Professional Documents
Culture Documents
Pattern Recognition
journal homepage: www.elsevier.com/locate/pr
art ic l e i nf o a b s t r a c t
Article history: Identity management through biometrics offer potential advantages over knowledge and possession
Received 9 February 2013 based methods. A wide variety of biometric modalities have been tested so far but several factors
Received in revised form paralyze the accuracy of mono-modal biometric systems. Usually, the analysis of multiple modalities
14 November 2013
offers better accuracy. An extensive review of biometric technology is presented here. Besides the mono-
Accepted 27 January 2014
modal systems, the article also discusses multi-modal biometric systems along with their architecture
Available online 5 February 2014
and information fusion levels. The paper along with the exemplary evidences highlights the potential for
Keywords: biometric technology, market value and prospects.
Biometrics & 2014 Elsevier Ltd. All rights reserved.
Fingerprint
Face
Iris
Retina
Behavioral biometrics
Gait
Voice
Soft biometrics
Multi-modal biometrics
http://dx.doi.org/10.1016/j.patcog.2014.01.016
0031-3203 & 2014 Elsevier Ltd. All rights reserved.
2674 J.A. Unar et al. / Pattern Recognition 47 (2014) 2673–2688
to the availability of resources and imposition of real constraints in light on the limitations of mono-modal biometric systems followed
terms of data collection and guarantee to achieve high accuracy, by an overview of multi-modal biometric systems. Finally, Section 5
(vi) acceptability: willingness of population to submit that attribute to presents the recent market trend and thorough analysis with
recognition system and (vii) circumvention: prone to imitation or examples towards adoption of biometric solutions followed by
mimicry in case of fraudulent attacks against the recognition system concluding remarks.
[1]. Based on the criteria, several distinctive human characteristics are
identified and tested. Instead of the broad categories (physiological,
behavioral and soft attributes), for convenience the physiological 2. Biometric systems
modalities can be further sub-divided into different sub-categories
according to their respective position in human body such as (i) hand A biometric system is a pattern recognition system which
region attributes, (ii) facial region attributes, (iii) ocular and periocular matches the salient or discriminatory features of acquired image
region attributes, (iv) behavioral attributes, and (v) medico-chemical (probe image) with the features of pre-stored images (gallery
attributes. The sub-division is illustrated in Fig. 1. image). For doing so, every biometric system comprises of (i)
The article is written with an intention to provide an in-depth Image acquisition module: this acquires the image of a biometric
overview of biometric technology to the readers. In contrast to trait and submits it to the system for further processing, (ii)
several reviews on the subject, this work provides additional Feature extraction module: processes the acquired image thereby
information which is lacking in other review articles, such as extracting the salient or discriminatory features, (iii) Matcher
module: matches the extracted features of probe image with those
i) The paper provides details about each biometric modality of gallery image to obtain a match score whereas, an embedded
along with prominent recognition techniques and public data decision making module verifies or rejects the claimed identity
sets available to the researchers. based on the match score and (iv) Database module: contains the
ii) The article contains a rough quantitative analysis of individual digital representation of previously acquired samples very often
techniques in terms of accuracy. termed as templates. Fig. 2 illustrates the composition of a typical
iii) The paper discusses multimodal biometrics in detail which has biometric system.
not been well covered in other review articles on the subject.
iv) The paper contains current market statistics along with future 2.1. Modes of operation
prospects for biometric based identification and authentication.
A biometric system operates in one of the following modes:
Rest of the article comprises of four sections, whereas, Section 2
provides basic concepts consisting composition, classification, opera- 2.1.1. Verification
tional scenarios and performance issues related to biometric systems. This is very often referred as positive recognition. The user after
Section 3 discusses biometric modalities as per aforementioned submitting the biometric signature to the system claims a parti-
classification along with advantages, disadvantages, matching tech- cular identity through a PIN, login name, etc. In response, the
niques and available data sources for researchers. Section 4 throws recognition system validates or voids user's claim by making a 1:1
J.A. Unar et al. / Pattern Recognition 47 (2014) 2673–2688 2675
(one-to-one) comparison between the submitted biometric signa- 2.2.2.1. False match rate (FMR) or false accept rate (FAR). This is the
ture and enrolled biometric signature associated with that parti- percentage measure of invalid matches, also referred as Type – I
cular identity. The applications include computer logins, ATMs, error. It is the number of times (in percentage) the system
e-commerce, access control and user authentication on mobile recognizes unauthorized users as genuine users. For a robust
devices [2]. biometric system, this error must be as low as possible.
