TELKOM
NIKA
, Vol.12, No
.2, June 20
14
, pp. 367~3
7
8
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v12i2.1975
367
Re
cei
v
ed O
c
t
ober 1
1
, 201
3; Revi
se
d Ap
ril 4, 2014;
Acce
pted April 20, 2014
A New Copyright Prote
c
tion for Vector Map using FFT-
based Watermarking
Shelv
i
e Nid
y
a Ne
y
m
an*
1
,
I N
y
o
m
an Prama Pradn
y
ana
2
, Benh
ar
d Sitohang
3
Schoo
l of Elect
r
ical En
gin
eeri
ng an
d Informa
tics (ST
E
I) - IT
B
Ged. Achmad
Bakrie Lt. 2Jl. Ganesh
a
No 1
0
, Bandu
ng, In
don
esia,
T
e
lp. +
62-22-2
502
26
0, F
a
x. +
62-2
2
-25
3
4
2
2
2
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: shelvi
e@i
pb.
ac.id
1
, prama.p
r
adn
ya
na
@itb.
a
c.id
2
, benh
ard
@
stei.itb.ac.id
3
A
b
st
r
a
ct
T
h
is study pr
opos
ed
a ne
w
approac
h o
f
copyrig
h
t pr
otection for v
e
ctor map
usi
ng ro
bus
t
wa
te
rm
a
r
kin
g
o
n
FFT a
l
g
o
r
ithm
. A co
p
y
righ
t
m
a
rke
r
in
se
rted
in
ve
cto
r
m
a
p
a
s
th
e wa
te
rm
a
r
k. I
n
ad
ditio
n
to
data ori
g
in
a
u
thentic
atio
n
ca
pab
iliti
e
s w
a
ter
m
ark,
RSA cry
p
togra
phi
c
a
l
g
o
rith
m
is us
ed
w
hen
ge
ner
ati
n
g
the w
a
ter
m
ark
.
Quality
mea
s
ure
m
e
n
t of t
he r
e
sults
w
a
s bas
ed
on
th
e thre
e c
hara
c
teristics of
di
git
a
l
w
a
termarki
ng:
(1) invis
i
bi
lity usin
g RMSE c
a
lcul
atio
ns
, (2)
fidelity w
i
th the farthest dist
ance
and (
3
) NC
calcul
atio
n
an
d g
e
m
otrica
l
l
e
vel
of r
obust
ness
ag
ains
t
attacks. Resu
lt of ex
per
iment
s show
ed
that
th
e
appr
oach
use
d
in this study s
u
ccee
d
e
d
in i
n
serting co
py
ri
g
h
t as w
a
termar
k on vector
ma
ps. Invisibi
lity test
show
ed
goo
d r
e
sults, de
mon
s
trated by RM
SE close
to
zero. F
i
de
lity of
the w
a
termar
k
ed
ma
p w
a
s
also
m
a
intained. Level of watermark robust
ness
against
geometric attacks on vect
or m
ap results
has
been
ma
inta
ine
d
w
i
thin the l
i
m
its that
these attack
s do not affect t
he w
a
termark
bit valu
e directl
y
.
Ke
y
w
ords
: copyright protection, vector map, robust
waterm
ark
i
ng, fast Fourier transfor
m
(FFT)
1. Introduc
tion
Over the
pa
st few d
e
cade
s, geospatial d
a
ta
produ
ctio
n process h
a
s
evolved
fro
m
pap
er
maps to
digit
a
l data
format
be
cau
s
e
of t
he infl
u
e
n
c
e of
the develo
p
ment
of com
puter technol
ogy
for g
eog
rap
h
i
c d
a
ta
colle
ction
devices su
ch
a
s
ge
ogra
phi
c
p
o
sitioning syste
m
s (GPS)
a
n
d
satellites that
provide accurate
spatial coordi
nate dat
a.Vector ma
ps as the fundamental
data of
a
geog
ra
phi
c
information system
(GIS) has repl
aced t
he role
of an
alog data
or
print [1] . This is
unde
rsta
nda
b
l
e becau
se
geo
spatial v
e
ctor
data
h
a
s the a
d
va
ntage of hig
h
pre
c
i
s
ion
data,
automated
p
r
ocesse
s a
n
d
lossle
ss
scalin
g co
mp
ared to the
data in pa
per form. Easie
r
prod
uctio
n
,
stora
ge
and
distri
bution
in digi
tal
map
s
tran
sa
ction
brin
gs
abo
ut o
t
her
con
s
e
que
nce
s
, nam
ely ea
sy mani
pulati
on an
d
a
c
qui
sition. It en
coura
ged
a n
eed a
m
ong
the
map produ
cers fo
r a m
ap produ
ctio
n mechani
sm
that can
facilitate copy right m
a
rker
[2].Furtherm
o
re, map con
s
umersal
s
o re
quire the a
b
il
ity to know the ownershi
p
identity of th
e
spatial
data
t
hey receive.In ad
diti
on,
re
gulatorsh
ave
the ne
ed
to v
e
ri
fy the vali
d
i
ty of the p
ubl
icly
distributed m
ap ownership.Digital
watermarking is
one of the best
solution
s that
can be
utilized
to solve the p
r
oble
m
.
