TELKOM
NIKA
, Vol.12, No
.2, June 20
14
, pp. 397~4
0
4
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v12i2.1915
397
Re
cei
v
ed
No
vem
ber 2
4
, 2013; Re
vi
sed
April 21, 201
4; Acce
pted
May 12, 20
14
Geometric Feat
ure Extraction of Batik Image Using
Cardinal Spline Curve Represen
tation
Aris Fanani, Ann
y
Yuniar
ti, Nanik Suciati
Dep
a
rtment of Informatics En
gin
eeri
ng, In
stitut
T
e
knolo
g
i S
epu
luh N
o
p
e
m
ber (IT
S
)
Surab
a
y
a, Indo
nesi
a
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: arisfana
nie
@
gm
ail.c
o
m, ann
y@
if.its.ac.id, nanik@
if.its.ac.id
A
b
st
r
a
ct
Batik is
an
Ind
ones
ian
n
a
tio
n
a
l
herita
g
e
w
h
i
c
h h
a
s
bee
n r
e
cog
n
ise
d
as
a w
o
rld
cult
ura
l
h
e
ritag
e
(w
orld h
e
ritag
e
)
. Batik is
w
i
d
e
ly
used
as
cl
othin
g
mat
e
ria
l
. T
he a
d
va
nce
m
e
n
t of tec
h
n
o
lo
gy a
llow
e
d
the
mater
i
al
o
p
ti
mi
sation
in
cl
oth
i
ng
des
ig
n. Geo
m
etric
a
l
info
rmati
o
n
ab
out
ab
atik i
m
ag
e
is r
equ
ire
d
i
n
a
mo
du
le for opt
imisin
g cloth
i
n
g
desi
gn w
i
th batik as
a raw
materi
al. Geo
m
etric fe
ature
extraction of b
a
tik
imag
e is us
ed
to hel
p a co
mp
uter to reco
gni
se batik'
s
p
a
ttern or
motif. T
h
i
s
researc
h
pro
poses
a
metho
d
for geo
metric feature
extracti
on of b
a
tik i
m
a
ge by
us
i
ng ca
rdin
al sp
lin
e cu
rve repr
esent
ation. T
he
metho
d
for geo
metric feature
extracti
on is d
i
vid
ed
i
n
to 2 pr
ocess
e
s, i.e., feature extractio
n
fo
r Klow
ong
an a
n
d
feature extracti
on for Isen-Ise
n
. Klow
ong
an
repres
ents th
e
pattern of bat
ik imag
e, w
hereas Isen-Ise
n
i
s
content
pattern
s of Klow
on
ga
n. F
eat
ure extraction of
Kl
ow
ong
an is
p
e
rfor
me
d by
de
leti
n
g
col
lin
ear
po
i
n
ts
from ob
ject bo
und
aries u
n
til
the
do
mi
na
nt poi
nts are obt
ain
ed. T
he
do
mi
na
nt points
are then us
ed
as
control p
o
i
n
ts. F
eature extra
c
tion of Isen-I
s
en is
perf
o
rmed by savi
ng
coord
i
nates
of every con
nec
ted
compo
nent w
h
i
c
h are
als
o
us
ed as c
ontro
l p
o
ints. Geo
m
etri
cfeature of
bati
k
imag
e
is re
pr
esente
d
as
a s
e
t
of control
poi
nt
s of klow
ong
an
and
isen-
ise
n
. Batik i
m
a
ge c
an b
e
reco
nstr
ucted by
draw
i
ng car
d
in
al s
p
li
ne
curve
usin
g
a
set
of co
ntrol
po
ints
in t
h
e
ge
o
m
etric
re
prese
n
tatio
n
.
T
he ex
peri
m
e
n
t show
s th
at the
reconstructe
d i
m
a
ge is vis
ual
l
y
simi
lar tothe
origi
n
a
l
batik i
m
a
ge.
Ke
y
w
ords
: Ba
tik, Geometric f
eature extracti
on,
Card
ina
l
sp
line, C
u
rve rep
r
esentati
o
n
1. Introduc
tion
Batik is
a craft that has
high a
r
tisti
c
value an
d h
a
s b
e
come
part of the
Indon
esi
a
n
culture (e
sp
e
c
ially Java
) for a long tim
e
. Etymologica
lly, the suffix "-thik" in the word "bathi
k", is
derived f
r
om
the wo
rd
dri
p
or tri
c
kle.
The
word
"m
bathik"
com
e
s from th
e
word "thi
k" whi
c
h
mean
s
small
[1]. Thus it
ca
n be
sai
d
that
"mbathi
k"
is
writing or drawing
complex
(small) obj
ects.
