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.
1
,
F
e
br
ua
r
y
2020
,
pp.
500
~
510
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
i
s
tekdikti
,
De
c
r
e
e
No:
21/E
/KP
T
/2018
DO
I
:
10.
12928/
T
E
L
KO
M
NI
KA
.
v18i1.
15020
500
Jou
r
n
al
h
omepage
:
ht
tp:
//
jour
nal.
uad
.
ac
.
id/
index
.
php/T
E
L
K
OM
N
I
K
A
R
S
T
i
n
var
ia
n
t
w
at
e
r
m
ar
k
in
g
t
e
c
h
n
i
q
u
e
f
or
ve
c
t
o
r
m
a
p
b
ase
d
on
L
C
A
-
t
r
a
n
sf
or
m
S
aleh
AL
-
ar
d
h
i
,
Vij
e
y
T
h
ayan
an
t
h
a
n
,
Abd
u
ll
ah
B
as
u
h
ail
Facu
l
t
y
o
f
C
o
mp
u
t
i
n
g
an
d
I
n
fo
rma
t
i
o
n
T
ec
h
n
o
l
o
g
y
(FCI
T
)
,
K
i
n
g
A
b
d
u
l
az
i
z
U
n
i
v
ers
i
t
y
,
J
ed
d
ah
,
Sau
d
i
A
ra
b
i
a
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
M
a
y
1
9
,
2019
R
e
vis
e
d
De
c
4
,
2019
Ac
c
e
pted
De
c
19
,
2019
T
h
e
d
an
g
ers
o
f
c
o
p
y
ri
g
h
t
p
r
o
t
ect
i
o
n
can
i
m
p
act
2
D
v
ect
o
r
ma
p
s
,
h
av
i
n
g
a
k
n
o
ck
-
o
n
effect
o
n
t
h
e
u
s
e
o
f
v
ect
o
r
d
a
t
a.
T
o
ach
i
e
v
e
i
n
v
ar
i
an
ce
p
r
o
p
er
t
y
,
u
n
i
fo
rm
RS
T
(ro
t
at
i
o
n
,
s
ca
l
i
n
g
a
n
d
t
ran
s
l
a
t
i
o
n
)
an
d
d
i
s
g
u
i
s
i
n
g
t
h
e
d
i
g
i
t
al
v
ect
o
r
map
’
s
i
n
fo
rma
t
i
o
n
b
y
i
m
p
l
em
e
n
t
i
n
g
d
i
s
t
o
r
t
i
o
n
c
o
n
t
ro
l
,
i
s
a
l
l
d
o
n
e
b
y
u
s
i
n
g
w
at
ermar
k
i
n
g
s
c
h
emes
.
Co
n
v
er
t
an
o
ri
g
i
n
al
map
,
t
h
en
en
g
rai
n
t
h
e
w
at
ermar
k
.
A
n
L
CA
a
l
g
o
ri
t
h
m
i
s
u
s
e
d
i
n
t
h
i
s
s
t
u
d
y
,
as
a
n
ew
l
y
p
ro
p
o
s
ed
w
ay
t
o
p
r
o
t
ec
t
t
h
e
v
ec
t
o
r
map
s
u
n
d
er
c
o
p
y
ri
g
h
t
.
T
h
e
p
ro
ced
u
re
i
s
o
p
erat
e
d
i
n
t
h
i
s
o
rd
er
:
1
)
u
s
e
an
o
ri
g
i
n
al
map
,
al
t
ere
d
b
y
t
h
e
L
CA
al
g
o
ri
t
h
m
,
2
)
u
s
e
t
h
e
c
o
effi
c
i
en
t
o
f
t
h
e
t
ra
n
s
f
o
rmat
i
o
n
t
o
e
n
g
rai
n
t
h
e
w
at
erm
ark
,
i
n
s
ert
i
n
g
t
h
e
res
u
l
t
i
n
g
fre
q
u
en
c
y
i
n
t
o
t
h
e
L
SB
w
a
v
e
,
3)
t
h
e
w
a
t
ermark
e
d
map
i
s
acq
u
i
re
d
b
y
u
s
i
n
g
t
h
e
i
n
v
ers
e
L
CA
map
t
ra
n
s
f
o
rmat
i
o
n
.
Fu
rt
h
er
i
n
v
es
t
i
g
at
i
o
n
s
d
i
s
co
v
ered
t
h
at
t
h
e
n
ece
s
s
ar
y
s
t
an
d
a
rd
s
o
f
fi
d
el
i
t
y
a
n
d
i
n
v
i
s
i
b
i
l
i
t
y
can
b
e
ac
h
i
e
v
ed
u
s
i
n
g
t
h
i
s
t
ec
h
n
i
q
u
e.
T
h
i
s
p
ro
ce
d
u
re
al
s
o
g
i
v
es
o
u
t
n
u
mero
u
s
fre
q
u
e
n
cy
d
o
ma
i
n
s
fo
r
d
i
g
i
t
a
l
w
a
t
ermark
i
n
g
;
a
s
w
e
l
l
a
s
b
e
i
n
g
res
i
l
i
e
n
t
t
o
s
i
g
n
al
an
d
g
e
o
met
r
i
c
i
n
v
as
i
o
n
s
.
K
e
y
w
o
r
d
s
:
C
opyr
ight
pr
otec
ti
on
Ge
ometr
ic
a
tt
a
c
ks
L
inea
r
c
e
ll
ular
a
utom
a
ta
tr
a
ns
f
or
m
(
L
C
AT
)
T
r
a
ns
f
or
m
wa
ter
mar
king,
Ve
c
tor
map
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
:
S
a
leh
AL
-
a
r
dhi,
F
a
c
ult
y
of
C
omput
ing
a
nd
I
nf
or
mation
T
e
c
hnolog
y
(
F
C
I
T
)
,
King
Abdula
z
iz
Unive
r
s
it
y,
J
e
dda
h,
S
a
udi
Ar
a
bia
E
mail:
s
_a
r
dhi@hot
mail
.
c
om
1.
I
NT
RODU
C
T
I
ON
F
r
om
the
making
of
pa
pe
r
maps
to
digi
tal
da
ta
t
ype
s
,
we
ha
ve
s
e
e
n
the
p
r
ogr
e
s
s
ion
of
ge
os
pa
ti
a
l
da
ta
ove
r
the
pa
s
t
de
c
a
de
s
.
I
mpr
ove
ments
in
c
omput
e
r
ha
r
dwa
r
e
r
e
lative
to
ge
ogr
a
phic
da
ta
c
oll
e
c
t
ion
tool
s
ha
ve
c
a
us
e
d
thi
s
.
I
ts
pr
of
icie
nc
ies
include
GPS
(
ge
ogr
a
phic
pos
it
ioni
ng
s
ys
tems
)
with
s
a
telli
tes
a
ble
to
r
e
tr
ieve
tr
us
ted
s
pa
ti
a
l
pos
it
ioni
ng
da
ta
.
T
he
f
unc
ti
ons
of
a
na
log
da
ta
o
r
p
r
int
ha
s
be
e
n
r
e
plac
e
d
b
y
ve
c
tor
maps
a
s
t
he
s
tanda
r
d
da
ta
unit
s
in
GI
S
(
ge
ogr
a
phic
inf
or
mation
s
ys
tems
)
[
1]
.
An
e
s
s
e
nti
a
l
c
ompone
nt
of
GI
S
a
r
e
ve
c
tor
maps
,
a
type
o
f
GI
S
s
pa
ti
a
l
da
ta
w
it
h
many
dis
ti
nc
ti
ve
qua
li
ti
e
s
,
c
ove
r
s
many
a
r
e
a
s
f
r
om
na
vigation
to
tr
a
f
f
ic
da
ta.
C
ur
r
e
nt
inves
ti
ga
ti
ons
i
ntend
to
c
opyr
ight
pr
otec
t
the
owne
r
s
o
f
thes
e
s
ys
tems
,
a
s
they
be
c
ome
mor
e
popular
a
nd
us
e
d
a
c
r
os
s
the
boa
r
d.
I
n
c
ompar
is
on
to
mul
ti
media
da
ta
whic
h
ha
s
f
ixed
r
e
lative
pos
it
ions
,
gr
a
phics
ha
ve
s
e
ve
r
a
l
indepe
n
de
nt
c
omponents
.
Gr
a
phics
c
a
n
a
ls
o
make
c
opyi
ng
s
c
a
r
c
e
due
to
it
s
a
bunda
nc
e
o
f
topol
ogy
a
nd
e
nginee
r
ing
inf
o
r
mation.
T
he
ince
nti
ve
f
o
r
r
e
s
e
a
r
c
h
int
o
DW
(
digi
tal
wa
ter
mar
king)
f
or
dig
it
a
l
ve
c
tor
maps
,
c
a
n
be
f
or
med
f
r
om
c
ombi
ning
thes
e
f
a
c
to
r
s
[
2]
.
De
c
idi
ng
on
who
owns
the
digi
tal
map
a
nd
whe
ther
it
ha
s
va
li
d
it
y
,
is
de
ter
m
ined
by
the
digi
tal
map
s
take
holder
s
,
due
to
DW
.
Alter
a
ti
on
whe
n
the
med
ia
point
is
a
lt
e
r
e
d
a
nd
pr
otec
ti
on
a
ga
ins
t
da
ta
e
xtr
a
c
ti
on
a
r
e
the
r
obus
t
be
ne
f
it
s
o
f
DW
.