2.1.2. Identification 2.2.2.2. False non – match rate (FNMR) or false reject rate (FRR). This
In this mode, the system tries to recognize the user by is the measure of times (in percentage) the system recognizes an
comparing the submitted biometric signature to all the enrolled authorized user as an impostor. It is the percentage measure of
signatures in the database by making 1:N (one-to-many) compar- rejecting valid inputs, also referred as Type – II error. From user's
isons without specific identity claim from the user. Identification is convenience point of view, this number must be low as much as
a crucial component in negative recognition where the user denies possible.
of holding a particular identity. In fact, negative recognition
prevents an individual from using multiple identities. The applica- 2.2.2.3. Equal error rate (EER). It refers to the point where false
tions include issuance of ID cards, passports, driving licenses, accept rate (FAR) and false reject rate (FRR) equals each other. This
border crossing and welfare disbursements [1,2]. is obtained through the Receiver Operating Characteristic (ROC)
which is a plot of FAR against the FRR and a trade-off between the
false accepts and rejects rates. This trade-off refers to the function
2.1.3. Screening
of the threshold value. If design of the system tolerates the input
This is an extension to identification where the biometric
variations and noise, the FAR increases and vice versa. Indeed,
system assures that a particular individual does not belong to a
equal error rate (ERR) is the measure of accuracy of a biometric
watch list of identities by performing 1:N (one-to-many) compar-
system. Nevertheless, our viewpoint considers sample acquisition
isons throughout the database. Example applications include air-
errors more important than the performance errors due to the fact
port security, surveillance activities, public place and public events
that they have a significant impact on the accuracy of biometric
security etc. [4]
systems in terms that good quality images provide better accuracy.
Moreover, the accuracy can be increased by decreasing such errors.
2.2. Performance measurements This can be achieved through the selection of appropriate imaging
sensors along with the development of robust artifact removal
The performance or accuracy of a biometric system is data algorithms.
dependent usually influenced by environmental and performance
factors. The environmental factors include temperature, humidity
and illumination conditions around the system, whereas, the 3. Biometric modalities
performance factors include capturing good quality images, com-
position of target user population, time interval between the This section provides an in-depth overview of different biometric
enrollment and verification phases and robustness of recognition modalities as per the classification provided in Fig. 1 above.
algorithms [2]. However, being a pattern recognition system, the
accuracy of a biometric system is usually measured in terms of 3.1. Hand region modalities
sample acquisition and performance errors as described below
[1,2]: Human hand contains rich texture information that provided
the foundations for early recognition systems based on finger-
prints. In addition to fingerprints, several attributes have been
2.2.1. Sample acquisition errors
identified and tested such as palmprint, hand geometry, finger
Environmental conditions surrounding the system give rise to
knuckle print, finger nail bed and hand vein pattern. However, all
acquisition errors. For instance, (i) Failure to enroll error: which is
the hand based attributes are an extension to fingerprint technol-
the percentage measure of times when the system rejects the
ogy. A fingerprint recognition system uses the texture of ridges
provided samples due to poor quality or noisy images and (ii)
and valleys present on the fingertips [2,5] whereby the ridge
Failure to capture error: that refers to the inability of imaging
endings (minutiae points) perform the recognition task and the
sensors to acquire the valid biometric sample. Very often, such
ridge flow classifies the fingerprints into one of the five categories
error prevails amongst the systems requiring impression based
such as arch, tented arch, left loop, right loop and whorl [6].
imaging of the modalities automatically. Normally, this error
An automated fingerprint recognition systems uses level 1, level
occurs due to dirty surface of the imaging sensors.
2 or level 3 features (refer Table 1). However, most of the
commercially available systems use level 2 features or minutiae
2.2.2. Performance errors points for recognition whereas level 3 features offer robust
These errors are used to measure the accuracy of biometric recognition against poor quality or latent fingerprint images [7].
systems in real environment. A brief description is given below: An automated palmprint recognition system considers the texture
2676 J.A. Unar et al. / Pattern Recognition 47 (2014) 2673–2688
Table 1
Summary of hand region modalities.