Batik as th
e famous
and
uniqu
e traditi
onal leg
a
cy i
n
Indone
sia i
s
also re
co
g
n
ise
d
as
a world
cultural h
e
rit
age. Th
e re
cog
n
ition of
batik a
s
wo
rl
d cultural
he
ritage
ha
s m
ade b
a
tik
m
o
re
famous
and
widely u
s
ed
as
clothing
material. Bati
k ha
s uni
qu
e motif cha
r
acteri
stics. T
h
e
decoratio
n a
nd bati
k
moti
fs are de
sig
n
ed through
h
u
man
cog
n
itive pro
c
e
s
ses that rep
r
e
s
e
n
t
artistic interp
retation
s of t
he
surrou
ndi
ngs. T
h
is is
rega
rd
ed a
s
one of th
e
most inte
re
st
ing
asp
e
ct
s to be
studied u
s
in
g sci
en
ce an
d techn
o
logy.
In developin
g
a system for fashion d
e
si
gn aut
omatio
n and mate
ri
al (batik) opti
m
isation,
a method to
extract ge
om
etric featu
r
e
s
of batik
ima
ge is
requi
re
d. The extra
c
tion method
can
help a
comp
uter to
re
cog
n
ise
patterns of a
bat
ik image. Extra
c
tion feat
ures
play an i
m
po
rtant
role in dete
c
t
i
ng the obje
c
t correctly [2].Geometri
c fe
ature extra
c
ti
on ha
s an importa
nt role
in
unde
rsta
ndin
g
obje
c
t sha
pe in batik
pattern
s or
motifs. Geo
m
etrical feature
s
of an
obje
c
t
are
c
on
structe
d
by a
set of
geomet
rical e
l
ements su
ch
as
point
s, lin
es,
curve
s
o
r
su
rfaces. T
h
e
feature
s
in this context include geo
metric feat
ure
s
of
object bou
n
dary (kl
o
won
gan in Java
n
e
se
langu
age
) an
d geomet
ric f
eature
s
of obj
ect filler (i
sen
-
ise
n
in Java
nese lang
uag
e).
A pro
c
e
ss of featu
r
e
extra
c
tion
is followe
d after the
process
of imag
e
segm
entation
[
3].Segmenta
t
ion is a very important
st
ep in obje
c
t reco
gnition. T
here a
r
e vari
ous
segm
entation
method
s that can be
used.
One of t
hem
is the thre
sh
olding meth
o
d
whi
c
h is oft
en
use
d
be
cau
s
e it is easy and intuitive. Another
met
hod is n
eutroso
phi
c-b
a
se
d segm
entati
on,
whi
c
h ha
s su
ccessfully se
parate
d
obje
c
ts and ba
ckground [4].
Many studie
s
o
n
obje
c
t rep
r
e
s
entatio
ns
have
bee
n cond
ucte
d. The
stu
d
ie
s incl
ud
e
polygon a
pproximation an
d cu
rve re
p
r
e
s
entatio
n. A set of points
of
object bo
und
ary (contou
r)
is
use
d
to obtai
n polygon a
p
p
roximatio
n
o
f
the obj
ect [5
]. Then, domi
nant point
s are obtaine
d from
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: 397 – 40
4
398
the contou
r
b
y
removin
g
collinea
r p
o
int
s
. If the
sh
ap
e of the
obj
ect contai
ns curve eleme
n
t, then
the polygo
n
approximatio
n ca
nnot
pro
v
ide sati
sfac
t
o
ry re
sult
s.
Another
app
roach
that ca
n
be
use
d
to
rep
r
ese
n
t curve
element
quite
well i
s
cu
rve
rep
r
e
s
entati
on
meth
od. Bezie
r
curve
is
a
polynomial
curve of n de
gree
whi
c
h combine
s
c
ont
rol point
s in image de
picti
on. One of the
wea
k
n
e
sse
s
of Bezie
r
cu
rve is th
e ab
sence of
lo
cal
co
ntrol
prop
erty be
cau
s
e
shifting
of o
n
e
control p
o
int
will affect the
overall
sh
ap
e of
the
curv
e. This
we
akness h
a
s
urg
ed an i
dea t
o
multiply
low
d
egre
e
Be
zie
r
curve
segm
e
n
ts calle
d ca
rdinal spli
ne. Grap
hic
obj
e
c
ts are split
i
n
to
several segments so
that
shifting of
cont
rol
points will m
odif
y
only som
e
part
s
of
curve
segm
ents.
