An Improved Dominant Point Feature for Online Signature Verification.

T.t

hrsrBsl1fifid,ffi*] sf $*curity

and lts Appllcations
vol.8. No,1 (2014)

n-Chief of the IISIA Journal
Jemal. H. AbawaiY,
Deakin University, Australia

piofessor ln Depa*ment of Computer Sclence at Purdue, USA

General Information of IJSIA
Blbllographlc Infortnatlon

r
o

ISSN:1738'9976
Publisher: SERSC

Scier,ce

&

Englneerlng Research

SupPist soCtetY

Conttct Informatloh
Sclence & Englneerlng Research
gupport coCletY
o Head office: 20 Virglnia Court, Sandy Bay, Tasrhania, Australia
Phorie
Email

no.: +51-3'9016-9027

: ilsla@sqrsc.org

Journal ToPies

technology
IIsIA aims to facilltate and support 5-e1-e.arch related to .security
followlns:'
the
lnclude
IJSIA
by
iorerii
ffirtiifinr. rhe topics

i

Baslc Toplcs:
a

SecurltY technology

a

Security Products or systems

q
-=l
'1

1

j

Secure installation, generatlon and operation

I
,1

I
I

{r

SecuritY assurance

and the

,'rF

ii.-iiii:i

!;:..t1;1::1s:. !.':::i:;_q.:'.

tnt€rnatisnar ro$rnar or security

"H,i!:

ffiiil1lil

Intdnrmttonal Jountal of'Sueurity and It* Apptleetiort

Janu*ry }A}J- Printsd
January }}L{Published

Publisher:
Fublishing

office:

SERSC
Science and Engineering Research Support soCiety
20 Virginia Coutt, Sandy Bay, tartnania, Australia

E-tnaii : sersc@sersc.org; ijsia@sersc.org
Tel: +61-3-9016-9021

Printing office:

Hanall Co. Ltd.
Tel: +82-2;2279-5494

L

&

E
E
ffi
i:.

Internationar Journar

"t

http ://6;.66;.o

:i::'fi:ii,|.ii:![Ti:l;

rs I I 0.

I 4257 / ilsia. 20 1 4. 8. l . 06

An Improved Dominant Point Feature for Ontine Signature
Verification

Danna Putral, Yogi Pratamaz, Oka Sudana3 and,qdi Purnawana
t'2'i'aDepartntent
of Information Techn?logy, (ldayana t)niversity, Indonesia
l
' ikgdarmaputra@gmail. com, yo gipratama. ib@gmail. com,
1
a
agungo kas @unud. ac. id, dos ena di@yaho o. c om

Abstract
Among the biometrie characteristic, signature forgery is the easiest way to do. Possibility
of signature forgery similarity might be reaehed perfectly. This paper intoduced a new
technique to improve dominant point feature system based on its location for online signature
verification. Dynamic Time lYarping is used to match two signature features vector. The

performance af system is tested by using 50 participants. Based on simulation result, system
d,ccuracy without presence of the simple and trained impostors is 99,65% with rejectian error
is A% and acceptanee error is 0.i5%. llthile the current systems arefaced with the simple and
trained impostors, system accuracy became 9/,.04% with rejection error is 1.6% and an
average af acceptance elror is 7.i6% with details as follaws; aeceptance error is 0.08%,

acceptance error of simple impostors is 4,4o/o, and acceptance error of trained impostors is
17.6%.The improved feature within fusion is produce better accuracy significantly than
dominant pointfeature. Accuracy of the improvedfeature withinfusion is 91.04%o, whereas
system dccuracy with just use the dominant pointfeature is 74.96%.

Keywords: Verification, Dominant Point, Biometric, Signature, Location of Dominant
Points

l.Introduction
Research and development of the biometric verilication of hurnan beings especially the
signatures has been widely applied. Several kinds of methods have been used to minimize the
level of signatrare forgery because signature is the easiest to forge when it compares to the
Possibility of signatures similarity might be reached
other biometric characteristics
people
perfectly. Few
realiee that the possibility of the direction of motion of the signature is
different for each person. It becomes the uniqueness of the signature itself, then for reasons
such as to minimize the possibility that the signature to be forged [2].

[].

Several methods have been applied to the biometrics(especially signatures) as
identification or distinguishing between people with each other, they ate domirant point [3],
stroke matching [4], based on writing speed [5], angle detection [6, 7], suppofi vector
machine [8], mouse based signatures [9], time sequence[0], localized arc pattern
[ll],dynamic RBF networks and time series motifs [12],4 features (pen position, time,
velocity, and pressure parameters) [3], local dominant orientation [14], etc.
Several studies have used dominant point as a research object or as ah object feature
extraction such as planar curves [15], digital curve [16], handwritten of some script [17, 19],
and also signature detection. Recognition rate of the previous study that used dominant point
as method for signature feature extraction in signature recognition is about 96% with 2A

rssN: 1738-9976 IJSIA
Copyrisht O 2014 SERSC

lnternational Journal of Security and lts Applications
Vol.8. No.1 (2014)

than 17 Yo, while

simple and trained impostors, the system cafl increase the accuracy more
*iifio"t simple and trained impostois can increase the accuracy more than 10 %. The online
*igntne verification system in this paper is very feasible to-be developed and applied for

uritfr.nti.otioo applicalions. For future work,

we would develop mobile

signature

authentication sYstem

References
tllD.Putra,"BiomefticSystem",Yogyakarta:ANDIOffset'Indonesia'(2009)'
ifiometriiAutqeglication Systerns', FI MU Report Series, (2000)'
itl Z. nifr" i"a V. lrtutyas,
t!iqll*.. Recoenition with Dominant Point Method"'
t3l F. Soedjianto, r