R
obus
t
DW
f
o
r
map
ve
c
tor
s
c
a
n
be
us
e
d
on
both
the
t
r
a
ns
f
or
mation
a
nd
s
pa
ti
a
l
dom
a
ins
,
a
nd
a
s
tr
ong
DW
a
ppli
c
a
ti
on
c
a
n
be
c
las
s
e
d
a
s
c
opyr
ight
pr
otec
ti
on
[
3]
.
Dis
c
r
e
te
wa
ve
let
tr
a
ns
f
or
m
(
DW
T
)
[
4]
,
dis
c
r
e
te
c
os
ine
t
r
a
ns
f
or
m
(
DC
T
)
[
5
]
,
a
nd
f
a
s
t
f
ou
r
ier
tr
a
ns
f
or
m
(
F
F
T
)
[
6]
a
r
e
a
ll
incl
ude
d
in
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
R
ST
invar
iant
w
ater
mar
k
ing
tec
hnique
for
v
e
c
tor
map
…
(
Saleh
A
L
-
ar
dhi
)
501
the
pr
incipa
l
tr
a
ns
f
or
m
a
lgo
r
it
hms
.
T
he
c
o
nve
nient
a
ppli
c
a
ti
on
o
f
the
s
pa
ti
a
l
domain
is
a
t
the
e
xpe
ns
e
of
r
obus
tnes
s
a
nd
invi
s
ibi
li
ty,
ther
e
f
or
e
to
c
opyr
ight
a
ppli
c
a
ti
ons
,
r
obus
t
DW
p
r
oc
e
dur
e
s
c
onc
e
ntr
a
te
on
the
tr
a
ns
f
o
r
mation
domain
[
7
-
9]
.
R
e
s
e
a
r
c
h
on
r
obus
t
DW
a
lgor
it
hms
in
the
t
r
a
ns
f
or
mation
domain
is
f
oc
us
e
d
on
im
a
ge
s
[
10
-
15]
a
nd
a
udio
[
16
-
20]
.
Ve
c
tor
maps
dif
f
e
r
f
r
om
mul
ti
m
e
dia
da
ta
be
c
a
us
e
they
a
r
e
not
li
mi
ted
to
f
unc
ti
ons
with
int
e
ge
r
s
,
howe
ve
r
dif
f
e
r
e
nt
methods
a
r
e
ne
e
de
d
to
e
nter
da
ta
int
o
a
ve
c
tor
map
.
I
W
T
(
int
e
g
e
r
wa
ve
t
r
a
ns
f
or
m
)
[
21]
,
DFT
[
2
]
a
nd
DC
T
[
22
,
23
]
c
a
n
b
e
us
e
d
by
r
obus
t
DW
in
the
ve
c
tor
map
tr
a
ns
f
or
m
a
ti
on
(
a
s
pr
oof
ha
s
s
hown)
,
ther
e
f
o
r
e
our
ne
w
p
r
oc
e
dur
e
is
c
ompar
e
d
to
the
DW
T
,
DC
T
a
nd
F
F
T
tec
h
niques
in
the
c
ur
r
e
nt
inves
ti
ga
ti
on.
F
or
e
a
c
h
c
o
-
or
dinate
s
e
que
nc
e
,
the
f
inal
12
digi
ts
of
the
de
c
im
a
l
a
r
e
take
n
a
nd
then
put
int
o
a
matr
ix
:
(
a
)
then
a
l
ter
e
d
int
o
4
ba
nds
,
(
b)
d
if
f
e
r
e
nt
ve
c
tor
da
ta
methods
a
s
s
oc
iate
d
with
c
opyr
ight
digi
tal
maps
ha
ve
be
e
n
e
xa
mi
ne
d,
[
24
-
26]
includi
ng
c
ha
nge
a
ble
wa
ter
mar
king
in
the
s
pa
ti
a
l
domain
a
nd
[
27]
wa
ter
mar
king
in
the
s
pa
ti
a
l
topol
ogy
domain.
T
he
pr
ogr
e
s
s
ion
of
c
opyr
igh
t
pr
o
tec
ti
ng
ve
c
tor
maps
is
the
a
im
of
thi
s
r
e
s
e
a
r
c
h.
I
mp
leme
ntation
of
the
L
C
A
tr
a
ns
f
or
mation
a
lgor
it
hm
is
the
a
dva
nc
e
d
tec
hnique
f
or
wa
ter
mar
king
ve
c
tor
maps
,
a
s
s
ugg
e
s
ted
by
thi
s
pa
pe
r
.
M
ult
im
e
dia
wa
ter
mar
king
ha
s
be
e
n
us
ing
c
e
ll
ular
a
utom
a
ta
tr
a
ns
f
or
m
(
C
AT
)
f
or
a
ve
r
y
long
ti
me
[
28
-
30]
.
E
mbedde
d
media
s
ti
ll
ha
s
no
r
e
s
e
a
r
c
h
r
e
s
ult
s
in
c
onne
c
ti
on
to
it
s
us
e
on
ve
c
tor
maps
.
Us
ing
th
e
L
C
A
a
lgor
it
hm
ha
s
the
be
ne
f
it
s
of
hi
gh
f
ideli
ty,
gr
e
a
t
ins
e
r
ti
on
r
e
s
ult
s
a
nd
high
invi
s
ibi
li
ty
[
31
]
.
Da
ta
ve
r
if
ica
ti
on
tr
a
it
s
a
r
e
p
r
e
s
e
nt
due
to
the
us
e
of
the
s
c
r
a
mbl
ing
method
[
25]
,
plus
it
is
not
li
mi
ted
to
one
tr
a
n
s
f
or
mation
plane
,
making
it
unique.
T
his
s
e
ts
it
a
bove
the
c
ur
r
e
nt
f
r
e
que
nc
y
wa
ter
mar
king
methods
a
nd
i
t
is
a
ble
to
a
c
c
e
ler
a
te
DW
via
it
s
mul
ti
-
f
r
e
que
nc
y
d
omains
.
Our
methods
f
or
the
pe
r
f
or
manc
e
mea
s
ur
e
ments
us
e
d
we
r
e
:
1)
NC
c
a
lcula
ti
on
wa
s
us
e
d
to
mea
s
ur
e
the
pe
r
f
or
manc
e
tec
hnique
,
2
)
e
va
luation
of
qua
li
ty,
ba
s
e
d
on
invi
s
ibi
l
it
y
with
R
M
S
E
c
a
lcula
ti
ons
,
3)
f
idelit
y
with
the
longes
t
dis
tanc
e
,
4)
R
e
s
is
tanc
e
a
nd
NC
c
a
lcula
ti
on
a
ga
ins
t
ge
ometr
ic
a
tt
a
c
ks
(
e
.
g.
R
S
T
)
.
T
he
ne
w
method
f
unc
ti
one
d
we
ll
,
a
s
is
e
vident
in
the
r
e
s
ult
s
of
the
invi
s
ibi
li
ty
tes
t,
whic
h
p
r
oduc
e
d
map
tes
t
da
ta
R
M
S
E
va
lues
.
NC
da
ta
a
nd
dis
tanc
e
c
ha
nge
we
r
e
withi
n
the
a
c
c
e
pted
thr
e
s
hold,
s
o
map
f
ide
li
ty
wa
s
high.
T
he
r
e
wa
s
a
high
leve
l
of
r
e
s
il
ienc
e
a
ga
ins
t
ge
ometr
ic
a
tt
a
c
ks
,
e
s
pe
c
ially
f
or
the
wa
ter
mar
k
a
s
int
e
gr
it
y
wa
s
e
s
s
e
nti
a
l.
T
he
r
e
a
r
e
f
our
e
xtr
a
s
e
c
ti
ons
a
t
t
he
e
nd
of
thi
s
pa
pe
r
.
An
e
xplana
ti
on
of
the
inv
e
s
ti
ga
ti
on
pr
oc
e
dur
e
s
us
e
d
in
thi
s
pa
pe
r
a
r
e
doc
umente
d
in
s
e
c
ti
on
2,
f
ol
lowe
d
by
the
r
e
s
ult
s
in
s
e
c
ti
on
3
,
then
s
e
c
ti
on
4
ha
s
a
c
onc
lus
ion
of
the
s
tudy.
2.
RE
S
E
AR
CH
M
E
T
HO
D
2
.
1.
L
in
e
ar
c
e
ll
u
lar
au
t
om
at
a
A
dyna
mi
c
a
l
in
a
dis
c
onti
nuous
f
r
e
que
nc
y
a
nd
ti
me
f
r
a
me,
s
ymbol
ize
d
by
the
L
C
A
tr
a
ns
f
or
m
is
s
hown
be
low
in
(
1)
.
E
a
c
h
c
e
ll
ha
s
a
r
e
s
tr
icte
d
g
r
o
up
of
s
tate
s
a
nd
f
o
r
m
a
latt
ice
s
tr
uc
tur
e
.
E
f
f
e
c
tual
a
nd
f
a
s
t
c
a
lcula
ti
ons
of
the
dis
c
r
e
te
tr
a
ns
f
o
r
mation
a
r
e
e
na
bled
by
the
a
lgo
r
it
hm
[
31]
.
(
C
t
+
1
)
T
=
M
n
.
(
C
t
)
T
(
m
o
d
2
)
(
1)
M
n
r
e
pr
e
s
e
nts
the
loca
l
t
r
a
ns
it
ion
matr
ix
,
given
that
n
=
5
k
:
M
n
=
(
1
1
1
0
0
…
…
…
…
0
0
0
1
1
1
1
0
…
…
…
…
0
0
0
1
1
1
1
1
…
…
…
…
0
0
0
0
1
1
1
1
…
…
…
…
0
0
0
0
0
1
1
1
…
…
…
…
0
0
0
…
.
