
f
L
BP,
CCR
an
d
IL
BP
feat
u
res
i
n
g
ra
n
i
t
e
t
e
x
t
u
re
cl
as
s
i
f
i
cat
i
o
n
,
”
M
a
c
h
i
n
e
V
i
s
i
o
n
a
n
d
A
p
p
l
i
ca
t
i
o
n
s
,
v
o
l
.
2
2
,
n
o
.
6
,
p
p
.
9
1
3
-
9
2
6
,
N
o
v
2
0
1
1
.
[2
1
]
Z
.
G
u
o
,
et
al
.
,
“A
Co
mp
l
et
e
d
Mo
d
el
i
n
g
o
f
L
o
cal
Bi
n
ary
Pat
t
ern
O
p
era
t
o
r
fo
r
T
e
x
t
u
re
Cl
as
s
i
fi
ca
t
i
o
n
,
”
IE
E
E
Tr
a
n
s
a
ct
i
o
n
s
o
n
Im
a
g
e
P
r
o
ces
s
i
n
g
,
v
o
l
.
1
9
,
n
o
.
6
,
p
p
.
1
6
5
7
-
1
6
6
3
,
J
u
n
2
0
1
0
.
[2
2
]
A
.
Y
.
H
i
n
d
i
,
et
a
l
.
,
“
A
N
o
v
e
l
T
ec
h
n
i
q
u
e
fo
r
D
at
a
S
t
e
g
an
o
g
ra
p
h
y
,
”
E
n
g
i
n
eer
i
n
g
,
Tech
n
o
l
o
g
y
&
A
p
p
l
i
e
d
S
c
i
e
n
ce
R
es
e
a
r
c
h
,
v
o
l
.
9
,
n
o
.
6
,
p
p
.
4
9
4
2
-
4
9
4
5
,
2
0
1
9
.
[2
3
]
T
.
So
n
g
an
d
H
.
Li
,
“
W
av
eL
BP
b
as
e
d
feat
u
res
f
o
r
i
mag
e
cl
as
s
i
f
i
cat
i
o
n
,”
P
a
t
t
er
n
R
ec
o
g
n
i
t
i
o
n
Let
t
er
s
,
v
o
l
.
3
4
,
n
o
.
1
2
,
p
p
.
1
3
2
3
-
1
3
2
8
,
Sep
2
0
1
3
.
[2
4
]
K
.
M.
Mat
ro
u
k
,
et
al
.
,
“In
v
es
t
i
g
at
i
o
n
an
d
A
n
a
l
y
s
i
s
o
f
A
N
N
Paramet
er
s
,
”
E
u
r
o
p
ea
n
Jo
u
r
n
a
l
o
f
S
c
i
en
t
i
f
i
c
R
es
e
a
r
ch
,
v
o
l
.
1
2
1
,
n
o
.
2
,
p
p
.
2
1
7
-
2
2
5
,
J
a
n
2
0
1
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.