Other
studie
s
related to
ba
tik image i
n
cl
ude
motif-ba
sed batik retrieval
system [6]
and
motif classification of batik image [7]. This re
search prop
osed
a method
of ge
ometri
c featu
r
e
extraction
of
batik im
age
and
rep
r
e
s
e
n
ted the feat
ure
s
a
s
cont
rol p
o
ints
of ca
rdinal
spli
ne
curve
s
. The
propo
se
d method con
s
ists of thre
e
parts: extra
c
tion of geo
metric featu
r
es,
rep
r
e
s
entatio
n of the features u
s
in
g ca
rdinal
spline curve, and reconstructio
n
of batik image.
2. Literature
Stud
y
This
se
ction
explain
s
the t
heori
e
s that
serve
a
s
th
e b
a
si
s for this rese
arch. T
h
e
y
inclu
de
a ne
utro
sop
h
i
c
a
p
p
r
oa
ch
to segme
n
tation a
nd th
e
Cardin
al
splin
e
.
Free
man
ch
ain
cod
e
, can
n
y
edge
dete
c
tion, and
co
n
necte
d comp
onent la
belin
g ar
e
not di
scusse
d furt
her in thi
s
study
becau
se tho
s
e method
s are very comm
on in image p
r
ocessin
g
.
2.1. Neutros
ophic appro
ach to s
e
gm
enta
tion
Neutrosophy
is a bra
n
ch
of Philosop
h
y t
hat studi
es the o
r
igin,
nature a
nd
scope of
neutralitie
s. Neutrosophy can be co
nsi
dere
d
a
s
a
p
r
opo
sition, th
eory, event,
con
c
e
p
t or e
n
tity.
Suppo
sed
th
at <A
> i
s
a
n
event or
entit
y, <No
n
-A
> i
s
n
o
t <A
>, a
nd
<Anti-A> i
s
the
op
po
site of
<A>. <Ne
u
t-A> is d
e
fined
as the a
dditi
on of not
<A
> an
d not <A
nti-A>. Fo
r e
x
ample, if <A> =
white, then <Anti-A> = bl
a
ck.
<No
n
-A
> = Blue, yello
w, red (othe
r
than white
)
. <Neut-A
> = Bl
ue,
yellow, red
(o
ther than white and bla
c
k).
In this pap
er, an image i
s
tran
sfo
r
me
d into a ne
u
t
roso
phi
c do
main. A pixel in the
neutro
so
phi
c
domain
ca
n b
e
rep
r
e
s
e
n
te
d as
P
{
T
,
I
,
F
}, meanin
g
th
at the pixel h
a
s
t
% of true,
i
%
of intermedi
ate, and
f
% of false, with th
e
value of
t
,
i
, and
f
ran
ge from 0 to 100.
In fuzzy logi
c,
the
i
value = 0.
Neutrosophi
c approa
ch to
segm
entation
based on
wa
tersh
ed m
e
th
od ha
s be
en
carrie
d
out [3]. Step
s in the neut
rosophi
c ap
p
r
oa
ch to
se
g
m
entation ba
sed o
n
wate
rshe
d method
is
descri
bed a
s
follows:
2.1.1. Mapping and de
ter
m
ination of {
T
, F}
At this point, mappin
g
and
determin
a
tio
n
of the imag
e matrix T do
main and F
domain
are
don
e. T i
s
the
obje
c
t
and F
is th
e
ba
ckgro
und.
To d
e
termi
n
e the valu
e
of T an
d F t
hat
inclu
de neut
roso
phi
c com
pone
nts, the followin
g
S-fu
nction i
s
used
:
,
,
,
,
0
0
,
,
1
,
1
,
,
1
,
,
whe
r
e
g
xy
is the inte
nsity
value
of pix
e
l
P(i, j)
. V
a
riable
s
a, b,
and
c
are
para
m
eters t
hat
determi
ne th
e sha
pe of the S-fun
c
tion
. The value of the variabl
es
a, b,
and
c
are
cal
c
ula
t
ed
based on the
histog
ram [8]:
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Geom
etric F
e
ature Ext
r
action of Batik Image Us
ing Cardinal Spline Cu
rve .... (Aris Fanani)
399
1.
Comp
ute the image hi
stog
ram
2. Determine
th
e
local m
a
xim
a
of the histo
g
ram,
Hi
s
ma
x
(g
1
), His
ma
x
(g
2
),…, His
ma
x
(g
k
)
3.
Cal
c
ulate the
averag
e valu
e of loca
l ma
xima with the followin
g
equ
ation:
∑
4. Determine
th
e
local m
a
xi
m
a
as a pea
k wh
ose heig
h
t exceed
s
His
ma
x
(g)
. As
s
u
me that
the pea
k is first discovered
as
g
mi
n
and then
g
ma
x
is found.