Digital
watermarkingi
s a
tech
niqu
e that
wo
rks
by i
n
serting
ce
rtain
informatio
n (referre
d
as waterm
ark) into a digital
media file, when u
s
ed o
n
a digital map,
the informati
on may co
nta
i
n
data whi
c
h is used to verif
y
the integrity or ow
ne
rsh
i
p of the map; prov
ided t
hat the inse
rtion
pro
c
e
s
s sho
u
ld result in
very sm
all
distortio
n
val
ue of the
re
sulting
map[
3]. Robu
st d
i
gital
watermarkin
g
is on
e of
digi
tal wate
rma
r
king te
chni
que
s that
ha
s
re
sistan
ce
ch
ara
c
teri
stic toward
data
content
s removal
an
d
modification
whe
n
its i
n
se
rtion me
dia
p
o
int ch
ang
es,
either due
to
attacks o
r
d
a
ta processi
ng. Du
rin
g
a
pplicat
ion
sta
ge, this te
ch
nique i
s
com
m
only u
s
ed
for
copyri
ght p
r
o
t
ection. Th
e
appli
c
ation
of rob
u
st
digita
l wate
rma
r
ki
n
g
techniq
u
e
s
on m
ap ve
ct
or
works on two types of d
o
main
s; sp
atial dom
ain
a
nd tran
sform
a
tion domai
n
[3]. The main
transfo
rm alg
o
rithm
s
a
r
e DFT
(
Discrete Fouri
e
r Tra
n
sform
), DW
T (
Disc
rete Wavelet
Trans
form
),
and DCT (
Di
sc
ret
e
C
o
si
n
e
Tran
sf
orm
) [4]. These digital waterm
arki
ng
s in the transfo
rmat
ion
domain
are
kno
w
n to b
e
robu
st to attack. Fo
rd
i
g
ital media
c
o
p
y
right prote
c
t
i
on appli
c
atio
n,
robu
st digital
watermarkin
g
techniqu
e
mainly wo
rk
on the tran
sf
ormatio
n
do
main; be
cau
s
e the
transfo
rmatio
n domain
ha
s so
me adv
antage
s in
term
s of invisibility and strong robu
stne
ss,
comp
ared wit
h
the spatial
domain
whi
c
h
have fr
agile
nature, an
d impleme
n
tatio
n
ease [5].
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 2, June 20
14: 367 – 37
8
368
At pre
s
ent, re
sea
r
che
r
s mo
stly co
ncentrate on
the
ro
bust
digital
waterma
r
ki
ng
a
l
gorithm
in tra
n
sfo
r
ma
tion do
main
for im
age[6]
-[11], and
au
di
o[12]-[16].
Ch
ara
c
teri
stics f
ile vecto
r
m
a
p is
very different
from the
im
age o
r
a
udio
as
wate
rma
r
k em
bed
ding
media.Fo
r it
take
s a
diffe
rent
techni
que
for inserting
a i
n
formatio
n in
the ve
ctor
m
ap. Some
exi
s
ting
re
sea
r
ch ha
s
pro
p
o
s
ed
the use of ro
bust di
gital waterma
r
ki
ng
onthe tra
n
sfo
r
mation
dom
ain of vecto
r
map u
s
ing
d
a
ta
transfo
rmatio
n algo
rithm
s
based
upon
DFT [2], I
W
T
(Intege
r
Wavelet T
r
a
n
sform) [1
7], and
DCT[1
8]-[20].
Only two ex
isting
studi
e
s
that were
specifi
c
ally ca
rrie
d
o
u
t for
the pu
rpo
s
e
s
of
vector m
ap
copyri
ght p
r
o
t
ection [2],[17]. Other
re
sea
r
ch
studi
es that
discuss the topi
c of
copyri
ght digi
tal maps
with different te
chni
que
s a
r
e
blind watermarking
with
DCT [21], zero
watermarkin
g
in the
spat
ial dom
ain [
22], re
cu
rsiv
e waterm
arki
ng in
spatial
domai
n [23
],
watermarkin
g
in spatial topl
ogy domain [
24], and reve
rsibl
e
wate
rm
arki
ng in spat
ial domain [2
5].
To e
nhan
ce
copyri
ght p
r
o
t
ection
perfo
rmance te
chni
que
s o
n
ve
ctor m
a
p
s
, this stud
y
prop
oses the
use of
rob
u
st digital watermarking
dom
ain tra
n
sfo
r
m
a
tion a
s
p
r
o
o
f of co
pyright
on
vector ma
p usin
g FFT d
a
ta transfo
rmation algo
ri
thm.FFT alg
o
rithm is very popular in
the
watermarkin
g
com
m
unity [
26] an
d neve
r
u
s
ed
to ve
ctor ma
p a
s
e
m
bedd
ed
me
dia. FFT
often
use
d
du
e to
its red
u
ced
comp
utation
a
l burden
while maintai
n
ing the q
uali
t
y of insertio
n
comp
ared wit
h
some oth
e
r
data tran
sformation algo
rithms [27]. In addition to ownership ma
rki
ng
purp
o
se, the
approa
ch ta
ken in thi
s
stu
d
y also
provides data orig
in
authe
nticat
ion
capa
bilities
via the RSA
publi
c
key
cryptog
r
a
phy
algo
rithm.
Wi
th the u
s
e
o
f
the algo
rith
m, data o
r
igi
n
authenti
c
atio
n can
obtain
e
d
se
cu
rity through o
w
n
e
rs
hip private
ke
y used to e
n
crypt copy
right
on
vector ma
ps.
