A Fingerprint Representation Technique based on Minutiae Quadrilateral Structure
Loading...
Date
04-10-18
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
MPUAT, Udaipur
Abstract
As the electronic banking, e-commerce and smartcard are widely used in our daily lives
and the need of increased security requirement in various applications ranging from
international border crossing to security information in the databases, automatic
recognition of persons has become a very important topic. A reliable identity
management system is needed in wide range of applications which include passports,
cellular telephones, ATM, and driver licenses. Traditional methods of person
identification are unable to satisfy the security requirements of electronically connected
society because PINs may be forgotten or guessed by an imposter and the tokens may be
lost or stolen.
Biometric systems are utilized broadly in many applications to recognize persons reliably.
Biometric recognition automatically recognises a person using the physiological and
behavioural characteristics. Biometric recognition provides a natural and reliable solution
for the problem of identity recognition in identity management systems. There are several
biometric modalities developed such as iris, face, signature, and palmprint which are used
in the current governmental and civil applications. Among all the biometric modalities,
fingerprint is a popular biometric trait which is used extensively in several applications
for person authentication due to its ease of capture, permanence distinctiveness.
As the use of biometric data is increasing in various applications, the concerns about the
security and privacy of biometric data is also increasing. A biometric is permanently
associated with a human; if this biometric data is compromised it cannot be revoked or
replaced. Therefore, biometric template protection schemes are required to secure the
original biometric data. Fingerprint samples exhibits intra-class variations like rotation,
translation and non linear deformation at each fingerprint image acquisition. Due to large
intra-class variations at each fingerprint image acquisition, fingerprint template protection
is not an easy task. One solution to reduce intra-class variation is to use the local
information provided by the fingerprint minutiae points such as local distance between
two points and local angles which are invariant with respect to fingerprint translation and
rotation.
This thesis is focused on proposing a new fingerprint representation using the information
provided by local quadrilateral structure to obtain a fixed length and alignment free
74
representation. We propose a fingerprint representation using the local neighbourhood
information of a minutia point. We construct quadrilateral structure using the nearest
neighbours of each minutia point and extract invariant features from these quadrilaterals.
The extracted features are quantised and mapped to a predefined 3D array. By
sequentially visiting the cells of 3D array, a fixed length 1D bit string is generated. The
bit string is applied with DFT to generate a complex vector and the complex vector is
multiplied with a random matrix generated by a user’s key, to generate final cancellable
template. The proposed method FS_QUAD is tested on FVC 2002 databases. We
compared our results with the two existing methods FS_INCIR and FS_AVGLO and
results show satisfactory performance.
Description
Fingerprint, Representation Technique based Minutiae, Quadrilateral Structure
Keywords
null
Citation
Menaria, L. AND jAIN , K