.
…
.
0
0
0
0
0
…
…
…
…
1
1
1
0
0
0
0
0
…
…
…
…
1
1
1
0
0
0
0
0
…
…
…
…
1
1
1
)
if
the
tr
a
ns
it
ion
matr
ix
of
a
c
e
ll
ular
a
utom
a
ti
on
(
An
)
wa
s
r
e
pr
e
s
e
nted
by
M
n
,
the
non
-
z
e
r
o
c
oe
f
f
icie
nts
will
be
c
ome
1
a
s
An
is
now
the
nth
o
r
de
r
pe
nta
-
diagona
l
matr
ix.
(
C
t
)
T
on
the
other
ha
nd
,
is
a
t
r
a
ns
pos
it
ion
of
a
li
ne
a
r
matr
ix
,
c
ompos
e
d
of
int
e
r
-
c
ha
nging
a
nd
r
a
n
dom
bi
na
r
y
numbe
r
s
a
r
e
s
hown
in
(
2)
.
(
C
t
)
T
=
M
n
−
1
.
(
C
t
+
1
)
T
(
m
o
d
2
)
(
2)
T
he
f
or
mul
a
be
low
is
tr
a
ns
it
ion
matr
ix
f
o
r
the
r
e
ve
r
s
e
of
An
,
given
that
n
=
5
k
:
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
.
1
,
F
e
br
ua
r
y
2020
:
500
-
510
502
M
n
−
1
=
(
M
5
−
1
B
B
⋯
B
B
T
M
5
−
1
B
⋱
⋮
B
T
A
T
⋱
⋱
⋮
⋮
⋱
⋱
M
5
−
1
B
B
T
⋯
B
T
B
T
M
5
−
1
)
,
whe
r
e
M
5
−
1
|
|
0
0
1
1
0
0
0
0
1
1
1
0
1
0
1
1
1
0
0
0
0
1
1
0
0
|
|
(
mod
2)
,
B
=
(
0
0
1
1
0
0
0
0
1
1
0
0
1
0
1
0
0
0
0
0
0
0
0
0
0
)
.
|
M
n
|
m
o
d
2
,
r
e
pr
e
s
e
nti
ng
the
t
r
a
ns
it
ion
matr
ix
s
tar
ts
with
f
ive
number
s
a
r
e
s
hown
be
low
in
(
3)
.
|
M
n
|
m
o
d
2
=
{
1
,
if
n
=
5k
or
n
=
5k
+
1
,
w
it
h
k
∈
N
0
,
o
th
e
r
w
i
s
e
}
(
(
3)
2.
2.
L
in
e
ar
c
e
ll
u
lar
au
t
om
at
a
t
r
an
s
f
or
m
T
his
inves
ti
ga
ti
on
invol
ve
d
doing
the
wa
ter
ma
r
k
e
mbedding
pr
oc
e
s
s
on
the
ve
c
tor
map
,
whic
h
include
d
the
tr
a
ns
f
or
mation
f
r
a
me
of
the
c
oo
r
dinate
s
of
the
ve
r
ti
c
e
s
.
T
his
pr
oc
e
dur
e
wa
s
us
e
d
on
the
c
oe
f
f
icie
nt
of
the
tr
a
ns
f
or
mation
r
e
s
ult
f
r
e
que
nc
y
of
the
ve
c
tor
map’
s
da
ta
.
F
igur
e
1,
s
h
ows
that
the
c
oor
dinate
s
a
r
e
c
o
nve
r
ted
int
o
a
n
L
C
A
tr
a
ns
f
or
m,
s
o
that
the
ve
c
tor
map
c
a
n
be
t
r
a
ns
late
d
int
o
the
f
r
e
que
nc
y
domain
of
a
s
ignal.
T
his
is
f
u
r
ther
s
uppor
ted
by
(
4
)
,
whic
h
s
hows
that
the
L
C
A
t
r
a
ns
f
or
m,
c
a
n
c
onve
r
t
the
hos
t
map’
s
v
x1
c
oor
dinate
.
F
igur
e
1.
Vis
ua
li
z
a
ti
on
of
L
C
A
tr
a
ns
f
or
m
(
)
=
∑
−
1
=
0
.
1
(
2
)
(
4)
w
he
r
e
,
T
(
M
)
=
hos
t
map's
domain
tr
a
ns
f
or
mat
ion
va
lue
,
M
n
=
L
C
A's
tr
a
ns
it
ion
matr
ix
,
v
x1
=
hos
t
map^
'
s
digi
tal
media
va
lue,
N
=
number
o
f
ve
r
t
ice
s
a
lt
e
r
e
d
b
y
the
f
r
e
que
nc
y
domain.
Note
:
in
(
5
)
s
hows
that
the
v
x1
co
-
or
dinate
unde
r
take
s
tr
a
ns
f
or
mation
with
the
L
C
A
tr
a
n
s
f
or
m,
a
ll
owin
g
f
or
the
e
nc
r
ypted
wa
ter
mar
k
pa
r
t
to
be
put
int
o
the
e
qu
a
ti
on
N
,
N
,
v
x1
,
T
(
M
)
,
M
n
.
v
x1
′′
=
v
x1
′
+
α
W
(
5)
w
he
r
e
,
α
=
e
mbedding
pa
r
a
mete
r
,
W
=
wa
ter
mar
k
pa
r
t.
Va
r
iations
of
the
ve
c
tor
map
a
r
e
dir
e
c
tl
y
pr
opor
ti
ona
l
to
the
e
mbedding
pa
r
a
mete
r
(
α
)
,
mea
nwhile
the
wa
ter
mar
k’
s
(
W
)
r
e
s
is
tanc
e
s
r
is
e
s
.
T
his
e
qua
ti
on
us
e
s
a
c
c
e
ptable
a
lt
e
r
a
ti
ons
to
the
ve
c
tor
map
,
a
huge
r
e
s
is
tanc
e
va
lue
a
nd
3
-
pa
r
t
α
va
lues
.
T
he
inve
r
s
e
o
f
the
L
C
A
tr
a
n
s
f
or
m
is
given
in
the
f
oll
owing
(
6)
:
In
6
be
low
,
r
e
pr
e
s
e
nts
the
L
C
A
tr
a
ns
f
or
m’
s
inver
s
e
:
v
x1
′′
,
(
)
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
R
ST
invar
iant
w
ater
mar
k
ing
tec
hnique
for
v
e
c
tor
map
…
(
Saleh
A
L
-
ar
dhi
)
503
iT
(
M
)
=
∑
N
−
1
n
=
0
M
n
−
1
.
v
x1
′′
(
m
o
d
2
)
(
6)
whe
r
e
:
iT
(
M
)
=
hos
t
map's
inver
s
e
domain
tr
a
ns
f
or
mation
va
lue
,
v
x1
′′
=
digi
tal
media
va
lue
of
thehos
t
map's
tr
a
ns
f
or
mation
.
2.
3.
Wa
t
e
r
m
ar
k
e
m
b
e
d
d
in
g
p
h
as
e
T
he
s
ugge
s
ted
s
c
he
me
is
a
ppli
e
d
in
the
f
r
e
que
nc
y
domain
a
nd
the
digi
tal
wa
ter
mar
k's
e
mbedde
d
model
is
r
e
pr
e
s
e
nted
in
F
igur
e
2.
T
he
wa
ter
mar
k
i
ns
e
r
ti
on
pr
oc
e
s
s
c
ons
is
ts
of
thr
e
e
s
teps
,
whe
n
a
pu
bli
c
ke
y
is
us
e
d
to
e
nc
r
ypt
(
thr
e
e
pa
r
ts
,
na
mely
a
ve
c
to
r
ma
p,
the
s
ize
of
L
C
A
tr
a
ns
it
ion
matr
ix
(
)
a
nd
a
wa
ter
m
a
r
k
that
s
c
r
a
mbl
e
s
the
e
leme
nts
)
a
nd
whe
n
a
pr
ivate
ke
y
is
us
e
d
to
de
c
r
ypt
.
T
he
pr
opos
e
d
s
c
he
me
r
e
li
e
s
on
the
li
ne
a
r
c
e
ll
ular
a
utom
a
ta
tr
a
ns
f
o
r
m
(
L
C
AT
)
a
lgo
r
it
hm,
a
nd
it
ope
r
a
tes
in
the
f
r
e
que
nc
y
domain.
Addi
ti
ona
ll
y,
the
pr
opos
e
d
s
c
he
me
doe
s
not
f
a
ll
unde
r
ge
ometr
ic
a
tt
a
c
ks
,
na
mely,
tr
a
ns
lation,
s
c
a
li
ng,
a
nd
r
otat
ion.
T
he
a
lgor
it
hm
of
the
s
c
he
me
is
a
s
f
oll
ows
:
F
igur
e
2
dis
plays
the
r
e
c
omm
e
nde
d
s
c
he
me,
us
e
d
in
the
digi
tal
wa
ter
mar
k’
s
e
mbedde
d
model
a
nd
th
e
f
r
e
que
nc
y
domain
.
T
he
r
e
a
r
e
th
r
e
e
ke
y
s
teps
to
wa
ter
mar
k
ins
e
r
ti
on;
whe
n
a
publi
c
ke
y
is
us
e
d
to
e
nc
r
ypt
na
mely
the
s
ize
of
,
wa
ter
mar
k
that
mi
xe
s
up
e
lem
e
nts
a
nd
a
ve
c
tor
map.