5.
Specify the lower limit of
grey le
vel B
1
d
an upp
er limit
of
B
2
:
∑
and
∑
whe
r
e
f
1
=
0,0
1
(o
btaine
d from the
re
sult
s of th
e ex
p
e
riment).
g_mi
n
is a
grey l
e
vel value
greate
r
than
0 and wa
s first discovered
. While
is a
gre
y
level
value gre
a
ter th
an 0
and it is the value that is fo
und the la
st.
6.
Determine th
e value of a and c pa
ram
e
ters:
1
,
i
f
(
a>B
1
)
,
a= B
1
if
(c>
B
2
), c= B
2
7.
Cal
c
ulate b p
a
ram
e
ter u
s
in
g the prin
cipl
e of maximum entropy:
1
,
,
whe
r
e
S
n
( ) i
s
Shann
on fu
nction
d
e
fine
d as:
,
,
,
1
,
1
,
Value pa
ram
e
ter
b
li
es
b
e
twee
n value
s
a
a
nd
c.
T
o
get the
opt
imal value of
b, the
che
c
king o
n
all po
ssibl
e value
s
of b
is necessa
ry. The optimal va
lue of
b
will
g
enerate
the greate
s
t value of maximum entro
py
H(X)
:
,
,
,
m
a
x
,
,
,
|
.
2.1.2. Enhancement
Having obtai
ned the ne
w image on n
eutro
sop
h
ic
domain, en
h
ancement proce
s
s is
done. T
h
is p
r
ocess
aim
s
to improve th
e imag
e
of t
he n
e
w
dom
ain. Enha
nce
m
ent p
r
o
c
e
s
s i
s
done
usi
ng i
n
tensity tran
sf
ormatio
n
. He
re is
a fun
c
ti
o
n
that is u
s
ed
to ma
ke im
provements to t
h
e
image in the
neutro
so
phi
c domain:
,
2
,
,
0
,
0
.
5
,
,
1
2
1
,
,
0,5
,
1
Whe
r
e
E (T(x
, y
)
is im
ag
e enha
ncem
ent of domai
n neutro
so
ph
ic.
T(x, y)
im
age in dom
a
i
n
neutro
so
phi
c.
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: 397 – 40
4
400
2.1.3. Thresh
olding
One way to retrieve the ob
jects from its
backg
rou
nd i
s
to sel
e
ct a t
h
re
shol
d valu
e T that
can
sepa
rate
the group
s from ea
ch oth
e
r. Thre
sh
old
value is determin
ed by using a he
uri
s
tic
approa
ch [9]:
1.
Determine th
e initial threshold
t
0
on
f(x,y
)
2. Separate
f
(
x,
y)
u
s
ing
t
0
, an
d then gro
up
them into 2 group
s of new
pixels,
F
1
and
F
2
3.
Find the mea
n
values of
μ
1
and
μ
2
from each gro
up
F
1
and
F
2
4.
Comp
ute a n
e
w thre
sh
old
value usi
ng the equ
a
tion t
1
= (
μ
1
+
μ
2
)/2
5.
Rep
eat ste
p
2 thro
ugh
4
so that the
differen
c
e
bet
w
e
en the val
ue
of t
n
– t
n-1
<
ε
(where
ε
= 0.0001
). If these conditi
ons a
r
e me
t, tn is define
d
thre
shol
d valu
e.
2.1.2. Cardin
a
l Spline
Cardinal
spli
n
e
is a
splin
e i
n
terpol
ation u
s
i
ng th
e pull
(tensio
n) to fo
rm a
curve.
Cardin
al
splin
e interpo
l
ation is a m
odificatio
n
of the quad
rati
c Bazi
er
spli
ne u
s
ing the
spli
cing p
r
o
c
ess
with C1 co
ntinuity [10]. One segme
n
t of the ca
rdin
al splin
e curve is define
d
by four cont
rol
points, the curve will inte
rpolat
e the
control poi
nts
and the fourt
h
must satisf
y the following
equatio
n:
1
1
2
1
2
1
1
2
2
1
2
1
1
1
0
1
0
0
1/
0
0
whe
r
e
is ten
s
ion p
a
ramet
e
r, u is the ve
ctor of
knot
s, and pi a
r
e the
control point
s. In this stu
d
y
,
used i
s
0.5.
