Biometric Technologies and Applications
Biometric Technologies and Applications
net/publication/221173670
CITATIONS READS
24 3,170
2 authors:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Patrick Shen-Pei Wang on 01 June 2014.
recognition accuracy, total cost of ownership, acquisition teraction which enable input of handwriting and signatures
and processing speed, intrinsic and system security, privacy [8,9].
and legal requirements, and user acceptance.
2.5 Faces
2.2 Multimodal biometrics
Face recognition systems detect patterns, shapes, and shad-
Through rapid advancements and developments of past ows in the face, perform feature extraction and recognition
years, there have been tremendous achievements have been of facial identity. In the broader view, it include all types of
reach in improvement of reliability of recognition. For facial processing such as tracking, detection, analysis and
example, there have been developed character recognition synthesis.
methods that can reach as high as almost 99.99% accuracy The most popular approach is based on eigenfaces,
rate [3]. However, real-life environments pose significant that represent the differences between the face under recog-
difficulties on biometric-based recognition systems. To in- nition and the enrolled ones in the database. The principle-
crease overall reliability, the contemporary biometric sys- component analysis using higher-order statistics is the un-
tems measure multiple physiological or behavioral traits. derlying mathematics for this facial pattern recognition.
This approach is called multimodal biometrics (see Hand- Many biometric systems are confused when identify-
book of Multibiometrics [5]. ing the same person smiling, aged, with various accessories
The most often multi-biometric data, employed in the (moustache, glasses), and/or in badly lit conditions. For ro-
biometric systems, include iris and retina of the eye, fin- bustness of recognition, advanced techniques such as mor-
gerprint, geometry and palmprint of the hand, and also face phable models and expression-invariant face representation
and ears. For example, face geometry is a highly dynamic methods. On the other hand, facial recognition tools can be
but reach topological structure (smile, lip, brow, and eye improved by training on a set of synthetic facial expressions
movements). Combining facial images with more static and appearance/environment variations generated from real
biometric such as fingerprint, is an example of the concept facial images [10,11].
of multibiometrics. The most advance on-going research in this area is
The multimodal human recognition is deployed in devoted to understanding of how humans can routinely per-
physical access security systems and other areas such as form robust face recognition, in order to improve machine
banking systems [1,5,6]. recognition of human faces. This research is relevant to
computer vision paradigm. A comprehensive references to
2.3 Fingerprints the current state-of-the-art approaches to face processing
can be found in [12].
Fingerprint is, perhaps, the oldest type of biometrics, used
in ancient world. The most popular utilization example of 2.6 Iris and retina
fingerprints is forensic investigations [7]. Today’s finger-
print readers is the most developed type of biometric sen- Iris recognition systems scan the surface of the iris to com-
sors. pare patterns. It was Daugman who first introduced so-
called Daugman code to represent an iris pattern. The
2.4 Signatures iris biometrics is considered to be the most reliable one,
as the plots for comparing the hamming distances for 9.1
Current interest in signature analysis is motivated by the million iris comparisons to the Beta-binomial distribution
development of improved devices for human-computer in- showed that the data fit the distribution remarkably well
[11]. Retina recognition systems scan the surface of the as passive surveillance, contact, cooperative, non-
retina and compare nerve patterns, blood vessels and such cooperative), environmental factors (light, smog, tem-
features. Various model of the iris, retina, and eye have perature), appearance (aging, camouflage).
been used to improve recognition, and can be found in [14-
16]. Therefore, synthetic biometric data plays an impor-
tant role in enhancing the security of biometric systems.
Traditionally, security strategies (security levels, tools, etc.)
2.7 Gait biometrics are designed based on assumptions about a hypothetical
robber or forger. Properly created artificial biometric data
Gait recognition is defined as the identification of a person provides for detailed and controlled modeling of a wide
through the pattern produced by walking [6,17]. The po- range of training skills, strategies and tactics, thus enabling
tential of gait as a biometric was encouraged by the consid- a better approach to enhancing the system’s performance.
erable amount of evidence available, especially in biome- This study aims to develop new approaches for the detec-
chanics literature [19,20]. A unique advantage of gait as tion of attacks on security systems.
biometrics is that it offers potential for recognition at a dis- The tools for analysis and recognition (direct biomet-
tance or at low resolution, when other biometrics might not rics) of human characteristics have their analogs in the syn-
be perceivable. As gait is behavioural biometrics there is thesis (Figure 1.
much potential for within-subject variation [21]. This in-
cludes footwear, clothing and apparel. Recognition can RECOGNITION AND SYNTHETIC
be based on the (static) human shape as well as on move- IDENTIFICATION OF BIOMETRIC DATA
BIOMETRIC DATA GENERATION
ment, suggesting a richer recognition cue. Model-based
techniques use the shape and dynamics of gait to guide the BEHAVIORAL
PHYSICAL
extraction of a feature vector. MODELS
DISTANCE
2.8 Other biometrics NEAR DISTANCE
CONTACT
MODELS
A variety of biometrics such as ear geometry [22], odour,
electrocardiogram, keystroke dynamics [1,5] and other KNOWLEDGE
DOMAIN
other physiological and behavioral characteristics are be- SIGNAL DOMAIN
ing investigated to be deployed for human authentication. REPRESENTATION
NON-
CORRELATED
3 Inverse Biometrics CORRELATED
DATA