T
his
c
a
n
ha
ppe
n
whe
n
a
pr
i
va
te
ke
y
is
us
e
d
to
de
c
r
ypt.
T
his
p
r
ogr
a
m
ope
r
a
tes
in
the
f
r
e
que
nc
y
domain
a
nd
de
pe
nds
on
the
L
C
AT
(
li
ne
a
r
c
e
ll
ular
a
utom
a
ta
tr
a
ns
f
or
m
)
a
lgor
it
hm
.
T
he
pr
ogr
a
m
is
a
ls
o
is
not
s
us
c
e
pti
ble
to
R
S
T
(
R
otation,
S
c
a
li
ng
a
nd
T
r
a
ns
lation)
,
whic
h
a
r
e
c
ons
ider
e
d
to
be
ge
ometr
ic
a
tt
a
c
ks
.
T
he
a
lgo
r
it
hm
goe
s
li
ke
thi
s:
−
S
e
lec
t
two
r
e
f
e
r
e
nc
e
ve
r
ti
c
e
s
,
v
f1
a
nd
v
f2
in
the
r
a
nge
(
1
≤
v
f1
,
v
f2
≤
n
)
a
s
the
ve
c
tor
map
of
M
,
i
n
or
de
r
to
a
s
s
ur
e
s
e
c
ur
it
y.
−
W
he
n
length
(
N)
is
t
r
a
ns
f
or
med
int
o
a
do
main
f
r
e
que
nc
y
(
with
no
r
e
f
e
r
e
nc
e
s
)
,
de
c
ide
on
the
numbe
r
o
f
ve
r
ti
c
e
s
in
the
map
f
il
e
(
M
)
.
−
Co
-
or
dinate
s
f
a
r
e
take
n
f
r
om
the
ve
r
tex
a
nd
c
onve
r
ted
int
o
a
L
C
AT
tr
a
ns
f
or
m
−
Us
ing
e
qua
ti
on
25
,
e
nc
r
ypt
the
f
a
c
tor
s
of
W
∗
,
whic
h
p
r
oduc
e
s
the
da
ta
s
e
que
nc
e
W
∗
=
{
w
i
∗
|
w
i
∗
∈
{0
,
1}
,
i
=
0,
1,
.
.
.
.
,
l
–
1}
−
P
ut
W
∗
int
o
the
f
in
a
l
two
c
ons
e
c
uti
ve
digi
ts
(
r
e
duc
e
s
the
im
pa
c
t
on
the
p
r
e
c
is
ion)
a
nd
a
s
s
ume
that
a
double
f
loating
point
16
-
digi
t
c
oor
dinate
va
lue
in
a
de
c
im
a
l
f
r
a
c
ti
ona
l
ve
r
s
ion.
T
he
e
mbedde
d
va
lue
is
in
the
r
a
nge
of
0
to
99
a
nd
doe
s
not
c
or
r
e
late
with
w
i
∗
.
I
f
we
a
s
s
ume
t
ha
t
the
int
e
ge
r
D
is
c
ompos
e
d
of
the
two
digi
ts
,
then:
W
∗
=
{
if
w
i
∗
is
0
th
e
n
D
≤
50
a
nd
s
a
v
e
d
at
th
e
p
o
s
it
io
ns
;
w
i
∗
=
1
,
o
th
e
r
w
is
e
}
(
7)
−
T
he
ini
t
ial
s
ha
pe
f
il
e
(
with
the
inver
s
e
L
C
AT
)
of
the
f
r
e
que
nc
y
domain
ve
c
tor
map
is
r
e
ins
tate
d
a
f
ter
the
wa
ter
mar
k
is
plac
e
d.
2.
4
.
Wa
t
e
r
m
ar
k
e
xt
r
ac
t
in
g
p
h
as
e
T
he
wa
ter
mar
k
e
xtr
a
c
ti
on
poin
t
a
nd
the
ins
e
r
ti
on
pr
oc
e
s
s
ha
ve
s
im
il
a
r
s
teps
,
but
in
the
oppos
ing
s
e
que
nc
e
.
T
he
thr
e
e
s
teps
us
e
d
in
the
e
xe
r
ti
on
pr
o
c
e
s
s
a
r
e
the
r
e
s
ult
s
of
the
ins
e
r
ti
on
p
r
oc
e
s
s
,
na
mely
the
two
r
e
f
e
r
e
nc
e
ve
r
ti
c
e
s
vf1
a
nd
vf2
,
f
ixed
s
ize
L
C
A
tr
a
ns
it
ion
matr
ix
(
Mn
)
,
a
nd
the
wa
ter
mar
ke
d
ve
c
tor
map,
a
s
s
hown
in
F
igu
r
e
3
.
I
n
the
opp
os
it
e
s
e
que
nc
e
,
the
wa
ter
mar
k
ins
e
r
ti
on
pr
oc
e
dur
e
a
nd
e
xt
r
a
c
ti
on
po
int
ha
ve
c
ompar
a
ble
pha
s
e
s
.
T
he
outcome
o
f
the
ins
e
r
ti
on
pr
oc
e
dur
e
,
p
r
oduc
e
s
thr
e
e
s
teps
f
or
the
e
xe
r
ti
on
pr
oc
e
dur
e
.
T
he
r
e
s
ult
s
a
r
e
Mn
(
f
ixed
s
ize
L
C
A
tr
a
ns
it
ion
ma
tr
ix
)
,
wa
ter
mar
ke
d
ve
c
to
r
map
F
igur
e
3
a
nd
two
r
e
f
e
r
e
nc
e
ve
r
ti
c
e
s
(
vf1
,
vf2
)
.
−
S
e
lec
t
v
f1
a
nd
v
f2
(
1
≤
v
f1
,
v
f2
≤
n
)
,
unde
r
pr
ivate
ke
y
k’
s
c
ontr
ol,
a
s
the
r
e
f
e
r
e
nc
e
ve
r
ti
c
e
s
f
o
r
the
ve
c
tor
map
o
f
M
−
De
ter
mi
ne
the
number
of
ve
r
ti
c
e
s
in
the
map
f
il
e
M
a
nd
the
length
(
N)
that
will
lat
e
r
be
tr
a
ns
f
o
r
med
int
o
a
domain
f
r
e
que
nc
y
without
the
r
e
f
e
r
e
nc
e
s
−
Onc
e
the
s
e
t
of
c
oor
dinate
s
f
or
e
a
c
h
f
e
a
tur
e
is
obta
ined,
it
will
be
tr
a
ns
f
or
med
int
o
a
L
C
A
T
tr
a
ns
f
or
m
−
T
he
e
mbedde
d
wa
ter
mar
k
loca
ti
on
a
nd
wa
ter
ma
r
k
bit
s
a
r
e
e
xtr
a
c
ted
us
ing
(
8)
:
W
∗
=
{
if
D
≤
50
th
e
n
w
i
∗
is
0
w
i
∗
=
1
,
o
th
e
r
w
is
e
}
(
8)
−
E
xtr
a
c
t
the
o
r
igi
na
l
e
mbedde
d
wa
ter
mar
k
s
e
que
nc
e
W
,
with
p
r
ivate
ke
y
k,
by
inver
s
ing
the
wa
ter
mar
k
pa
tt
e
r
n
.
−
Ge
t
the
wa
ter
mar
k
by
r
e
buil
ding
the
wa
ter
ma
r
k
pa
tt
e
r
n.
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
.
1
,
F
e
br
ua
r
y
2020
:
500
-
510
504
−
S
e
lec
t
two
r
e
f
e
r
e
nc
e
ve
r
ti
c
e
s
,
v
f1
a
nd
v
f2
in
the
r
a
nge
(
1
≤
v
f1
,
v
f2
≤
n
)
a
s
the
ve
c
tor
map
of
M
,
i
n
or
de
r
to
a
s
s
ur
e
s
e
c
ur
it
y.
−
W
he
n
length
(
N)
is
t
r
a
ns
f
or
med
int
o
a
do
main
f
r
e
que
nc
y
(
with
no
r
e
f
e
r
e
nc
e
s
)
,
de
c
ide
on
the
numbe
r
o
f
ve
r
ti
c
e
s
in
the
map
f
il
e
(
M
)
.
−
Co
-
or
dinate
s
of
e
a
c
h
f
e
a
tur
e
a
r
e
ga
ined
a
nd
then
take
n
f
r
om
the
ve
r
tex
a
nd
c
onve
r
ted
int
o
a
L
C
AT
tr
a
ns
f
or
m
.
−
T
he
(
8)
is
us
e
d
to
e
xtr
a
c
t
the
e
mbedde
d
wa
ter
mar
k
e
d
pa
r
ts
a
nd
loca
ti
on
.
W
∗
=
{
if
D
≤
50
th
e
n
w
i
∗
is
0
w
i
∗
=
1
,
o
th
e
r
w
is
e
}
(
8)
−
R
e
ve
r
s
e
the
wa
ter
mar
k
pa
tt
e
r
n,
s
o
that
you
c
a
n
us
e
the
pr
ivate
ke
y
(
k
)
to
e
xtr
a
c
t
the
or
igi
na
l
e
mbedde
d
wa
ter
mar
k
s
e
que
nc
e
(
W
)
.
−
R
e
c
ons
tr
uc
t
the
wa
ter
mar
k
pa
tt
e
r
n
,
in
or
de
r
to
ge
t
t
he
wa
ter
mar
k
.
F
igur
e
2.