U
ser can ea
si
ly to authenticate data
o
r
ig
in of the map by using
RSA public
key.
The
perfo
rma
n
ce
techiqu
e
is m
e
a
s
ured t
h
rou
gh
simil
a
rity test u
s
in
g
NC
cal
c
ulati
on. Th
e
test is u
s
e
d
to determine t
he succe
s
s o
f
this tech
niq
ue in the i
n
se
rtion of
copyri
ght into a ve
ctor
map. The
re
sults
of this
study mea
s
u
r
ed o
b
je
ctively, not base
d
on pe
rcepti
on a
s
on
so
me
existing re
se
arch[17],[18],[28].Quality measure
m
ent
of the results of this study
was b
a
sed
on
three
of the
chara
c
te
risti
c
s of digital
wat
e
rma
r
ki
ng. T
hey areinvisi
b
ility
using roo
t
mean sq
uared
error
(RMSE)cal
c
ulatio
ns,
fidelity with t
he farth
e
st
di
stan
ce
cal
c
ul
ation, an
d no
rmali
z
ed
cro
s
s
correl
ation (NC), an
d the
strength of
wa
termar
k rob
u
s
tne
ss agai
n
s
t
gemot
rical attacks su
ch as
transl
a
tion, ro
tation and scaling [29].
The result finding
s sho
w
e
d
that the ap
proa
ch
used i
n
this
study succee
ded i
n
i
n
se
rting
copyri
ght as watermark
on
vector maps.
Invisi
bility
on the experi
mental
result
findings showed
good
results,
demon
strated
by
RMSE va
lues ge
nerat
e
d
from
the
m
ap te
st d
a
ta v
a
lue
s
: bel
ow
1
or
clo
s
e
to
zero. Fi
delity
of the m
ap i
s
also m
a
intai
ned, in
dicate
d by the
di
stance
shift an
d NC
values in th
e
rang
e
deem
e
d
a
c
ceptabl
e
according
to stand
ard
s
. Le
vel
of waterm
ark
robu
stne
ss
again
s
t geom
etric atta
cks
on vecto
r
ma
p re
sults h
a
s
been mai
n
tai
ned withi
n
the limits that these
attacks do not affect the watermark bit
value direct
ly or still within the specified value extract
i
on
limits.
In the n
e
xt part
of this p
aper
we
will
explain th
e t
e
ch
niqu
es u
s
ed o
n
th
e a
p
p
roa
c
h
develop
ed in
this stu
d
y, followe
d by an
elabo
rati
on
of the experim
ental re
sult
s
and an
alyse
s
of
the perfo
rma
n
ce of the ap
proa
ch. Th
e final se
ction of
this pape
r cl
ose
s
with
con
c
lu
sion
s.
2. Res
earc
h
Method
2.1. Fast
Fo
urier
Trans
f
orm
Fast
F
o
u
r
ier Tran
sfo
r
m (F
FT)
i
s
an alg
o
rithm
used t
o
re
present a
sig
nal i
n
a
d
i
screte
time and
a freque
ncy d
o
m
a
in. FFT i
n
g
eneral i
s
u
s
e
d
to
calculate
the di
screte
tran
sform
a
tion
of
the DF
T q
u
ickly an
d efficie
n
tly. FFT is
u
s
ed
to d
e
crea
se th
e
compl
e
xities of th
e
DFT. G
ene
ral
l
y,
the FFT form
ulation can b
e
descri
bed a
s
it
is in the equation (1) o
r
(3) [30].
H(
k)
=
∑
(1)
W
N
=
/
=
cos
(
2
π
/N
) - j
si
n
(
2
π
/N
) (
2
)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
A New Copy
right Protec
tion for Vec
t
or M
ap Us
ing FFT-based ....
(S
helv
i
e Nidy
a Ney
m
an)
369
H(
k)
=
∑
2
/
/
+
∑
2
1
/
(3)
is the domain transfo
rmat
ion value,
is the digital media block value,
is the a
m
ount
of the
data th
at will
be
alte
red
to b
e
a
freque
ncy
dom
ain. While fo
r the inve
rse f
o
rmul
ation
of
the
FFT is u
s
ing t
he equ
ation (4).
h(
n)
=
∑
∗
∗
(4)
is the real value of the complex figure
,
is the imaginary value of the complex
figure, and
is the wate
rmarked
comp
lex sequ
en
ce
value.
2.2. Trans
f
orm
T
r
ansform
In this rese
arch, the
watermark in
sertio
n as
a copyri
ghts ma
rker
on the ve
ctor map
i
s
con
d
u
c
ted
on
the tran
sformation d
o
mai
n
for the
co
ordinate
of verti
c
e
s
.The
wate
rmark in
sertio
n
pro
c
e
s
s is
co
ndu
cted o
n
th
e co
efficient
of the tr
an
sfo
r
mation
re
sult
frequ
en
cy of the vecto
r
m
a
p
data. To tran
sform ve
ctor map into d
o
main
fre
que
ncy si
gnal, the vecto
r
m
ap coordinat
e i
s
modified into
a compl
e
x se
quen
ce
with the formulation (5) [2].
(5)
is the a
b
sci
s
sa of the v
e
ctor
map
coordi
nate a
n
d
is the
ordinate of the
vector m
ap
c
o
ordinate. While the
used is the coo
r
dinate ind
e
x occu
rre
d in the map file
mentione
d. The
techni
que u
s
ed for the wat
e
rma
r
k in
se
rti
on is this fo
rmula (6
).
