A Bayesian Approach for Modeling Sensor Influence on Quality, Liveness and Match Score Values in Fingerprint Verification
1. A Bayesian Approach for Modeling Sensor
Influence on Quality, Liveness and Match Score
Values in Fingerprint Verification
Ajita Rattani1, Norman Poh2 and Arun Ross1
1
Dept. of Computer Science and Eng., Michigan State University, USA
http://www.cse.msu.edu/~rossarun/i-probe/
2
Dept. of Computing, University of Surrey, UK
2. Abstract
• Recently a number of studies in fingerprint verification have
combined match scores with quality and liveness measure in order
to thwart spoof attacks.
• However, these approaches do not explicitly account for the
influence of the sensor on these variables.
• In this work, we propose a graphical model that account for the
impact of the sensor on match scores, quality and liveness measures.
• Effectiveness of the proposed model has been assessed on the
LivDet 2011 fingerprint database using Biometrika and Italdata
sensors.
4. Spoof Attacks
• A spoofing attack occurs when an adversary replicate the
biometric trait of another individual in order to circumvent
the system.
• These attacks pose the most severe threat to biometric
systems as
– they can be easily performed using simple techniques
and commonly available materials and,
– do not require any knowledge of internal functioning
of the system.
4
6. Fingerprints and Spoof Attacks
•
Studies have shown that fake fingerprints can be easily fabricated using
commonly available materials like silicone, latex etc., [Matsumoto et al.,
2002, Yambay et al., 2012].
•
Fingerprint liveness detection algorithms have been proposed as a countermeasure against spoof attacks.
Matsumoto et al., Impact of artificial gummy fingers on fingerprint systems, Opt. Sec.
Counterfait Deterrence Techniq. IV 4677, 275-289.
Yambay et al., Livdet2011 – fingerprint liveness detection competition, ICB, 2012
6
8. Fingerprint Liveness Detectors
• They aim to discriminate live biometric samples from the
spoofed (fake) artefact.
• The algorithms for fingerprint liveness detection examine the
textural, anatomical and/or physiological attributes of the
finger.
• The output of these liveness detection algorithms is a singlevalued numerical entity referred to as liveness measure.
8
10. Liveness Detector and Biometric System
• Liveness detection algorithms are not designed to
operate in isolation; rather, they have to be
integrated with the overall fingerprint recognition
system.
• Accordingly, recent studies have combined match
scores generated by a fingerprint matcher with
liveness values, as well as image quality, in order
to render a decision on the recognition process.
11. State of the Art
Reference
Variables
combined
Scheme
Database
Marasco et
al., BTAS,
2012
fingerprint
match scores +
liveness values
Bayesian
Belief Network
LivDet 2009
Chingovska
et al.,
CVPRW,
2013
face match
scores +
liveness values
Logistic
Regression
Replay attack,
2011
Rattani et al.,
ICB, 2013
fingerprint
match scores +
quality +
liveness values
Density-based
fusion
framework
LivDet 2011
12. Problem Statement
• However, in the aforementioned schemes, the influence of
the sensor on the 3 variables - match scores, liveness
values and quality has not been considered.
• Such a consideration is essential for several reasons:
(a) the quality of an image is impacted by the sensor used;
(b) most liveness measures are training-based and are
impacted by the sensor that was used to collect live and spoof
data;
(c) understanding sensor influence, can help in facilitating
sensor interoperability for fingerprint matchers and liveness
detectors.
13.
14. Contributions
• Development of a graphical model for fusing match
scores, liveness measures and quality values while
accounting for sensor influence;
• Implementation of the proposed model using a Gaussian
Mixture Model (GMM) based Bayesian classifier;
• Evaluation of the proposed model using fingerprint data
from two different sensors in the LivDet 2011 fingerprint
database.
15. Notations
• y = match score between pair of fingerprint
images.
• q = quality of a fingerprint image
• l = liveness measure of a fingerprint image
• k = class label i.e., {C, I, S}
• d = fingerprint acquisition sensor
17. Contd..,
k
•
d
y
l
•
q
c) Proposed Model C
Conjectures:
1. Match scores, quality and liveness measures are sensor dependent
(d
{y, q, l})
2. Further their exist no significant correlation between quality (q) and liveness
measures (l).
24. Conclusions
• Recent studies have addressed the security of
fingerprint verification system by combining
match scores with quality and liveness measure.
• We advance the state of the art by modeling
sensor influence on them.
• This is realized through a graphical model that
account for the impact of sensor on liveness,
quality and match score values
25. Contd..,
• Experimental investigations on the LivDet11 fingerprint
database indicate that
– Existing training-based fusion framework cannot
operate effectively in a multi-sensor scenario
– The proposed graphical model can effectively operate
in a multi-sensor environment
• The effectiveness of the proposed model can be further
enhanced by improving the sensitivity of the underlying
liveness detector