3. Geometric
feature ex
tr
action o
f
Ba
tik Image
Geomet
ric fe
ature
extracti
on of bati
k
i
m
age
ha
s a
n
impo
rtant role in u
nde
rstandi
ng
obje
c
t shap
e
s
in
a
bati
k
im
age. G
eom
etric featu
r
e
s
are featu
r
e
s
of
obje
ct con
s
tructed
by
a
se
t of
geomet
rical elements li
ke points, line
s
, curve
s
or
surface
s
. Geom
etric featu
r
es in a batik imag
e
are g
eomet
ri
c features
of pattern
(kl
o
wonga
n in
the
Javan
e
se lan
guag
e) a
n
d g
e
ometri
c feat
ure
of filler pattern (isen-i
s
en
in the Javanese language).
3.1 Geome
t
ric fea
t
ure e
x
traction o
f
klo
w
o
nga
n
Klowon
gan i
n
the batik im
a
ge is th
e arch
etypal
form of
the batik im
a
ge. Klowo
nga
n of a
batik im
age
i
s
sho
w
n i
n
Fi
gure
1. G
e
o
m
etric extr
a
c
tion featu
r
e
of klo
w
on
gan
is sh
own
in Fi
g
u
re
2a.
a
b
Figure 1. (a)
Batik image; (b) Klowong
an
Geomet
ric fe
ature extracti
on
process b
egin
s
by ente
r
ing
RGB im
age a
n
d
con
v
erting it
to greyscale.
After imag
e
convertion
to
greyscal
e, filtering
is d
one
to elimi
n
ate
noise. Denoi
sed
image will
be
segm
ented u
s
ing neutroso
phic-ba
se
d
th
re
shold as de
scribe
d
in se
ction
2.1.
F
r
o
m
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Geom
etric F
e
ature Ext
r
action of Batik Image Us
ing Cardinal Spline Cu
rve .... (Aris Fanani)
401
the seg
m
ent
ation re
sult,
conn
ecte
d
comp
one
nt labeling
can be d
e
te
rmine
d
usi
n
g 8
-
neigh
borhoo
d
.
Each co
mp
onent lab
e
lin
g will dete
r
mi
ne dire
ction
d
a
ta of object
in the image
b
y
mean
s of ch
ain cod
e
me
thod and co
ordin
a
te po
si
tion object
s
will be sto
r
e
d
. Chain co
de
algorith
m
use
d
in the extraction
8-co
nne
cted is a
s
foll
ows [11]:
Determine th
e pixel of the obje
c
t who
s
e
val
ue is
in top left row; assume this
pixel
P
0
Determine
the
dir
v
a
ri
able
(for
di
re
ction
s
). Set
dir
=
7 (s
inc
e
P
0
is
the top-left pixel in the
objec
t).
Traverse
the
3x3 nei
ghb
o
u
rho
od
of cu
rre
nt pixel
s
. Begin the
se
arch at th
e p
i
xels in th
e
dire
ction dir +
7
(m
od 8)
for even dir or
dir
+
6
(mo
d
8) for
odd
di
r (Figu
r
e
3
(b
-d)). Ta
ble
1
sho
w
s the direction of the
dir and the n
e
x
t dir
that resulted in anticl
o
ckwi
se di
re
ction:
Table 1. Di
re
ction an
d nex
t direction
dir
0
1
2
3
4
5
6
7
dir+7(
mod 8
)
7
0
1
2
3
4
5
6
dir+6 (
mod 8
)
6
7
0
1
2
3
4
5
The first foreground pixel
will
be the new boundary
element. Update
dir
.
Stop when t
he cu
rrent b
ound
ary ele
m
ent
P
n
is equal to the seco
nd elem
e
n
t
P
1
and the
previou
s
bo
u
ndary pixel
P
n-1
is equal to the first boun
dary elem
ent
P
0.
Figure 3 sh
o
w
s the d
e
termination of
P
0
and determi
nation of the dire
ction
s
usi
ng ch
ain
cod
e
alg
o
rith
m. Geom
etri
c featu
r
e
red
u
ction
of
kl
o
w
on
gan
i
s
p
e
rform
ed by finding
d
o
min
ant
point in
ea
ch
sh
ape
of ob
ject b
a
sed o
n
the
di
re
ctio
n chain
co
de
. Step by ste
p
to d
e
termi
n
e
dominant poi
nt by removing collinear points [4]:
Select thre
e points a
s
initi
a
l point. For e
x
ample
P
i
, P
j
,
dan
P
k
Define thresh
old value, (
d
t
)
Delete point Pj with dis
t
anc
e
value (
d
) o
f
the straight l
i
ne from
Pi
and
P
k
.
If
d
≤
d
t
. distance (
d
) i
s
cal
c
ulate
d
as:
Rep
eat step
c, and sto
p
if P
j
= P
i
.
Figure 4
sh
o
w
s a m
e
thod
to obtain
dom
inant p
o
ints.