F’
=
F
+
α
W
(6)
F’
is the wa
termarke
d freque
ncy coe
fficient,
F
is the initial freque
ncy coe
fficient,
α
is the
modificatio
n
amplitude, an
d
W
is the waterma
r
k bit.
On the (6
) formulatio
n abo
ve, the bigge
r
α
use
d
, the big
ger chan
ge
s will hap
pen o
n
the vector
map file, but the watermark re
sista
n
ce is
s
t
r
o
ng
er
. T
h
is r
e
s
e
ar
ch
us
es
α
value
as
big a
s
2,
with
an a
c
ceptabl
e vecto
r
ma
p
cha
nge
s, an
d
a
highresi
s
tan
c
e value[31].
2.3.
The Embedd
ing Waterma
r
k Phase
The em
bed
d
ed mo
del of
digital watermark of t
he v
e
ctor map
da
ta is
sho
w
n i
n
Figu
re 1.
In the watermark in
se
rtio
n process, th
ree i
nput
s a
r
e u
s
ed. T
hey
are
vecto
r
map, a
co
pyright
inse
rted i
n
a
s
the
wate
rma
r
k,
and
the
p
r
ivate
ke
y of
the RSA
cryp
tograp
h al
go
rithm. The
first
stage of the inse
rtion pro
c
e
ss i
s
by lookin
g for the
coordinate o
n
the vector
map that will
be
saved into a
list which
can be tran
sf
orme
d into a domain freq
uen
cy. After
getting a set
of
coo
r
din
a
te from the vertex point
of each feature, t
hat coordinate
therefore wil
l
be transfo
rmed
into a com
p
l
e
x sequ
en
ce
. The next stage is t
he readin
g
or th
e byte watermark from t
he
copyri
ght ma
rke
r
file and encrypted wit
h
an RSA pr
i
v
ate key algo
rithm of the copyright hol
d
e
r.
The pu
rpo
s
e
of this encryp
t
ion is to pro
v
ide secu
rity for the data origin a
u
then
ticity, which is a
guarantee
th
at the
data
source
is
from
a l
egitimate
party. The
waterma
r
k e
n
cryption
re
sult
then
transfo
rme
d
i
n
to a bit set and saved int
o
a list. After both input
s a
r
e re
ady, the next stage is
to
do the
compl
e
x se
que
nce
saving
an
d le
ngthen
the
b
y
te wate
rmark in
a
differe
nt file. Late
r
,
the
compl
e
x seq
uen
ce will be
used to extract the wate
rmark with the
non-bli
nd wa
termarkin
g
. The
next step is tran
sformi
ng the FFT towa
rds vecto
r
ma
p file to beco
m
e a freque
n
c
y domain a
n
d
the watermark insertion o
n
the real re
su
lt figur
es of the tran
sform
a
tion domai
n mentione
d. And
after the e
n
tire
waterm
ark has
bee
n in
serte
d
, t
hen
the next step
is to resto
r
e
the freq
uen
cy
domain ve
cto
r
map to retu
rn to its origin
al sha
pe file with iFFT.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 2, June 20
14: 367 – 37
8
370
2.4.
The Extr
acti
ng Wa
termar
k Phase
The wate
rma
r
k extra
c
tion
point is ba
sically t
he same
with the inse
rtion pro
c
e
ss with a
reverse
step
s. In the extraction
pro
c
e
s
s, we
us
e th
e three i
nput
s that are th
e re
sults
of the
inse
rtion p
r
o
c
ess such a
s
t
he complex
seque
nce of
th
e re
al vecto
r
map file, waterma
r
ked ve
ctor
map,
an
d RS
A
algo
rithm publi
c
key.
You can
s
ee
t
he stage
s of
the wate
rma
r
k
extra
c
tion on
Figure 2.
Figure 1. Wat
e
rma
r
k e
m
be
dded
Watermarked vect
or
m
ap
(
9
) Generation
Coordinates of Modified
V
(
8
) Transformation into
Complex Sequence
(
a
k
)
(
7
) iF
FT Transformation
W
a
term
ark
RS
A En
cryp
tio
n
o
f
th
e
Wa
t
e
r
m
a
r
k
(
6
) Embedded of
Wa
t
e
r
m
a
r
k
(
4
) Calculation of
Complex Number FFT
(
3
) F
F
T Transformation
(
1
) E
x
traction coordinates
of vertices
(
2
) Transformation into
Complex Sequence
(
a
k
)
Original vector
m
ap
(
5
) Calculation of
Magnitu
de &
P
has
e
Figure 2. Wat
e
rma
r
k extracted pha
se
W
a
term
ark
(
8
) R
S
A Descryption of the
Wa
t
e
r
m
a
r
k
(
7
) Generation of list bytes
of E
n
cryption W
a
termark
W
a
term
aked
vector map
(
1
) E
x
traction coordinates
of vertices
(
2
) Transformation into
Complex Sequence
(
a
k
)
(
6
) E
x
traction of
W
a
termark
Original vector
m
ap
(
3
) F
F
T Transformation
(
4
) Calculation of Complex
Nu
mb
er FFT
(
5
) Calculation of
Magnitu
de &
P
has
e
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
9
30
A New Copy
right Protec
tion for Vec
t
or Map
Us
ing FFT-based ....