W
a
ter
mar
k
e
mbedding
F
igur
e
3.
W
a
ter
mar
k
e
xtr
a
c
ti
on
3.
RE
S
UL
T
S
A
ND
DI
S
CU
S
S
I
ON
3.
1.
E
xp
e
r
im
e
n
t
al
r
e
s
u
lt
s
F
or
the
e
va
luation
o
f
da
ta
type
(
.
s
hp)
,
ther
e
a
r
e
two
s
ha
pe
f
il
e
maps
.
T
wo
ve
c
tor
maps
make
up
the
f
il
e
type
E
S
R
I
s
tanda
r
d
[
32
]
,
whic
h
a
r
e
a
s
pot
he
ight
a
nd
c
oa
s
tl
ine
map
of
T
a
ylo
r
R
ooke
r
y
[
33
]
.
C
opyr
ight
mar
ke
r
wa
s
r
e
p
r
e
s
e
nted
by
a
bit
map
pictur
e
.
T
he
P
C
us
e
d
f
or
th
is
s
tudy
wa
s
W
indows
10
pr
of
e
s
s
ional,
QG
I
S
ve
r
s
ion
3.
0
,
16
GB
memor
y
a
nd
2.
3
GH
z
.
P
r
og
r
a
ms
us
e
d
we
r
e
python
a
nd
M
AT
L
AB
.
T
he
s
e
c
r
e
t
pa
r
ts
a
s
s
oc
iate
d
with
e
a
c
h
tr
a
n
s
f
or
m
c
oor
dinate
ha
d
a
M
n
=
30
,
α
in
LSB
a
nd
T
=
1
f
or
it
e
r
a
ti
ve
e
mbedding.
T
he
invi
s
ibi
li
ty
o
f
our
a
pp
r
oa
c
h
wa
s
dis
playe
d
in
the
f
ir
s
t
tes
t.
F
igu
r
e
4
s
hows
ve
c
t
or
maps
wa
ter
mar
ke
d,
us
ing
the
method
s
ugge
s
ted
in
s
e
c
ti
on
3.
5.
T
his
p
r
oduc
e
d
the
wa
ter
mar
ke
d
types
dis
playe
d
in
F
igur
e
5.
T
he
r
e
c
omm
e
nde
d
pe
r
f
or
manc
e
tec
hnique
a
na
lys
is
f
or
thi
s
s
tudy
wa
s
mea
s
ur
e
d
by
the
NC
c
a
lcula
ti
on.
I
t
wa
s
a
ls
o
us
e
d
to
e
xa
mi
ne
the
s
im
il
a
r
it
ies
be
twe
e
n
the
or
ig
inal
wa
ter
mar
k
,
be
f
o
r
e
a
nd
a
f
ter
the
e
xtr
a
c
ti
on
(
va
lues
r
a
nging
0
-
1)
.
T
he
s
uc
c
e
s
s
of
the
wa
ter
mar
king
us
a
ge
is
indi
c
a
ted
by
how
hig
h
the
NC
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
R
ST
invar
iant
w
ater
mar
k
ing
tec
hnique
for
v
e
c
tor
map
…
(
Saleh
A
L
-
ar
dhi
)
505
va
lue
is
.
T
he
lar
ge
r
the
NC
va
lue
is
,
the
mor
e
s
im
i
lar
it
y
ther
e
is
be
twe
e
n
im
a
ge
s
.
I
n
(
9)
,
w
(
ini
t
ial
va
lue)
a
nd
w
i
∗
a
r
e
the
e
xtr
a
c
ted
wa
ter
mar
ks
.
T
a
ble
1
s
hows
that
the
or
igi
na
l
a
nd
e
xtr
a
c
ted
wa
ter
mar
ks
a
r
e
the
s
a
me,
a
s
the
e
va
luation
r
e
s
ult
f
or
NC
is
a
ppr
oxim
a
tely
1,
plus
they
ha
ve
the
s
a
me
wa
ter
mar
k
c
ontent
a
n
d
length.
T
his
pr
oc
e
dur
e
wa
s
a
ble
to
input
c
opyr
ight
s
a
s
a
wa
ter
mar
k
(
without
les
s
e
ning
it
s
qua
li
ty)
,
a
s
a
wa
ter
mar
k
e
xtr
a
c
ted
a
ga
in
f
r
om
the
ve
c
tor
map
f
il
e
,
doe
s
not
ne
e
d
to
go
th
r
ough
dim
e
ns
ions
or
c
ontent
c
ha
nge
s
.
(
a
)
(
b)
F
igur
e
4.
T
e
s
t
2D
ve
c
tor
maps
:
(
a
)
s
pot
he
ight
map
of
T
a
ylor
R
ooke
r
y,
(
b)
c
oa
s
tl
ine
map
of
T
a
ylor
R
ooke
r
y
(
a
)
(
b)
F
igur
e
5.
T
he
wa
ter
mar
ke
d
2D
ve
c
tor
maps
o
f
F
ig
ur
e
4
map
NC
=
∑
w
i
M
i
=
0
X
w
i
∗
√
∑
(
w
i
M
i
=
0
)
2
X
√
∑
(
w
i
∗
)
2
M
i
=
0
(
9)
T
a
ble
1.
R
e
s
ult
of
s
im
il
a
r
i
ty
tes
t
be
twe
e
n
or
igi
na
l
wa
ter
mar
ks
a
nd
e
xtr
a
c
ted
wa
ter
mar
k
M
a
p
M
a
p T
ype
F
e
a
tu
r
e
s
/v
e
r
ti
c
e
s
O
r
ig
in
a
l
W
a
te
r
ma
r
k
E
xt
r
a
c
te
d
W
a
te
r
ma
r
k
NC
a
P
oi
nt
355/
355
0.998909
b
P
ol
yl
in
e
18/
4279
0.998308
3.
2.
I
n
vis
ib
il
it
y
e
valu
a
t
ion
T
a
ble
2
s
hows
the
two
pa
r
a
mete
r
s
that
we
r
e
us
e
d
pr
e
vious
ly
in
the
invi
s
ibi
li
ty
mea
s
ur
e
ment
(
a
s
r
e
f
e
r
e
nc
e
a
na
lys
is
f
or
c
a
lcula
ti
ng
R
M
S
E
)
,
whic
h
s
ubs
e
que
ntl
y
de
c
ides
on
the
a
lt
e
r
a
ti
on
be
twe
e
n
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
.
1
,
F
e
br
ua
r
y
2020
:
500
-
510
506
the
int
e
r
polate
d
wa
ter
mar
k’
s
r
e
s
ult
s
a
nd
the
be
g
inni
ng
of
the
ma
p
f
il
e
.
T
he
e
qua
ti
on
f
or
the
m
a
king
of
the
R
M
S
E
,
is
s
hown
be
low:
N
,
v
x1
∗
′′
R
M
S
E
=
(
1
∑
|
=
0
−
∗
′′
|
)
(
10)
whe
r
e
,
=
the
number
o
f
ve
r
tex
map
ve
c
tor
s
,
1
=
e
quivale
nt
x
c
oor
dinate
s
in
the
ini
ti
a
l
ve
c
tor
m
a
p.
1
∗
′′
=
e
quivale
nt
x
c
oor
dinate
s
in
the
r
e
c
ove
r
e
d
ve
c
tor
map
.
T
a
ble
2.
R
M
S
E
s
be
twe
e
n
the
r
e
c
ove
r
e
d
maps
a
nd
t
he
or
igi
na
l
maps
(
×
10
-
9)
(
T
=
1)
M
a
p
D
C
T
D
W
T
FFT
P
r
opos
e
d (
L
C
A
T
)
a
5.0917×
10−
4
1.
3014
×
10−
4
2.9023×
10−
5
1
.
1317
×
10−
9
b
1.8514×
10−
3
3.5872
×
10−
4
2.3342×
10−
5
3.1562×
10−
9
3.
3.
F
id
e
li
t
y
e
valu
a
t
ion
Digit
a
l
wa
ter
mar
king
is
untr
a
c
e
a
ble
by
human
s
e
ns
e
s
,
making
it
a
ve
r
y
r
e
li
a
ble
s
ys
tem,
plus
it
doe
s
n’
t
s
igni
f
ica
ntl
y
wor
s
e
n
the
media
f
il
e
int
e
r
polation.
M
or
e
ove
r
,
with
r
e
ga
r
ds
to
R
M
S
E
,
the
f
ur
thes
t
c
ha
nge
s
c
a
n
be
ga
uge
d.
A
pos
it
ion
c
ha
nge
is
indi
c
a
ted
by
the
f
u
r
thes
t
dis
tanc
e
,
whic
h
is
c
a
us
e
d
by
wa
ter
mar
k
int
e
r
polation
int
o
ve
c
tor
f
i
les
.
T
his
is
g
a
ined
via
a
n
a
s
s
e
s
s
ment
c
ompar
ing
the
or
igi
na
l
ve
c
tor
map
f
il
e
,
ve
c
tor
map
f
i
le
(
c
ontaining
the
wa
ter
mar
k)
a
n
d
c
oor
dinate
d
ve
r
tex.
T
o
c
onve
r
t
the
f
ur
thes
t
dis
ta
nc
e
int
o
mete
r
s
,
QG
I
S
is
im
pleme
nted
.
T
he
lengthies
t
noti
c
e
a
ble
s
hif
t
in
da
ta
a
na
lys
is
is
60
c
m,
a
s
s
e
e
n
in
T
a
ble
3,
ther
e
f
or
e
the
f
ur
thes
t
dis
tanc
e
va
lue
s
ti
ll
maintains
a
s
igni
f
ica
nc
e
leve
l
of
pr
e
c
is
ion
in
the
ve
c
tor
ma
p,
whic
h
is
dis
playe
d
in
F
igur
e
6.