After domin
a
n
t point
s a
r
e
obtaine
d
from the
red
u
ction
process, coo
r
dinate
s
of th
e
d
o
m
i
nant p
o
int of
ea
ch
obje
c
t
are
sto
r
ed
in
a
database that
will be used for image re
construction usi
ng cardinal spline.
3.2 Geome
t
ric fea
t
ure e
x
traction o
f
isen-ise
n
Isen-i
s
e
n
i
s
klowo
ngan
fille
r of b
a
tik im
a
ge.
Example
isen
-isen
of b
a
tik ima
ge i
s
sho
w
n
in Figure 5. Geomet
ric fe
ature extra
c
ti
on proce
s
s
of
isen
-ise
n is
sho
w
n in Fi
g
u
re 2
b
. Geo
m
etric
feature extra
c
tion of i
s
en
-ise
n
is
don
e by inco
rp
o
r
ating b
a
tik
RGB ima
ge,
conve
r
t it into
grayscal
e an
d inse
rt the image segm
entation re
su
lts. Bounda
ry detection of
the graysca
l
e
image i
s
do
n
e
usi
ng Cann
y method. Isolation of
ise
n
-isen i
s
carried out to obt
ain ise
n
-i
se
n
o
f
batik ima
ge.
This p
r
o
c
e
ss is do
ne by
multip
lying th
e matrix ima
ge of Canny
edge
dete
c
tion
results with t
he matrix of ero
s
ion
-
segm
ented im
ag
e. The next pro
c
e
ss i
s
co
nn
ected
comp
o
nent
labeling and
storing the isen-i
s
en
coordinates i
n
to the database
wh
ich
will further be used as a
control point i
n
the re
con
s
truction by u
s
in
g a cardinal
spline.
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: 397 – 40
4
402
Ci
tr
a RG
B
Ba
t
i
k
Se
g
m
e
n
t
a
s
i
de
ng
an
t
h
r
e
s
h
o
l
d
be
r
b
as
i
s
ne
utr
o
s
o
ph
i
c
C
o
nn
ec
t
e
d
C
o
mp
on
en
t
La
be
l
i
n
g
P
e
ne
nt
u
a
n
ar
ah
b
a
t
a
s o
b
je
k
de
ng
an
C
h
a
i
n
Cod
e
Re
d
u
k
s
i
f
i
t
u
r
ge
ome
t
r
i
k
l
owon
ga
n
D
a
tab
a
s
e
fi
t
u
r
ge
ome
t
r
i
k
l
owon
ga
n
D
e
n
o
isin
g
de
ng
an
M
e
a
n
F
ilt
e
r
in
g
Ub
a
h
k
e
Gr
a
y
s
c
a
l
e
C
i
tr
a G
r
ay
s
c
a
l
e
B
a
ti
k
,
C
i
tr
a
S
e
g
m
entas
i
,
Ci
t
r
a
RG
B
D
e
tek
s
i
T
epi
C
anny
C
i
t
r
a G
r
ay
s
c
al
e bat
i
k
Is
ol
a
s
i
I
n
ter
i
or
Ci
t
r
a
=
Ha
s
i
l
Ca
n
n
y
-
er
o
s
i
(
C
i
tr
a
S
e
g
m
entas
i
)
D
a
tabas
e fi
tur
geom
et
r
i
i
s
en-
is
e
n
R
eduk
s
i
f
i
tur
geom
et
r
i
i
s
en
-
i
s
e
n
C
onnec
t
e
d
C
o
m
ponent
Label
i
n
g
a
b
Figure 2.(a
) Geomet
ric fe
ature extra
c
ti
on of klo
w
on
gan; (b
) Geo
m
etric feat
u
r
e
extraction of
isen
Figure 3. (a)
Determinatio
n
P
0.
(b-d)
De
terminin
g the nex
t directio
n
Figure 4. Re
moval pro
c
e
s
s colli
nea
r poi
nt.