(She
lv
ie Nidya Ney
m
an)
371
The first stag
e of the extra
c
tion p
r
o
c
e
s
s is to loo
k
for the co
ordi
nat
e from the
sh
ape file
vector
map t
hat consi
s
ts
of wate
rma
r
k and
save
it
in a li
st to b
e
tran
sfo
r
m i
n
to a freque
ncy
domain. The
next stage is to calculate
the compl
e
x seq
uen
ce an
d FFT to obtain the comp
lex
figure from th
e sh
apefile fil
e
. The list i
s
con
s
i
s
t of the
encrypted
waterma
r
k
bit value gotte
n fro
m
the deviation
of the
com
p
lex figure from the
re
al
map a
n
d the
wate
rma
r
ke
d map. T
h
en
, by
usin
g the
the
RSA p
ubli
c
key, de
crypt
the wate
rm
ark fo
r the i
n
itial waterm
ark as a
co
pyrig
h
t
marker. T
he
decryption p
r
oce
s
s can b
e
don
e w
hen
the en
crypt
ed waterm
ark u
s
e
s
the
ri
ght
publi
c
and p
r
i
v
ate key set, so we ca
n assure that
the copyri
ght is from a legitima
te party.
3. Resul
t
s
and
Discus
s
ion
3.1. Experimenta
l
Results
Vector map
u
s
ed
a
s
the
ev
aluation
data
is tw
o
sha
pef
ile (.shp
)
file t
y
pe ESRI
sta
ndard
that wa
s buil
t
from point f
eature
s
with
4008 ve
rtex
and lin
e feat
ure
s
with 75
18 vertex. A
s
a
copyri
ght ma
rke
r
, we u
s
e
d
three bitma
p
type pi
cture files with these mea
s
u
r
ements fo
r e
a
ch:
178
byte (2
9
x 29 pixel
s
),
154
byte (2
2
x 23 pixel
)
an
d 17
4 byte
(2
8 x 28
pixels). The le
ngth
of
the copyrig
h
ts m
a
rke
r
bit i
s
limited
to t
w
ice
sm
alle
r
than the
vert
ex amou
nt o
n
the ve
cto
r
map
file.
The pe
rform
a
nce te
chni
qu
e analysi
s
de
veloped in thi
s
re
sea
r
ch was mea
s
u
r
e
d
through
NC
cal
c
ulati
on. The NC calculation
wa
s don
e to
analyze the
similarity be
tween the in
itial
watermark be
fore the in
sertion and the e
x
tracted
wa
te
rmark result with the valu
e ran
g
ing fro
m
0
to 1. The hig
her the
NC v
a
lue, the more simila
r
both
image
s, therefore it
ca
n b
e
stated that
the
watermarkin
g
usage te
chni
que su
cce
ss
is highe
r.
Th
e NC calculat
ion re
sult use
d
the equatio
n
(7) with
is the initial watermark and
’
is the extracted
waterm
ark result that ca
n
be seen in
Table 1 [2].
|
⋅
|
√
⋅
√
⋅
(7)
Table 1. Re
sult of similarit
y
test betwee
n
origin
al wat
e
rma
r
k
with e
x
tracted
wate
rmark
Ma
p
Size of Original
Watermark (b
yte
)
O
r
iginal
W
aterm
a
rk
Extracted
W
aterm
a
rk
Size of Extracted
Watermark
(b
y
t
e
)
NC
Linestring
178 178
1
154 154
1
174 174
1
Point
178 178
1
154 154
1
174 174
1
Table 1
sho
w
s that the
entire evalu
a
t
ion dat
a results in NC i
s
1 with the same
watermark l
e
ngth an
d
con
t
ent. Similarity value is
eq
ual to 1
between the
initial
wate
rma
r
k
with
the waterm
ark extra
c
tion
result
s
sho
w
that bot
h
w
a
te
r
m
ar
k id
en
tic
a
l.
W
a
ter
m
ar
k c
a
n
be
r
e
-
extracted fro
m
the vector
map file and i
t
will not
go throu
gh si
ze o
r
conte
n
t cha
n
ge
s. Therefore,
we can co
nfirm that this tech
niqu
e su
ccee
ded in
in
sertin
g co
pyri
ghts a
s
a wa
termark with
o
u
t
cha
ngin
g
the watermark qu
ality.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
9
30
TELKOM
NIKA
Vol. 12, No. 2, June 20
14: 367 – 37
8
372
3.2. In
v
i
sibilit
y
Ev
a
l
uation
Invisibility measu
r
em
ent uses two pa
ram
e
ters
a
s
refe
rence analysi
s
is the calcula
t
ion of
RMSE. The
calcul
ation of t
he di
stortio
n
betwe
en th
e
begin
n
ing
of the ma
p file a
nd the
re
sult
of
interpol
ated
watermark
was
co
ndu
cted
in the
RMSE
mea
s
u
r
eme
n
t. The RMSE
formulatio
n u
s
ed
is ba
sed o
n
Equation (8) [3
2].
RMSE =
∑∑
,
,
,
(8)
N den
otes th
e numb
e
r of
vertex map vector,
,
is the value of co
mplex se
que
nce of e
a
rly
maps at
coo
r
dinate
,
,
,
is the
value of com
p
lex seq
uen
ce of the map
result at coo
r
dinate
,
.
