17-12-2012, 05:44 PM
Robustness of Offline Signature Verification Based on Gray Level Features
1Robustness of Offline.pdf (Size: 2.86 MB / Downloads: 107)
Abstract
Several papers have recently appeared in the literature
which propose pseudo-dynamic features for automatic
static handwritten signature verification based on the use of gray
level values from signature stroke pixels. Good results have been
obtained using rotation invariant uniform local binary patterns
LBP plus LBP and statistical measures from gray level
co-occurrence matrices (GLCM) with MCYT and GPDS offline
signature corpuses. In these studies the corpuses contain signatures
written on a uniform white “nondistorting” background,
however the gray level distribution of signature strokes changes
when it is written on a complex background, such as a check or
an invoice. The aim of this paper is to measure gray level features
robustness when it is distorted by a complex background and
also to propose more stable features. A set of different checks and
invoices with varying background complexity is blended with the
MCYT and GPDS signatures. The blending model is based on
multiplication. The signature models are trained with genuine
signatures on white background and tested with other genuine and
forgeries mixed with different backgrounds. Results show that a
basic version of local binary patterns (LBP) or local derivative
and directional patterns are more robust than rotation invariant
uniform LBP or GLCM features to the gray level distortion when
using a support vector machine with histogram oriented kernels
as a classifier.
INTRODUCTION
BIOMETRICS is playing an increasingly important role in
personal identification and authentication systems. Several
technologies that have been developed in this area are based
on fingerprints, iris, face, voice, the handwritten signature, hand,
etc.
Handwritten signatures occupy a very special place in this
wide set of biometric traits. The main reason is tradition: handwritten
signatures have long been established as the most widespread
means of personal verification. Signatures are generally
accepted by governments and financial institutions as a legal
means of verifying identity.Moreover, verification by signature
analysis requires no invasive measurements and people are used
to this event in their day to day activities
A handwritten signature is the result of a complex process depending
on the psychophysical state of the signer and the conditions
under which the signing process occurs. Although complex
theories have been proposed to model the psychophysical
mechanisms underlying handwriting and the ink processes, signature
verification is still an open challenge since a signature is
usually judged to be genuine or a forgery on the basis of only a
few reference specimens [1].
Gray-Level Distortion at Database
TheMCYT and GPDS960Gray signatures were blended with
the check database to obtain the synthetic signature database,
that is, a new database with distorted gray levels. There are
many different types of blending modes: darken, multiply, color
or linear burn, lighten, color or linear dodge, etc. We used the
multiply blend mode which multiplies the check image by the
signature one. As we overlay gray level strokes, each stroke results
in a new darker gray level. The pixels outside of the strokes
are unaffected because the white signature background does not
generate a change [23].
GRAY LEVEL BASED SIGNATURE FEATURES
Local Patterns
The local binary pattern (LBP) operator is defined as a gray
level invariant texture measure in a local neighborhood [24].
The original LBP operator labels the pixel of an image by
thresholding the 3 3 neighborhood of each pixel and concatenating
the results binomially to form a number. Assume that a
given image is defined as .
Histogram of Local Patterns
The gray level image is transformed to LBP
LDP or LDerivP code matrices. Each code matrix contains
information about the structure to which the pixel belongs:
the stroke edge, stroke corners, stroke ends, inside the stroke or
background, etc. We model the distribution of the local pattern
by spatial histogram to avoid losing the location of the different
structures inside the image. Therefore, the image is divided into
a number of adjacent regions [18]. After conducting several
experiments and testing a range of smaller and greater region
sizes, the best equal error ratio performance was obtained when
dividing the image into four equal vertical blocks and three
equal horizontal blocks which overlapped by 60% [19].
CONCLUSION
Histograms of local binary, local directional and local
derivative patterns used for texture measures are proposed for
offline automatic signature verification. These parameters were
evaluated with different classifiers such as nearest neighbor and
SVM. SVM was evaluated with different kernels such as the
classical RBF and the histogram-oriented kernels GHI and
kernels. The results show that the kernel provides the best
results with local derivative patterns. These results improve
those reported in [14].
The SVM with kernel has proved to be more robust
against the gray level distortion when the signatures are mixed
with bank checks. With this configuration, the robustness
against gray level distortion is very similar with the basic LBP,
LDP and LDerivP features.