Fingerprint Ridge flow, ridge pattern, singular points, ridge skeleton, ridge k-nearest neighbor [24], FFT [25], GA [26,27], DTW [28], ACO CASIA, Sfinge, FVC
flow, ridge ending, ridge contours, ridge kernel, orientation [29], Graph matching [30,31], Neural networks [32–34], SVM 2004 DB1, FVC
field, island, spur, crossover, learned feature, sweat pores, dots [34,35], HMMs [36], Bayesian [37], Adaboost [6], Fuzzy logic 2006, NUERO
& incipient ridges [38,39], Corner detection [40], Decision trees [41] technology
Palmprint Ridges, singular points, minutiae points, principal lines, Edge maps [42–46], PCA [47,48], LDA [48–50], ICA [48,51], DCT CASIA, PolyU, IIT
wrinkles, palm texture, mean, variance, moments, center of [52], Zernike moments [53], Hu invariant moments [54], Mean Delhi
gravity & density, spatial dispersivity, L1-norm energy [55,56], HMM [21], Directional line detector [55], Wavelets
[57–60], LBP [61], SVM [62]
Hand geometry Length & width of fingers, aspect ratio of finger or palm, Correlation co-efficient [63], Absolute distance [64,65], Bosphorus
length, thickness & area of hand, hand contour, hand Mahalanobis distance [66,67], Euclidean distance [68], Bayes
coordinates and angles, Zernike moments, skin folds and classifier [69], Mean alignment error [70], Hamming distance
crease pattern [22], GMM [22,71], L1 cosine distance [65], SVM [72]
Hand vein Vein bifurcation & ending Adaptive thresholding [73], Morphological gradient operator —
pattern [74], PCA [75], LDA [76], FFT [72], Feature point distance [73],
Vein triangulation and shape [20], SVM [77], SIFT [78], LBP
[79], Curvelet transform [80]
Finger knuckle Texture of lines, orientation, magnitude Localized Radon Transform [81], PCA [81], Gabor filters [82], PolyU (FKP)
print BLPOC [17], LDA [82], OE-SIFT [83], Phase congruency [18], ICA database
[82]
Fig. 3. Hand region modalities: (a) fingerprint [6] (b) palmprint, (c) hand geometry, (d) hand vein pattern [20], (e) finger knuckle print.
consisting of principal lines and secondary creases present on the advantage of low cost imaging sensors and small template size
palmar region of human hand [8,9] thereby acquiring the low and makes hand region attributes a favorable choice for a variety of
high quality images. For matching, the recognition system extracts applications as compared to other biometric signatures. Conver-
ridges, singular points and minutiae points from high quality sely, distorted images, physical contact with imaging device, high
image, principal lines, wrinkles and palm texture from low quality user cooperation, hand diseases (arthritis) and natural imaging
image. In terms of matching algorithms (refer Table 1), the system contaminants (dead cell, scars, cuts, wet and dry skin) as well as
applies line [10], sub-space [11], local and global statistics based dirty or oily surface of imaging sensors are some of the challenges
approaches [12,13]. A hand geometry system takes into account associated with hand based biometrics [2,20,23]. Comparatively,
the length, width, aspect ratio of fingers or palm as well as length, fingerprint technology occupies the top layer amongst all other
thickness, area, skin folds and crease patterns of human hand hand region modalities due to diverse applications whereas all
[14,15]. A finger knuckle print system (although new in biometric other attributes well suit the verification tasks. However, it is
modalities) utilizes the texture of lines present on the outer part of known that the artificial gummy fingers and recognition from
the fingers near phalangeal joint [16–18] whereas the finger nail latent fingerprint images challenge the authenticity of fingerprint
bed refers to the measurement of the distance between parallel recognition systems. Nevertheless, the recent trend indicates more
dermal structures present under the finger nail [19]. A hand vein interest in vein technology as compared to other hand based
recognition system utilizes the vein bifurcations and endings modalities. This is due to the hardness of vein pattern against
beneath the skin of human hand [20]. Most of the image acquisi- forgery. Table 1 displays feature sets, promising recognition
tion methods in all the hand based modalities are impression techniques and public data sources available to each hand based
based requiring optical, thermal, silicon or ultrasonic imaging characteristic. Fig. 3 displays all the hand region modalities.