Table 2. Re
sult of invisibility test between origi
n
al m
ap with watermarked ma
p
Table 2 sho
w
s that the RM
SE values ob
tained fo
r all
data analy
s
is
use
d
in this rese
arch
prod
uces a
value
belo
w
1
and
close to
ze
ro.
Th
e res
u
lts o
f
th
is
s
t
u
d
y
s
h
ow an
impr
o
v
e
m
en
t
comp
ared to
previou
s
simil
a
r
study[4].Th
e
refo
re, the
tech
niqu
es u
s
ed to
gen
erat
e a
goo
d
RM
SE
data analy
s
is output value
indicate
s tha
t
the occu
rre
n
ce
of geom
etrical
disto
r
tion scale, d
u
e
to
the copyri
ght marking o
n
the map, is ve
ry low.
The lo
w disto
r
tion shows that the
presen
ce of a
watermark on
media interp
olation is diffi
cult to be det
ected by the
human
sen
s
e
s
.
Figure 3 sho
w
re
sult of th
e overlay the
origin
al map
and corre
s
p
o
ndin
g wate
rmarked
map u
s
ing
re
d dottes
and
purpl
e dotte
s. Gree
n box m
a
rks a
shift in
the co
ordi
nat
e
s of the ve
rtex
of the two ma
ps. Th
e figu
re can b
e
vie
w
ed
that
the distorsio
n
ca
use
d
b
y
w
a
ter
m
a
r
k
e
m
be
dd
in
g
pro
c
e
s
s is
small eno
ugh
and the
wate
rmarke
d ma
p
pre
s
e
r
veves the geo
sp
atial inform
ation
in
the origin
al m
ap with hig
h
pre
c
isi
on.
3.3. Fidelit
y
Ev
a
l
uation
The fidelity a
s
pe
ct of
digit
a
l waterm
arki
ng
co
nc
ep
t is d
e
f
in
ed
as
th
e
w
a
te
rma
r
k
c
a
nn
o
t
be dete
c
ted
by huma
n
se
nse
s
a
nd
do
es
not si
gnifi
cantly de
grad
e the q
ua
lity
of the me
dia
file
interpol
ation
[31]. Besides RMSE, farthest chan
ges occurred
will also
be measured. Farthest
distan
ce i
s
a
position
shift that occu
rs due to
the waterma
r
k inte
rpolatio
n into
the vector m
ap
files. The farthest dista
n
ce is obtain
e
d
by co
mpa
r
in
g the entire
coo
r
din
a
te’s
vertex betwe
e
n
origin
al vecto
r
map file an
d vector ma
p file c
ontain
i
ng wate
rma
r
k. Farthe
st d
i
stan
ce is th
en
conve
r
ted int
o
meter usi
n
g Quantum
softwa
r
e
GIS
.
According to the Geog
raphi
cal Surv
e
y
Institute of Ja
pan, the ch
a
nge
s t
hat ca
n be tolerate
d is equ
a
l to 75 cm on the
actual si
ze [
31].
Based
on
Ta
ble 3, vi
sible
shift in th
e lo
nge
st po
sitio
n
o
c
curred
in
the d
a
ta a
n
a
l
ysis i
s
equ
al to
Ma
p
W
aterm
a
rk
Size of Watermark
(b
y
t
e
)
RMSE
Linestring
178 0.00000005
154 0.00000005
174 0.00000005
Point
178 0.005115
154 0.005115
174 0.005201
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
9
30
A New Copy
right Protec
tion for Vec
t
or Map
Us
ing FFT-based ....
(She
lv
ie Nidya Ney
m
an)
373
0.506 m
e
ters by 51
cm,
or in oth
e
r
wo
rds, the
c
han
g
e
s th
at o
c
curred i
s
not
more than
75
cm.
In
this
case, the
cal
c
ulatio
n result of th
e f
a
rthe
st
di
stan
ce valu
e a
ppl
ied to the
dat
a analy
s
is ca
n
still maintain
a level of vector map precisi
on or preserv
e
accuracy level of the data.
Figure 3. Orig
inal and
wate
rmarke
d map
overlaid with
each other.
Table 3. Re
sult of fidelity
test between
origin
al map
with wate
rma
r
ke
d map
3.4. Robu
stn
ess
Ev
aluation
The te
sting p
r
oce
s
s was th
en pe
rformed
to det
ermine
the level of
waterma
r
k ro
b
u
stne
ss
from the
tech
nique
s
devel
oped
to d
eal
with
attacks that have
b
een
pre
pared
. Thre
e type
s of
geomet
ric attacks,
nam
ely tran
slation,
rotation, an
d
scalin
g atta
cks. Attacks
we
re ap
plied
to t
he
Ma
p
W
aterm
a
rk
Size of Watermark
(b
y
t
e
)
Farthest Distanc
e
(meter)
Linestring
178 0.25
154 0.25
174 0.26
Point
178 0.5
154 0.5
174 0.51
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
9
30
TELKOM
NIKA
Vol. 12, No. 2, June 20
14: 367 – 37
8
374
test data in the form of vector
map
s
that have spatial
feature
s
wi
th
4008 vertex p
o
ints that have
been in
se
rte
d
wate
rma
r
ks sized 17
8 b
y
tes (29 x 29
pixels).Attacks
ca
rri
ed ou
t on test vect
or
maps that ha
ve been inserted watermark used feat
ures provided b
y
the software Quantum G
I
S.
