05-07-2012, 03:45 PM
Personal Authentication Using Signature Recognition
Personal Authentication.pdf (Size: 427.9 KB / Downloads: 85)
Abstract.
In this paper, a problem of personal authentication through the use of
signature recognition is described. The methods of verification include both online
(or dynamic) and off-line (static) signature verification algorithms. The
dynamic methods covered, are based on the analysis of the shape, speed, stroke,
pen pressure and timing information. While the static methods involve general
shape recognition techniques. The paper gives a brief historical overview of the
existing methods and presents some of the recent research in the field.
Introduction
A problem of personal verification and identification is an actively growing area of
research. The methods are numerous, and are based on different personal
characteristics. Voice [1], lip movements [2], hand geometry [3, 4], face [5, 6, 7],
odor [8, 9], gait [10, 11], iris [12], retina [13], fingerprint [14] are the most commonly
used authentication methods. All of these psychological and behavioral characteristics
are called biometrics. The biometrics is most commonly defined as measurable
psychological or behavioral characteristic of the individual that can be used in
personal identification and verification [15]. The driving force of the progress in this
field is, above all, the growing role of the Internet and electronic transfers in modern
society. Therefore, considerable number of applications is concentrated in the area of
electronic commerce and electronic banking systems.
The Nature of Human Signatures
It is reasonable to start this part with a general definition of what a signature is.
According to American Heritage Dictionary [19] signature can be defined as: “the
name of a person written with his or her own hand; the act of signing one's name”
[19].
Second definition refers to the whole process of signing, and brings us to the
assumption that the way the signature is made is a part of this signature [18]. Which
further leads to a hypothesis that the characteristics of the process of signing (i.e.
velocity, pen pressure, stroke etc.) are unique to every individual. [16] suggests that
the signature consists of a series of rapid movements. It is supposed that the features
of the process of signing originate from the intrinsic properties of human
neuromuscular system, which produces the aforementioned rapid movements.
Knowing that this system is constituted by a very large number of neurons and muscle
fibers, is possible to declare, based on the central limit theorem, that a rapid and
habitual movement velocity profile asymptotically tends toward a delta-lognormal
equation [16]. This statement explains stability of the characteristics of the signatures.
Thus, the signature can be treated as an output of a system observed in a certain time
interval, necessary to make the signature.
Off-line Signature Verification
Off-line signature verification problem has attracted a great deal of attention in the
past years; many results have been obtained [22-38]. However, these results are far
from being perfect, and do not give the accuracy required for many security problems.
For many years the problem of signature verification has generally been solved by
some authority or clerical employee, however with the invention of computers and
scanning devices the trend has been towards automation of the whole process.
During a period of more than 20 years many approaches to the problem of
automatic off-line signature verification have been created. The techniques used
include: 2D transforms [22], histograms of directional data [23-25], curvature [26],
horizontal and vertical projections of the writing trace of the signature [27], structural
approaches [28], local measurements made on the writing trace of the signature [29],
the position of feature points located on the skeleton of the signature [30]. One of the
best results in this area has been reported in [31], where the error rate was less then
one percent [32].
On-line signature verification
On-line signature verification is based on dynamic characteristics of the process of
signing. Since time-dependent way of representing the signature contains more
information, the accuracy of the recognition is significantly higher. The design and
implementation of the on-line signature verification systems involves data acquisition,
feature extraction, feature selection, decision-making, and performance evaluation
[38]. But, at the same time, dynamic signature verification process requires special
equipment to gather the information necessary for the verification process. Most
common is a digitizing tablet, which registers not only the trajectory and speed of the
process of signing, but also the pressure and pen tip position. These unique
characteristics allow verification of the genuine signatures.
Neural Networks
At some point of time neural network approach has become a cure-all tool. No
wonder that eventually this approach has been applied to the problem of automatic
dynamic signature verification. Let us constrict the description of the work done in
the area to two most interesting works [44, 45].
[44] constructed a three layer artificial neural network, trained using supervised
learning with back propagation. Momentum (h) and learning rate (l) equal to
respectively 0.9 and 0.1. Training was stopped when the maximum error reached the
value of 20%. Number of input neurons varied between 28-40, and the network
contained one hidden layer of log2n neurons, where n is the number of input neurons.
And the output contains one neuron, producing a result of either genuine or forged
signature judgment.
Conclusions
In this paper we have considered a problem of personal authentication through the use
of signature recognition. Both on-line and off-line methods have been described. The
method of signature verification, reviewed in this paper, benefits the advantage of
being highly acceptable by potential customers as compared to the rest of biometric
solutions. The driving force of the progress in this field is, above all, the growing role
of the Internet and electronic transfers in modern society. Therefore, considerable
number of applications is concentrated in the area of electronic commerce and
electronic banking systems.