sensors [20–22]. However, some exceptions are hand geometry
systems that use a camera to acquire 3D image of human hand
[22] and hand vein recognition system that images the vein 3.2. Facial region modalities
structure beneath the skin thereby exposing hand to near infrared
light [20]. Nevertheless, fingerprinting technology occupies a Human facial region has been an interesting topic for researchers
prominent place in other hand based biometrics due to the longest due to the fact that human face is the most natural biometric trait
serving method and higher degree of uniqueness as well as high used to recognize fellow beings since centuries [2]. Although being
acceptance. And this makes it a suitable choice for verification as the most natural biometric characteristic, the non-linear structure of
well as recognition applications. On the contrary, other hand based human face makes it complex pattern recognition problem as well as
modalities suit to verification task only [2,20,23]. The potential an active area of research in computer vision applications [84]. As a
J.A. Unar et al. / Pattern Recognition 47 (2014) 2673–2688 2677
Table 2
Summary of facial region modalities.
Face Distance between eyes, mouth, side of nose, entire PCA [95,96], LDA [97], Self-organizing map & FERET, AR faces, MIT, CVL, XM2VTS, Yale
face image, corner points, contours, gender, goatee, convolutional network [98], Template Matching [99], face, Yale face B, 3D RMA, CASIA,
roundness of face, edge maps, pixel intensity, local LEMs [100], EBGM [101], DCP [102], LBP [103], CSML GavabDB
and global curvatures [104], SVM [105], DBN [106–108], NMF [109], SIFT
[110,111], HMM [112], HOG-EBGM [113]
Ear Shape Size, length, width & height of helix rim, triangular Vornoi distance graphs [114], LDA [115], Force field XM2VTS, UND, UCR, USTB (Dataset 1,
fossa, anti-helix, concha, lobule, step edge magnitude, transform [116,117], GA [118], PCA [119,120], Active Dataset 2, Dataset 3, Dataset 4), WPUT-
color, curvature, contours, edge information, shape shape model [121], NMF [122], Gabor filters [123–125], DB, IIT Delhi, IIT Kanpur, ScFace, YSU,
indices, registered color, range image pair ICA [126], Wavelets [127], SIFT [128,129], SURF [130], NCKU, UBEAR
LBP [131–138], Moment invariants [133], SVM [134],
ICP [135], Mesh-PCA [136], Local surface patch [137]
Tongue print Width, thickness, curvature of tongue contour, cracks, 2D Gabor filter [93,139] —
texture
result, the research community invested considerable efforts in the suggested the utilization of 3D features instead of 2D features
development of robust automated face recognition systems. [90,94]. However, due to the challenges stated above as well as
An automated face recognition system establishes the identity medium level of distinctiveness, facial recognition systems are very
of an individual based on the computation of 2D or 3D features much prone to recognition errors. Table 2 provides an overview of
(refer Table 2) [84,85]. We acknowledge that the 3D facial facial region modalities in terms of feature sets, prominent recogni-
recognition is the new trend proposed to rectify the basic tion techniques and public data sources. All the facial region
problems associated with 2D facial recognition such as image modalities have been presented in Fig. 4.
degradation, poor alignment and movement of head and facial
expressions. However, these systems cannot guarantee reliable 3.3. Ocular region modalities
identification in presence of artifacts such as application of
cosmetics and plastic surgery. Moreover, a person's face may Human ocular region attained considerable attention of research-
change or be changed over time which may have a significant ers in the past decade due to the fact that this region possesses most
impact on the accuracy of such systems. In addition, the expensive accurate, highly reliable, well protected, stable and almost impossible
imaging hardware is another factor which limits the use of such to forge biometric signatures, for instance, retina, iris and sclera vein
systems. In order to develop robust face recognition systems, the pattern. A retinal identification system takes into account the unique
research community proposed the idea of human recognition and invariant structure of blood veins present on human retina to
based on facial thermograph which shows the heat radiation establish the identity [140,141]. In fact, a retinal scan system estab-
patterns of human face due to the presence of vascular structure lishes the identity by examining either the landmarks (position and
beneath the human skin [86]. Few studies [87,88] claimed to bifurcations of blood vessels) [142] or measuring the area of
achieve better accuracy in terms of face detection, localization and reference (fovea, optic disk) [143]. An automated iris recognition
segmentation against the facial images acquired in visible light. system uses the unique and stable texture consisting of crypts,
Besides human face, human ear, teeth and tongue also qualify the furrows, corona and freckles of human iris [144,145]. Consecutively,
criteria for being the biometric modalities. An automated ear sclera based identification system utilizes the vascular pattern of
recognition system takes into account the shape of human ear to blood vessels present in the sclera region of human eye [146].