NC
cal
c
ulatio
ns pe
rformed
on the
wate
rmark extra
c
t
i
on re
sults to
determin
e
the extent of the
cha
nge
s that occur d
ue to
these atta
cks. Another
type of testing was al
so cond
ucted to see the
impact of RS
A implement
ation as th
e data ori
g
in
a
u
thentication
toward watermark ro
bu
stn
e
ss
level.
Tran
slatio
n at
tack wasdon
e by
ch
angin
g
the
po
sition
of
some
vert
ice
s
by
movi
ng
som
e
coo
r
din
a
tes o
n
test data. The test re
sult
s ca
n be see
n
in Table 3.
Table 4. Re
sult of translati
on attacks te
st
Amount of
vertexs
Translation
Watermark
Extracted
(w
ith RSA)
NC
(w
ith
RSA)
Watermark
Extracted
(w
ithout RSA)
NC
(w
ithout RSA)
1 0.1
1
1
1 0.2
1
1
1 0.3
1
1
1
0.4 k
Not detected
-
0.9939
2
0.1
1
1
2
0.2
1
1
2
0.3
Not detected
-
0.8935
(-) 2
0.1
1
1
(-) 2
0.2
Not detected
-
0.9885
3
0.1
1
1
3
0.2
Not detected
-
0.727
Rotation
atta
ck was don
e
by rotating
th
e ent
ire
te
st data coo
r
din
a
tes
i
n
the ra
nge of
-
0.01 to 0.009
degree
s. The
test re
sults a
r
e sh
own in T
able 4.
Scaling atta
ck wa
s do
ne by enlargi
ng
the size
of the test map st
arting from 1.
0001 to
1.0009. The t
e
st re
sult
s are sho
w
n in T
able 5.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
9
30
A New Copy
right Protec
tion for Vec
t
or Map
Us
ing FFT-based ....
(She
lv
ie Nidya Ney
m
an)
375
Table 5. Re
sult of rotation attacks test
Rotation (
° )
Watermark
Extracted
(w
ith RSA)
Watermark
Extracted
(w
ithout RSA)
NC
(w
ith RSA)
NC
(w
ithout RSA)
-0.01
Not detected
Not detected
0
0.1035
-0.009
Not
detected
0
0.3300
-0.008
Not
detected
0
0.9035
-0.007
1
1
-0.006
1
1
0.006
1
1
0.007
1
1
0.008
Not
detected
0
1
0.009
Not
detected
0
1
Table 6. Re
sult of scalin
g attacks test
Scaling
Watermark
Extracted
(w
ith RSA)
Watermark Extra
c
ted
(w
ithout RSA)
NC
(w
ith RSA)
NC
(w
ithout RSA)
1.0005
Not detected
0 0.9980
1.0004
Not detected
0 0.9984
1.0003
Not detected
0 0.9997
1.0002
Not detected
0 1
1.0001
1 1
1.0009
1 1
1.0008
1 1
1.0007
1 1
1.0006
1 1
After implementing the three types of
atta
cks, the techniqu
es
use
d
did not
always
manag
e to do waterm
ark extraction
with NC value
e
qual
s to one. The evaluation scena
rio tha
t
failed to dete
c
t the wate
rm
ark
wa
s 4
0
%. Table 4,
5, a
nd 6 sho
w
th
at for the atta
cks that
chan
ge
the watermark bit, this technique
will not be able to be extracted.
One of the
examples is t
he
transl
a
tion att
a
ck that
cau
s
es a f
r
ictio
n
towa
rd
s the v
e
rtex which causes th
e in
serted
wate
rm
ark
bit value ch
a
nge
s and th
e
extraction
cannot be
don
e.
Beside it i
s
se
en o
n
the rotation att
a
ck
results on Table 4, the bit val
ue change as much as
1 will make
t
h
e watermark extraction fai
l
.
Therefore, the FFT algorit
hm will be abl
e to maintain
the real watermark value on a certain li
mit,
but when t
h
e
attack
can
no lo
nge
r b
e
maintain
ed
by the FFT,
the waterm
ark bit val
ue
will
cha
nge a
nd the extractio
n
can n
o
long
er be execute
d
.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 2, June 20
14: 367 – 37
8
376
We
ca
n
say t
hat this watermark extracti
on fa
ilu
re
is caused
by a
q
u
ite big
di
stortion on
the waterm
ark in
se
rtion
m
ap result. On
e of the
fa
ct
ors that affe
cts the
disto
r
ti
on valu
e i
s
t
h
e
limitation value u
s
ag
e in
the wate
rma
r
k in
sertio
n
pro
c
ess
.
T
h
is
re
s
e
a
r
c
h
us
ed
a
-
0
.4
to
0.6
limitation ra
n
ge to dete
r
mi
ne the
wate
rmark value
so
it can
be
extracted
preci
s
ely. The limitation
value is affected by the
usage of th
e m
odificatio
n
a
m
plitude as much
a
s
2 or the
integ
r
atio
n
that
happ
ened in t
he FFT calcul
ation value.
The wate
rma
r
k extra
c
tion
su
ccess is al
so a
ffecte
d
b
y
the usage o
f
the RSA asymmetric
cryptog
r
a
ph
algorith
m
a
s
the data
o
r
igi
n
auth
entic
ation
se
rvice.