T
a
ble
3.
R
e
s
ult
of
f
idelit
y
tes
t
be
twe
e
n
o
r
igi
na
l
ma
ps
with
wa
ter
mar
ke
d
mas
M
a
p
τ (
m)
D
C
T
D
W
T
FFT
F
a
r
th
e
s
t
D
is
ta
nc
e
(
me
te
r
)
L
C
A
T
a
0.5
0.45
0.4
0.3
0.050
b
0.5
0.5
0
.43
0
.38
0.068
(
a
)
(
b)
F
igur
e
6.
Or
igi
na
l
a
nd
wa
ter
mar
ke
d
map
ove
r
laid
with
e
a
c
h
other
3.
4
.
Ro
b
u
s
t
n
e
s
s
e
valu
at
ion
E
xa
mi
ning
the
s
tr
e
ngth
o
f
the
wa
ter
mar
k
is
the
ne
xt
s
tage
.
Adva
nc
e
d
methods
a
r
e
us
e
d
to
c
ha
ll
e
nge
the
thr
e
e
ge
ometr
ic
a
tt
a
c
ks
,
c
las
s
e
d
a
s
R
S
T
(
r
otati
on,
s
c
a
li
ng
a
nd
t
r
a
ns
lation
)
.
T
he
R
S
T
a
ls
o
include
s
a
s
ignal
ope
r
a
ti
on
a
tt
a
c
k
o
f
thr
e
e
c
omponents
;
ve
r
tex
de
letion,
ve
r
tex
mod
if
ica
ti
on
a
nd
ve
r
tex
ins
e
r
ti
on
.
At
tac
ks
in
the
f
or
m
of
a
ve
r
tex
point
e
mbedde
d
in
wa
ter
mar
ks
a
nd
ve
c
tor
maps
with
s
pa
ti
a
l
f
e
a
tur
e
s
,
we
r
e
us
e
d
to
tes
t
on
the
da
ta.
I
n
or
de
r
to
c
onduc
t
a
tt
a
c
ks
on
the
tes
t
ve
c
tor
maps
,
QG
I
S
s
of
twa
r
e
is
us
e
d
a
nd
the
r
e
s
ult
s
o
f
the
e
xtr
a
c
ti
on
to
de
ter
mi
ne
the
im
pa
c
t
of
the
a
tt
a
c
ks
T
a
ble
4
,
a
r
e
f
ound
out
by
doing
NC
c
a
lcula
ti
ons
.
T
a
ble
4
c
a
n
be
dis
ti
nguis
he
d
int
o
thr
e
e
pa
r
ts
:
1)
R
otation
a
tt
a
c
k:
r
otating
the
t
e
s
t
da
ta
c
oor
dinate
s
by
30°
to
180°,
2)
s
c
a
li
ng
a
tt
a
c
k:
e
n
lar
ging
the
tes
t
map
f
r
om
0
.
25
to
4
.
0,
a
nd
3)
t
r
a
ns
lation
a
tt
a
c
k:
Alter
the
pos
it
ioni
ng
of
the
ve
r
t
ice
s
by
mov
ing
the
c
oor
dinate
s
.
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
R
ST
invar
iant
w
ater
mar
k
ing
tec
hnique
for
v
e
c
tor
map
…
(
Saleh
A
L
-
ar
dhi
)
507
T
a
ble
4.
E
xpe
r
i
ment
r
e
s
ult
s
f
o
r
r
obus
tnes
s
3.
4.
1.
Rot
a
t
ion
a
t
t
ac
k
T
ur
ning
a
ve
c
tor
map
,
in
o
r
de
r
to
r
e
duc
e
the
W
DR
(
W
a
ter
mar
k
De
duc
ti
on
R
a
te)
is
the
de
f
ini
ti
on
o
f
r
otation.
F
igur
e
7
s
hows
that
ther
e
wa
s
a
r
o
tation
f
r
om
ρ=30
°
(
low
po
int
)
to
ρ=18
0°
(
high
point
)
.
O
nly
ve
r
y
s
mall
e
f
f
e
c
ts
on
the
qua
li
ty
of
the
e
mbedde
d
wa
te
r
mar
k
we
r
e
s
hown
by
the
tes
ti
ng.
W
e
now
ha
ve
e
videnc
e
that
the
r
e
c
omm
e
nde
d
W
V
map
is
s
uf
f
icie
nt
e
nough
to
r
e
s
is
t
s
uc
h
a
n
a
tt
a
c
k.
WAT
E
RM
AR
K
DE
T
E
CT
I
ON
RA
T
E
ON
ROT
AT
I
ON
AT
T
AC
K
R
otation
(
ρ=30
°)
R
otation
(
ρ=60
°)
R
otation
(
ρ=90
°)
R
otation
(
ρ=18
0°)
F
igur
e
7.
Qua
li
ty
o
f
wa
ter
ma
r
k
ve
c
tor
map
a
f
ter
r
o
tation
a
tt
a
c
ks
3.
4.
2
.
T
r
an
s
lat
ion
a
t
t
ac
k
S
c
a
li
ng
(
f
or
thi
s
s
tudy)
,
is
de
f
ined
a
s
r
e
s
izing
a
map
a
long
it
s
a
xe
s
,
s
o
that
the
W
D
R
c
a
n
be
r
e
duc
e
d.
T
he
s
c
a
li
ng
r
a
nge
wa
s
ς
=
0.
25
(
low)
to
ς
=
0.
25
(
high)
.
Akin
to
the
r
otation
a
tt
a
c
ks
,
s
c
a
li
ng
a
tt
a
c
ks
a
ls
o
ha
ve
only
a
s
mall
im
pa
c
t
on
the
W
DR
a
nd
F
igu
r
e
8
s
hows
that
the
wa
ter
mar
k
e
mbedde
d,
jus
t
a
bout
ma
na
ging
to
s
ur
vive.
W
e
c
a
n
c
onc
lude
that
unde
r
a
s
c
a
li
ng
a
tt
a
c
k,
thi
s
method
of
L
C
AT
wa
ter
mar
king
is
only
modes
tl
y
r
obus
t.
3.
4.
3.
T
r
an
s
lat
ion
a
t
t
ac
k
R
e
loca
ti
ng
the
whole
map
s
hif
ti
ng
it
int
o
a
s
pe
c
if
ic
dir
e
c
ti
on,
is
the
main
tr
a
it
of
the
tr
a
ns
lation
a
tt
a
c
ks
.
T
he
map
us
e
d
in
thi
s
r
e
s
e
a
r
c
h,
tr
a
ns
late
d
f
r
om
-
1.
2
,
-
2.
3
m
to
-
0.
6
m
,
0
.
7
m
in
F
igur
e
9
.
T
h
e
s
tudi
e
s
only
s
howe
d
a
mi
nor
e
f
f
e
c
t
on
the
qua
li
ty
of
e
mbe
dde
d
wa
ter
mar
k,
a
pa
r
t
f
r
om
the
e
xc
e
pti
o
ns
a
t
-
7.
9,
-
0.
6
m,
0.
7
m
,
a
nd
2
.
6.
T
he
m
ini
mal
im
pa
c
t
c
ould
be
due
to
the
t
r
a
ns
lation
a
tt
a
c
ks
,
s
o
we
c
a
n
c
onc
l
ude
that
the
L
C
AT
wa
ter
mar
k
ing
tec
hnique
is
moder
a
tely
r
e
s
is
tant
to
t
r
a
ns
lation
a
tt
a
c
ks
.
W
a
ter
mar
k
e
xtr
a
c
ti
on
wa
s
not
a
lwa
ys
a
tt
a
ined,
by
the
a
t
tac
k
tec
hniques
(
wit
h
NC
=
1)
.
Out
of
the
e
nti
r
e
e
va
luation
s
c
e
na
r
io,
th
e
f
a
il
ur
e
r
a
te
of
the
wa
ter
mar
k
de
tec
ti
on
wa
s
89
%
.
F
r
o
m
th
is
,
we
know
that
thes
e
e
xtr
a
c
ti
on
methods
c
a
n’
t
be
us
e
d
in
a
tt
a
c
ks
that
c
ha
nge
the
wa
ter
mar
k
bit
(
T
a
ble
4)
.
F
r
iction
a
ga
ins
t
the
ve
r
tex
is
made
by
T
As
,
c
ha
nging
the
W
B
V
(
W
a
ter
mar
k
B
it
Va
lue)
a
nd
s
toppi
ng
e
x
tr
a
c
ti
on.
I
n
the
mea
nti
me,
r
otation
a
tt
a
c
k
r
e
s
ult
s
s
how
that
the
wa
ter
mar
k
e
xtr
a
c
ti
on
wil
l
be
uns
uc
c
e
s
s
f
ul
is
the
bit
va
lue
c
ha
nge
s
by
up
to
1.
I
f
the
L
C
AT
a
l
gor
it
hm
c
a
n’
t
main
tain
it
s
f
ixed
ge
nuine
wa
ter
mar
k
leve
l
dur
ing
a
n
a
tt
a
c
k,
then
the
W
B
V
c
ha
nge
s
a
nd
e
xtr
a
c
ti
on
be
c
omes
una
tt
a
inable
.