a
b
Figure 5. (a)
Batik image ; (b) Isen
-ise
n
3.3 Batik ima
g
e recon
s
tr
u
c
tion of c
a
rd
inal spline curv
e representa
tion
Re
sult of geo
metric fe
ature extractio
n
of
isen
-isen and klowong
an
is rep
r
e
s
e
n
ted
in
a
set of domi
n
a
n
t points. Thi
s
set of domi
nant poi
nt
s th
at has
been
stored in th
e d
a
taba
se
will b
e
use
d
a
s
a
co
ntrol p
o
int in t
he bati
k
ima
g
e
re
co
nstructi
on u
s
ing
a
ca
rdinal
spli
ne
as d
e
scribed
in
se
ction 2.2.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
9
30
Geom
etric F
e
ature Ext
r
action of Batik Image Us
ing Cardinal Spline Cu
rve .... (Aris Fanani)
403
0
50
100
150
200
25
0
300
350
0
50
100
150
200
250
300
350
0
10
0
20
0
300
400
500
600
0
100
200
300
400
500
600
0
50
10
0
150
20
0
25
0
30
0
350
0
50
10
0
15
0
20
0
25
0
0
50
100
15
0
20
0
250
30
0
350
40
0
45
0
500
0
50
10
0
15
0
20
0
25
0
30
0
35
0
40
0
45
0
0
50
10
0
150
200
25
0
30
0
350
400
45
0
0
50
10
0
15
0
20
0
25
0
30
0
35
0
4. Results a
nd Analy
s
is
The
data
use
d
in thi
s
stud
y is the
ima
g
e
of b
a
tik
ma
dura.
Image
s of bati
k
m
a
d
u
ra
we
re
obtaine
d by t
a
kin
g
di
re
ct p
hotogr
aph
s from the b
a
tik
craft
s
men.
A
se
rie
s
of ex
perim
ents
we
re
con
d
u
c
ted to evaluate the
prop
osed sy
stem. T
able 2 sho
w
s the re
sults of expe
riment.
Table 2. The
Re
sult of Experime
n
t
N
o
Ba
ti
k
Ima
g
e
Kl
o
w
on
ga
n
Ise
n
-
i
s
en
R
e
su
l
t
1
Size : 687
x 63
0
T
i
me : 881.98
secon
d
2
Size : 381
x 39
2
T
i
me : 95.22 secon
d
3
Size : 376
x 29
2
T
i
me :62.87 second
4
Size : 530
x 46
9
T
i
me
: 1152.7
7
secon
d
5
Size : 640
x 48
0
T
i
me : 342.61
secon
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: 397 – 40
4
404
The ori
g
inal i
m
age, si
ze a
nd pro
c
e
s
sin
g
time are sh
own in
colum
n
1. Black an
d white imag
e as
input from
kl
owo
nga
n extraction p
r
o
c
e
s
s is
sho
w
n
i
n
colum
n
2. Isolated ima
g
e
of Isen-i
s
e
n
is
sho
w
n in
col
u
mn 3. The
results of th
e pro
p
o
s
ed
algorith
m
are
in colu
mn 4
.
The re
sult
of
con
n
e
c
ted co
mpone
nt lab
e
ling of the klowo
ngan
an
d isen
-i
sen i
m
age i
s
not sho
w
n b
e
cau
s
e it
contai
ns the
sa
me
bla
c
k and
white i
m
age
a
s
the
klo
w
o
nga
n
and i
s
e
n
-i
se
n, only th
at
the
con
n
e
c
ted co
mpone
nt
lab
e
ling contai
n
s
o
ne obje
c
t whi
c
h
i
s
con
necte
d
to 8-n
e
ighb
ourhoo
d
.
In
all the batik i
m
age
s used i
n
the experim
ent, our al
go
rithm can inte
rpolate a set of control poi
nts
whi
c
h
are
gi
ven ba
se
d o
n
cardinal
sp
line
cu
rve re
pre
s
entatio
n
of klo
w
o
nga
n
and
i
s
en
-ise
n.
Visually, the
result of
re
co
nstru
c
tion
u
s
i
ng
cardinal
spline give
s
si
milar
re
sults
to the o
r
igin
a
l
image. From
the experim
e
n
t, the third
batik ima
ge
has th
e faste
s
t pro
c
e
s
sing
time, i.e. 62.87
se
con
d
s,
whil
e the fou
r
th
batik ima
ge
has th
e
lon
g
e
st p
r
o
c
e
ssin
g
time, i.e. 1152.77
se
co
n
d
s.
Based
on
the
image
p
r
o
c
e
ssi
ng
re
sult o
f
klo
w
on
gan
and i
s
e
n
-i
sen
,
the third b
a
tik im
age
ha
s t
h
e
least
con
n
e
c
ted comp
one
nt labelin
g, which
re
sult
s i
n
fewe
r
set
s
of cont
rol p
o
i
n
ts. The
fourth
batik im
age,
on the
other
hand, h
a
s the
most
con
n
e
c
ted co
mpo
n
e
n
t labelin
g, which
also results
on mo
re
sets of control
po
ints. Th
e p
r
o
c
e
ssi
ng time
in this expe
ri
ment is not
d
e
termin
ed
by the
image
size. Instea
d, it is
determi
ned
b
y
the set of
control
points
use
d
in the
repre
s
e
n
tation
of
cardinal splin
e
curve.