T
h
is i
s
evaluat
ed by
the
sa
me
impleme
n
tation techniq
ue
but witho
u
t u
s
ing
RSA al
g
o
rithm. Th
e result
can
be
see
n
on
Tabl
e a,
V, and VI, NC value incre
a
se
d and sh
owe
d
that
the waterm
ark
robu
stne
ss h
a
s be
com
e
better
and the su
ccess level for the waterm
ark extrac
tio
n
is getting highe
r. The RSA asymm
e
tric
cryptog
r
a
ph
algorith
m
usage is p
r
ove
n
to decre
a
s
e the rob
u
st
ness level from the tech
nique
use
d
. This rese
arch u
s
e
d
RSA asymmetric
cr
yp
tograp
h algo
rithms to au
thentify the java
se
curity library. This algorit
hm is a very se
n
s
itive one
toward
s en
crypted data chang
es. Whe
n
an e
n
crypted
data i
s
havi
ng a
sli
ght
chang
e, then
this d
a
ta
will
not b
e
a
b
le
to be
de
crypt
e
d
further. The
r
efore, any ch
ange
s hap
pe
n on the bi
t after the attack will ca
use failing wate
rm
ark
data extraction. This will
gi
ve us choi
ces, whet
her to increase the wate
rm
ark robust
ness or to
increa
se the
data origi
n
au
thenticatio
n.
Eventhough t
he robu
st lev
e
l wa
s q
u
ite l
o
w, but
the
tech
niqu
e wa
s abl
e to mai
n
tain the
watermark fo
r so
me test
s
that can b
e
seen on
ta
ble
4, 5, dan 6. Some extra
c
tion p
r
o
c
e
s
ses
sho
w
that the re
sult of the ex
tracte
d watermark is exactly the
same as
the origin
al
wate
rmark
and the
NC v
a
lue sho
w
s 1
.
The rob
u
stn
e
ss towa
rd
s
attacks ha
pp
ened b
e
cau
s
e it cau
s
ed
some
cha
nge
s
on t
he valu
e of
the
seq
uen
ce
compl
e
x on
t
he ve
ctor ma
pping
the
FF
T calculation
that
will be
sprea
d
for every F
FT value. Thi
s
will ma
ke t
he ch
ang
es
happ
ened
do
esn’t affect m
u
ch
and the FF
T value is
still in the wate
rm
ark
extra
c
ti
on
value limit. The re
sult will
be differe
nt when
we
do th
e in
sertio
n o
n
th
e spa
c
ey d
o
m
ain b
e
cau
s
e the
co
ordi
n
a
te value
ch
ange
s
will aff
e
ct
dire
ctly towa
rds th
e in
sert
ed waterm
ark bit valu
e. So, the calcula
t
ion sp
re
ad d
one by th
e F
F
T
whi
c
h is on
e
of the transform dom
ain
method
s that can maintai
n
the waterm
ark b
e
tter th
an
usin
g the sp
a
t
ial method.
Therefore, th
e ro
bu
stne
ss level of the
develop
ed t
e
ch
niqu
e is
determi
ned
b
y
som
e
things
su
ch a
s
the qu
ality of the asymm
e
tric
ke
y algo
rithm used; e
x
traction limit
used; the
kin
d
s
of the frequ
ency do
main
algorithm, a
nd so
me ot
h
e
r impo
rtant
things
su
ch
as data le
n
g
th
stora
ge an
d o
t
her rel
a
ted p
r
og
rammi
ng tech
niqu
es.
4. Conclusio
n
The co
ncl
u
si
ons of this
re
sea
r
ch are:
1.
Rob
u
st wate
rm
arking
tech
nique ba
se
d on the domai
n transfo
rm
with FFT wa
s succe
ssf
ully
con
d
u
c
ted to embed a
cop
y
right marke
r
into vector m
ap.
2.
The invi
sibilit
y and fidelity
level sh
own
by this
exp
e
ri
mental
re
sult
s of th
e tech
nique
proves
that the
simil
a
rity an
d fide
lity level of
the
wate
rma
r
ked
vecto
r
m
ap i
s
ke
pt. T
he di
sto
r
tion
scale
re
pre
s
e
n
ted by
the
RMSE value i
s
clo
s
e to
zero
and th
e fa
rth
e
st di
stan
ce
differen
c
e
is
51 cm.
3.
By doing insertio
n in the freque
ncy
domain, an
y change
s o
c
cur on the
vector map
coordinate will be
spread on t
he
other frequency
domai
n val
ue so it will
not affect
signifi
cantly t
o
the inserted copyri
ght. It w
ill m
a
ke t
he inserted c
opyright
becomes more
reliabl
e towards a
n
y cha
n
g
e
s. The l
e
vel
of the
r
o
bu
s
t
ne
ss
techniqu
e towa
rd
s an
y translatio
n
attacks, rotati
on, and scal
e
chan
ging rea
c
he
s
60% of the wh
ole eval
uation sce
narios.
4.
The
usa
ge
of
FFT f
r
equ
en
cy do
main
al
gorithm
can
maintain th
e
pre
c
isi
on l
e
vel of the
resul
t
vector ma
p, with an a
c
cep
t
able disto
r
tio
n
scale.
The usage
of
RSA asy
mmetric cryp
tograp
h key
algo
rithm g
i
ves
the
dat
a
o
r
igin
authenti
c
atio
n se
rvice but
it will decrea
s
e the robu
stn
e
ss level of the inserted
copyright towa
rds
any geomet
ri
c attack hap
p
ened.
Evaluation Warning : The document was created with Spire.PDF for Python.