A
tt
a
c
k T
ype
D
e
te
c
ti
on R
a
te
(
N
C
)
R
ot
a
ti
on (
ρ=
30°)
0.939542
R
ot
a
ti
on (
ρ=
60°)
0.921861
R
ot
a
ti
on (
ρ=
90°)
0.906505
R
ot
a
ti
on (
ρ=
180°)
0.898085
S
c
a
li
ng (
ς
=
0.25)
0.913221
S
c
a
li
ng (
ς
=
0.5)
0.890227
S
c
a
li
ng (
ς
=
2.0)
0.884843
S
c
a
li
ng (
ς
=
4.0)
0.880051
T
r
a
ns
la
ti
on (
−
1.2 m,
-
2.3 m)
0.901241
T
r
a
ns
la
ti
on (
4.2 m , 5.6 m
)
0.896940
T
r
a
ns
la
ti
on (
2.6 m,
–
7.9 m)
0.765425
T
r
a
ns
la
ti
on (
–
0.6
m,
0.7 m)
0.558168
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
.
1
,
F
e
br
ua
r
y
2020
:
500
-
510
508
WAT
E
RM
AR
K
DE
T
E
CT
I
ON
RA
T
E
ON
S
CA
L
I
NG
AT
T
AC
K
S
c
a
li
ng
(
ς
=
0.
25
)
S
c
a
li
ng
(
ς
=
0.
5
)
S
a
l
S
c
a
li
ng
(
ς
=
2
.
0)
S
c
a
li
ng
(
ς
=
4.
0
)
F
igur
e
8.
Qua
li
ty
o
f
wa
ter
ma
r
k
im
a
ge
a
f
ter
s
c
a
li
ng
a
tt
a
c
ks
WAT
E
RM
AR
K
DE
T
E
CT
I
ON
RA
T
E
ON
T
RA
NSL
AT
I
N
AT
T
AC
K
(−
1.2 m,
-
2.3m)
(
4.2 , 5.6 m)
(
2.6,
–
7.9)
(
-
0.6 m,
-
0.7m)
T
RA
NSL
AT
I
ON
(
−
1.
2
m
,
-
2.
3m
)
(
4.
2,
5.
6
m)
(
2.
6
m
,
–
7.
9
m)
(
–
0.
6
m,
0.
7
m)
F
igur
e
9.
Qua
li
ty
o
f
wa
ter
ma
r
k
ve
c
tor
map
a
f
ter
tr
a
ns
lation
a
tt
a
c
ks
3.
5
.
Co
m
p
ar
at
ive
s
t
u
d
y
o
f
wat
e
r
m
ar
k
i
n
g
t
e
c
h
n
i
q
u
e
s
T
he
r
e
a
r
e
va
r
ious
wa
ter
mar
king
tec
hniques
de
s
c
r
ibes
in
thi
s
s
tudy.
W
he
n
they
a
r
e
tes
ted
on
D
W
T
,
F
F
T
a
nd
DC
T
a
lgor
i
thm
s
;
f
idelit
y
,
ge
ometr
ica
l
a
t
tac
ks
a
nd
be
tt
e
r
invi
s
ibi
li
ty
is
s
hown
by
the
li
ne
a
r
c
e
ll
ular
a
utom
a
ta
[
31]
.
F
oll
owing
thi
s
,
1D
i
mage
f
il
e
s
w
e
r
e
wa
ter
mar
ke
d
by
C
AT
;
de
s
pit
e
the
f
a
c
t
that
2
-
D
C
A
ne
e
de
d
to
be
c
onve
r
ted
be
twe
e
n
two
dim
e
ns
ions
,
s
pe
c
if
ica
ll
y
a
long
with
r
e
pe
ti
ti
on.
T
he
s
a
me
ke
ys
,
ve
r
ti
c
e
s
a
nd
maps
s
hown
in
T
a
ble
5
we
r
e
us
e
d,
to
c
o
mpar
e
them
to
our
pr
ogr
a
m
[
31]
.
T
a
ble
6
s
umm
a
r
ize
s
the
r
e
s
is
tanc
e
of
the
f
our
tec
hniques
a
ga
ins
t
the
R
S
T
a
tt
a
c
ks
(
in
r
e
lation
to
wa
ter
mar
k
i
mm
unit
y
to
e
li
mi
na
ti
on
a
nd
inadve
r
tent
de
gr
a
da
ti
on)
.
L
C
AT
c
omes
out
on
top,
a
s
a
de
tec
ti
on
r
a
te
of
0.
76;
indi
c
a
tes
that
a
c
ompl
e
te
wa
ter
mar
k
r
e
c
ove
r
y
is
s
ti
ll
f
e
a
s
ibl
e
.
T
a
ble
5.
P
a
r
a
mete
r
s
a
na
lys
is
of
both
DC
T
,
DW
T
,
F
F
T
a
nd
L
C
AT
P
a
r
a
me
te
r
s
a
na
ly
s
is
E
va
lu
a
ti
on M
e
tr
ic
s
L
C
A
T
D
W
T
FFT
D
C
T
I
nvi
s
ib
il
it
y
hi
gh
me
di
um
hi
gh
me
di
um
F
id
e
li
ty
hi
gh
L
ow
me
di
um
L
ow
G
e
ome
tr
ic
a
l
A
tt
a
c
ks
89%
77%
83%
74%
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
R
ST
invar
iant
w
ater
mar
k
ing
tec
hnique
for
v
e
c
tor
map
…
(
Saleh
A
L
-
ar
dhi
)
509
T
a
ble
6.
W
a
ter
mar
k
de
tec
ti
on
r
a
te
c
ompar
is
on
f
or
DC
T
,
DW
T
,
F
F
T
,
L
C
AT
unde
r
va
r
ious
a
tt
a
c
ks
A
tt
a
c
k T
ype
D
C
T
D
W
T
FFT
L
C
A
T
R
ot
a
ti
on (
ρ=
30°)
0.87
0.85
0.88
0.94
R
ot
a
ti
on (
ρ=
60°)
0.84
0.86
0.90
0.92
R
ot
a
ti
on (
ρ=
90°)
0.80
0.80
0.86
0.91
R
ot
a
ti
on (
ρ=
180°)
0.62
0.80
0.79
0.90
S
c
a
li
ng (
ς
=
0.25)
0.47
0.62
0.72
0.91
S
c
a
li
ng (
ς
=
0.5)
0.65
0.70
0.78
0.89
S
c
a
li
ng (
ς
=
2.0)
0.87
0.90
0.93
0.88
S
c
a
li
ng (
ς
=
4.0)
0.78
0.73
0.84
0.88
T
r
a
ns
la
ti
on (
−
1.2 m,
-
2.3 m)
0.77
0.83
0.88
0.90
T
r
a
ns
la
ti
on (
4.2 , 5.6)
0.82
0.83
0.87
0.90
T
r
a
ns
la
ti
on (
2.6,
–
7.9)
0.71
0.73
0.80
0.85
T
r
a
ns
la
ti
on (
-
0.6 m,
-
0.7 m)
0.68
0.60
0.67
0.76
M
a
nipul
a
ti
on
of
r
e
s
ult
s
on
the
wa
ter
mar
k
ins
e
r
t
ion
map,
is
the
c
a
us
e
of
wa
ter
mar
k
e
xt
r
a
c
ti
on
f
a
il
ur
e
.
T
he
dis
tor
ti
on
va
lue
is
a
f
f
e
c
ted
by
the
us
e
of
the
li
mi
tation
va
lue
in
the
wa
ter
mar
k
i
ns
e
r
ti
on.
T
he
wa
ter
mar
k
va
lue
is
f
ound
by
us
ing
a
bit
matr
ix
s
ize
of
M
n
≤
30,
in
or
de
r
to
a
ll
ow
f
or
e
xt
r
a
c
ti
on.
L
im
it
a
ti
on
va
lue
is
a
f
f
e
c
ted
by
modi
f
ica
ti
on
a
mpl
i
tudes
of
M
n
≥
35
bi
ts
.
T
he
tec
hniques
of
s
ome
o
f
the
tes
ts
,
mana
ge
d
to
ke
e
p
the
s
a
me
wa
ter
mar
k,
e
xtr
a
c
ted
a
nd
or
igi
na
l
wa
ter
mar
ks
we
r
e
the
s
a
me
a
nd
NC
=
1
,
in
s
pit
e
of
the
low
r
obus
tnes
s
.
F
or
e
ve
r
y
L
C
AT
va
lue,
modi
f
ica
ti
ons
on
the
va
lue
o
f
the
s
e
que
nc
e
c
ompl
e
x
on
the
ve
c
tor
mapping,
a
n
L
C
AT
c
a
lcula
ti
on
wa
s
s
pr
e
a
d
f
or
e
a
c
h
one
.
T
his
wa
s
done
to
ke
e
p
the
L
C
A
T
v
a
lue
withi
n
the
li
mi
ts
of
the
wa
ter
mar
k
e
xt
r
a
c
ti
on.
Alte
r
a
ti
ons
to
the
c
oor
dinate
va
lue,
d
ir
e
c
tl
y
a
f
f
e
c
ts
the
ins
e
r
ted
wa
ter
mar
k
bit
va
lue;
whe
n
ins
e
r
ti
on
ha
ppe
ns
on
th
e
s
pa
c
e
d
-
out
domain.
T
his
will
p
r
oduc
e
c
ontr
a
s
ti
n
g
r
e
s
ult
s
.
T
he
tr
a
ns
f
or
m
domain
c
a
lcula
ti
on
s
pa
nning
the
L
C
AT
is
be
tt
e
r
a
t
c
ons
e
r
ving
the
wa
ter
mar
k,
in
c
o
mpar
is
on
to
the
s
pa
ti
a
l
tec
hnique.