5. Conclusio
n
Experimental
re
sults
sh
o
w
that
the propo
sed syste
m
ca
n pe
rform geom
etri
c
feature
s
extraction
of
batik ima
ge,
re
pre
s
e
n
t th
e featu
r
e
s
in
to so
me
co
n
t
rol p
o
ints of
ca
rdi
nal
spli
ne
curve,
an
d
re
con
s
tru
c
t th
e
batik ima
ge
by usi
ng th
e
cardinal
splin
e curve
repre
s
entatio
n. Th
e
result sho
w
s that recon
s
tru
c
ted imag
e is vis
ually simil
a
r to the origi
nal batik ima
ge.
Ackn
o
w
l
e
dg
ement
This pa
pe
r is a part of a rese
arch fund
ed
by Indone
sian
Dire
cto
r
ate Gene
ral
of Highe
r
Educatio
n (DIKTI) throug
h
Strategis
Nasio
nal
(Stra
nas) Research G
r
ant, 20
13, and i
s
al
so
s
u
pported by J
I
CA Predic
t ITS.
Referen
ces
[1]
Kus
w
a
ji. Men
g
ena
l Se
ni Bati
k di Yo
g
y
ak
ar
ta. Yog
y
ak
arta
. Pro
y
ek Pe
ng
emba
nga
n Per
m
useum
an
Yog
y
ak
arta.
[2]
Ahsan M, M
Dzulk
i
fli.
F
e
atures Extract
i
on for
Objec
t
Detectio
n
Based
on
Interest Poi
n
t.
T
e
leco
mmunic
a
tion, Co
mputi
ng, Elec
tron
ics
and Co
ntrol (T
ELKOMNIKA).
201
3; (11): 271
6-27
22.
[3]
G Patil B, Mane N, Su
bb
arama
n
S.
Iris F
eature Extraction a
nd C
l
assificati
on
u
s
ing F
P
GA.
Internatio
na
l Journ
a
l of Electr
ical
a
nd Co
mp
uter Engi
ne
erin
g (IJECE).
2012: 214-
22
2.
[4]
Z
ang M, Z
hang L, Chen
g H.
Neutroso
phic
Approc
h to Segmentati
on Ba
sed on W
a
ters
hed Meth
od
.
Scienc
e Direct
on Sig
nal Pr
oc
essin
g
.
201
0; (90): 151
0–
151
7
.
[5]
Po
yato C, Ma
drid-C
uav
es F
,
Medin
a
-C
arni
cer R.
Polyg
o
n
Appr
oxi
m
ati
on of Di
gita
l
Plan
ar Curv
es
T
h
roug
h Break
Point Sup
p
res
s
ion
.
Scie
nce
Direct on Patte
rn Reco
gniti
on.
201
0; (43): 14
– 25.
[6]
Eka R. Pe
ng
e
m
ban
gan
Siste
m
T
e
mu Kemb
ali
Citra B
a
tik
Meng
gun
aka
n
T
r
ansformasi W
a
velet
ya
ng
Dirotasi
dan M
u
lti La
ye
r Perc
ept
ron. Master
T
hesis. Suraba
ya: Postg
a
rd
ua
ted IT
S: 2011.
[7]
Arisan
di B. Peng
en
ala
n
Mo
tif Batik deng
an Rotate
d W
a
vel
e
t F
ilter dan Ne
ural N
e
tw
o
r
k. T
hesis.
Surab
a
y
a: Pos
t
gradu
ated IT
S: 2009.
[8]
Che
ng H,
W
a
n
g
X.
Microc
alci
fication
Detecti
on usin
g
F
u
zz
y Lo
gic and
Sc
ale Spac
e
A
p
p
r
oach.
IEEE
.
200
3: 363-
375.
[9] Gonzal
es
RC.
Digita
l
Imag
e
Processi
ng
. Ne
w
Jersey
: Prentice-Hall, In
c., Upp
e
r Sad
d
le
River. 20
02:
609-
610.
[10] Fussel
D.
Interpol
ating C
u
rve
s
.
T
e
xas: Univ
ersit
y
of T
e
xas
at Austin. 200
2
:
18-20.
[11]
Amizah N,
Mohamm
ad Z
J
. A
pplic
ation
of F
r
eeman
Chai
n C
odes
: An Alternati
v
e Reco
gn
itio
n
T
e
chnique for
Mala
ysi
an C
a
r Plates.
IJCSNS Internation
a
l
Journ
a
l of
Co
mp
uter Scie
nc
e and N
e
tw
ork
Security
. 200
2
: 222-22
7.
Evaluation Warning : The document was created with Spire.PDF for Python.