T
he
r
obus
tnes
s
of
the
method
de
pe
nds
on,
F
DA
(
f
r
e
que
nc
y
domain
a
l
gor
it
hm)
,
e
xtr
a
c
ti
on
li
mi
t
,
r
e
late
d
pr
ogr
a
mm
ing
methods
,
a
s
ymm
e
tr
ic
a
lgor
it
hm
ke
y
qua
li
ty
a
nd
the
da
ta
s
tor
a
g
e
length.
4.
CONC
L
USI
ON
F
indi
ng
a
good
f
r
e
que
nc
y
wa
ter
mar
king
s
c
he
me
f
or
2D
ve
c
tor
map
d
r
a
wings
wa
s
the
pr
incipa
l
a
im
of
thi
s
pa
pe
r
.
T
he
f
r
e
que
nt
p
r
oblems
with
wa
ter
ma
r
king
s
c
he
mes
of
ve
c
tor
maps
a
r
e
invi
s
ibi
li
ty
,
high
f
idelit
y
a
nd
we
a
k
r
obus
tnes
s
.
F
ur
ther
mor
e
,
the
s
c
he
me
ne
e
ds
the
or
igi
na
l
c
ove
r
whe
n
e
xtr
a
c
ti
ng
the
wa
ter
mar
k,
be
c
a
us
e
it
is
‘
Non
-
bli
nd’
.
A
wa
ter
mar
king
tec
hni
que
(
that
wa
s
L
C
AT
domain
tr
a
ns
f
or
med)
wa
s
e
mbedde
d
int
o
the
ve
c
tor
map.
I
nvis
ibi
li
ty
s
howe
d
that
li
ke
ne
s
s
in
f
ideli
ty
s
tage
s
,
in
the
wa
ter
mar
ke
d
map
a
r
e
uphe
ld.
T
he
R
M
S
E
(
dis
tor
ti
on
s
c
a
le)
s
taying
c
ir
c
a
z
e
r
o
a
n
d
the
dis
tanc
e
f
r
om
the
or
igi
na
l
ve
c
tor
map
be
ing
10%
o
r
les
s
,
a
ls
o
dis
playe
d
thi
s
.
T
he
im
pa
c
t
of
the
f
r
e
que
nc
y
domain
s
c
a
tt
e
r
s
th
r
oughout
va
r
ious
f
r
e
q
ue
nc
ies
,
r
e
s
tr
a
ini
ng
im
pa
c
t
a
nd
upping
thi
s
method’
s
de
pe
nda
bil
it
y.
L
C
AT
F
DA
wa
r
r
a
nts
the
ve
c
tor
map’
s
e
xa
c
tnes
s
a
nd
r
obus
tnes
s
is
s
hown
in
89%
o
f
e
xa
mi
ne
d
c
a
s
e
s
.
RE
F
E
RE
NC
E
S
[1
]
K
.
T
.
Ch
an
g
,
“
In
t
r
o
d
u
ct
i
o
n
t
o
G
eo
g
rap
h
i
c
In
f
o
rma
t
i
o
n
S
y
s
t
ems
,
”
McG
raw
-
H
i
l
l
,
2
0
1
2
.
[2
]
V
.
T
ao
,
X
.
D
eh
e
,
L
.
Ch
e
n
g
m
i
n
g
,
an
d
S.
J
i
a
n
g
u
o
,
“
W
at
e
rmark
i
n
g
G
IS
D
a
t
a
fo
r
D
i
g
i
t
al
Ma
p
Co
p
y
r
i
g
h
t
Pro
t
ect
i
o
n
,
”
Pro
ceed
i
n
g
s
o
f
t
h
e
24
th
In
t
e
r
n
a
t
i
o
n
a
l
Ca
r
t
o
g
r
a
p
h
i
c
Co
n
f
er
en
ce
s
(ICC)
,
p
p
.
1
-
9
,
2
0
0
9
.
[3
]
A
.
A
b
u
b
a
h
i
a
,
a
n
d
M.
C
o
cea,
A
d
v
a
n
cemen
t
s
i
n
G
IS
Map
Co
p
y
r
i
g
h
t
Pr
o
t
ec
t
i
o
n
Sch
eme
s
-
a
Cri
t
i
ca
l
Re
v
i
e
w
,
”
M
u
l
t
i
m
e
d
i
a
To
o
l
s
a
n
d
A
p
p
l
i
ca
t
i
o
n
s
,
v
o
l
.
7
6
,
n
o
.
1
0
,
p
p
.
1
2
2
0
5
-
1
2
2
3
1
,
2
0
1
7
.
[4
]
Y
.
L
i
n
g
,
C.
F.
L
i
n
,
an
d
Z
.
Y
.
Z
h
an
g
,
“
A
Z
ero
-
W
a
t
erma
rk
i
n
g
A
l
g
o
ri
t
h
m
f
o
r
D
i
g
i
t
a
l
Map
Bas
e
d
o
n
D
w
t
D
o
ma
i
n
,
”
Co
m
p
u
t
er
,
i
n
f
o
r
m
a
t
i
cs
,
cy
b
er
n
et
i
cs
a
n
d
a
p
p
l
i
ca
t
i
o
n
s
,
v
o
l
.
1
0
7
,
p
p
.
5
1
3
-
5
2
1
,
2
0
1
2
.
[5
]
J
.
W
u
,
Q
.
L
i
u
,
J
.
W
an
g
,
L
.
an
d
G
ao
L
,
“
A
Ro
b
u
s
t
W
at
ermark
i
n
g
A
l
g
o
r
i
t
h
m
fo
r
2
D
C
A
D
en
g
i
n
eer
i
n
g
g
rap
h
i
c
s
Bas
ed
o
n
D
c
t
an
d
C
h
ao
s
Sy
s
t
em,
”
A
d
v
a
n
ce
s
i
n
s
wa
r
m
i
n
t
e
l
l
i
g
e
n
c
e,
v
o
l
7
9
2
9
,
p
p
.
2
1
5
-
2
2
3
,
2
0
1
3
.
[6
]
S.
N
.
N
ey
man
,
I.
N
.
P.
P
rad
n
y
a
n
a,
an
d
B.
Si
t
o
h
a
n
g
,
“
A
N
ew
Co
p
y
r
i
g
h
t
Pro
t
ect
i
o
n
fo
r
V
ec
t
o
r
Ma
p
U
s
i
n
g
fft
Ba
s
ed
W
at
ermar
k
i
n
g
,
”
TE
LKO
M
NIKA
Tel
ec
o
m
m
u
n
i
c
a
t
i
o
n
Co
m
p
u
t
i
n
g
E
l
ect
r
o
n
i
c
s
a
n
d
Co
n
t
r
o
l
,
v
o
l
.
1
2
,
n
o
.
2
,
p
p
.
3
6
7
-
3
3
7
,
2
0
1
4
.
[7
]
S.
A
L
-
ard
h
i
,
V
.
T
h
ay
a
n
an
t
h
a
n
,
an
d
A
.
Ba
s
u
h
ai
l
.
“
Co
p
y
r
i
g
h
t
Pr
o
t
ec
t
i
o
n
a
n
d
Co
n
t
e
n
t
A
u
t
h
e
n
t
i
cat
i
o
n
Bas
e
d
o
n
L
i
n
ear
Cel
l
u
l
ar
A
u
t
o
mat
a
W
at
ermar
k
i
n
g
f
o
r
2
D
V
e
ct
o
r
Ma
p
s
,
”
A
d
va
n
c
es
i
n
C
o
m
p
u
t
er
V
i
s
i
o
n
C
V
C
2
0
1
9
,
pp
.
7
0
0
-
7
1
9
,
20
19
.
[8
]
S.
A
L
-
ard
h
i
,
V
.
T
h
a
y
an
a
n
t
h
an
,
a
n
d
A
.
Bas
u
h
ai
l
,
“
Frag
i
l
e
W
a
t
ermark
i
n
g
b
a
s
ed
o
n
L
i
n
ear
Cel
l
u
l
ar
A
u
t
o
mat
a
u
s
i
n
g
Man
h
a
t
t
a
n
D
i
s
t
an
ce
s
fo
r
2
D
V
ec
t
o
r
Map
,
”
In
t
e
r
n
a
t
i
o
n
a
l
Jo
u
r
n
a
l
o
f
A
d
va
n
ced
Co
m
p
u
t
e
r
S
ci
en
ce
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
(IJA
CS
A
),
v
o
l
.
1
0
,
n
o
.
6
,
2
0
1
9
.
[9
]
S.
A
L
-
ard
h
i
,
V
.
T
h
ay
an
a
n
t
h
an
,
an
d
A
.
Bas
u
h
ai
l
,
“
A
W
at
ermark
i
n
g
Sy
s
t
em
A
rch
i
t
ec
t
u
re
u
s
i
n
g
t
h
e
Cel
l
u
l
ar
A
u
t
o
ma
t
a
T
ran
s
f
o
rm
fo
r
2
D
V
ec
t
o
r
Map
,
”
In
t
e
r
n
a
t
i
o
n
a
l
Jo
u
r
n
a
l
o
f
A
d
va
n
ced
Co
m
p
u
t
e
r
S
ci
en
ce
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
(IJA
CS
A
),
v
o
l
.
1
0
,
n
o
.
6
,
2
0
1
9
.
Evaluation Warning : The document was created with Spire.PDF for Python.