establish the identity based on 2D or 3D features (refer Table 2) However, a few studies talked about the authenticity of sclera
[89,90]. This includes different imaging techniques such as taking the vasculature as a biometric characteristic [146,147]. Consequently,
photograph of ear, pushing the ear against a flat surface to get an ear the conformity requires large scale studies. Although the established
mark, taking the thermograph of ear and measuring the acoustic traits (retina, iris) belong to the same region but utilize different
transfer function of a sound wave projected on human ear [90–92]. imaging procedures. For example, retinal vascular pattern that can
Identification based on tongue print is relatively new in biometric only be imaged by exposing the human eye to infra-red light whereas
modalities. The potential advantages associated with tongue print iris texture can be acquired either by illuminating human eye with
include protection from external environment and its squirm which near infra-red light or in visible wavelength light. Comparatively, iris
proves the liveliness of the subject. In order to establish identity, a occupies a prominent position in ocular biometric signatures due to
tongue print identification system uses the geometrical features such noninvasive image acquisition and clear visibility at a distance [144]
as width, thickness and curvature of tongue contour, crack features whereas the intrusive imaging procedure of retinal scan systems
and texture features of human tongue [93]. The conformity of tongue restrict their use to military purposes only instead of large scale
print as biometric attribute requires large scale studies. In terms of deployment in public places. Besides imaging iris texture in NIR light,
dental biometrics, due to no public deployment of automated dental the ocular biometric researchers attempted to perform recognition
biometric systems and extensive usage of the same in postmortem on iris texture acquired in visible light, at a distance and on the move
and forensic applications, we consider dental biometrics beyond the [148–152]. However, the development of synthetic iris images from
scope of this paper. The comparative analysis of all the facial region stored iris codes [153] opened a new debate in iris template
modalities reveal almost the common advantages such as, non- protection. Nevertheless, higher uniqueness, stability over time,
intrusive and contactless image acquisition thereby resulting in the highly secure and hardness to forge makes ocular biometrics a
higher public acceptance. Parts from advantages, the challenges suitable choice for verification as well as recognition tasks. Conver-
include sensitiveness to illumination conditions, pose variations, sely, sensitivity to illumination, higher degree of cooperation from
ageing conditions and poor recognition accuracy in presence of the subjects, high cost of imaging sensors and reflections from
partial or full occlusions. In order to overcome these challenges ambient light sources are some of the challenges associated with
and to achieve better recognition accuracy, the research community ocular biometrics. A comprehensive overview in terms of feature sets,
2678 J.A. Unar et al. / Pattern Recognition 47 (2014) 2673–2688
Fig. 4. Facial region modalities: (a) face, (b) facial thermograph [88], (c) ear shape, (d) tongue print.
Table 3
Summary of ocular region modalities.
Iris Color, shape and iris texture 2D Gabor filters [144,154,155], Wavelets [156–159], LoG filter [160], DCT [161], Ordinal CASIA, UBIRIS, WVU,
(crypts, furrows, corona, freckles) measures [162,163], ICA [164], PCA [165,166], LDA [167], LBP [168,169], WCPH [170], MMU1, MMU2, IIT Delhi
Neural Networks [171,172], SVM [173], SIFT [168,174], Adaboost [175], Texton histogram [162],
Weight map [176], Directionlets [177], GA [178]
Retina Vein bifurcations, area of optic Principal bifurcation orientation (PBO) [140], DB-ICP [179], Gabor wavelet [180], SIFT [181], VARIA
disk or fovea SFR [182]
Sclera Vein bifurcations ANN [146], SURF [183], Direct correlation [183], Minutiae matching [183] —
prominent recognition algorithms and public data sources is pre- dynamics system establishes the human identity through analyzing
sented in Table 3. Fig. 5. illustrates all the ocular region modalities. the typing rhythms of an individual [193]. A speaker verification
system uses the vocal characteristics of an individual to establish
3.4. Medico-chemical biometrics the identity either by imposing the fixed vocabulary constraints
(text dependent) or in a dynamic way i.e. without imposition of
Parts from the classical classification, we classify body odor, vocabulary constraints on the individuals [194,195]. A signature
Deoxyribonucleic Acid (DNA), heart sound and Electrocardiogram recognition system establishes the identity based on the analysis of
(ECG) as medico-chemical biometrics due to the fact that the an individual's signature characteristics produced to the system
recognition process requires specialized medical/chemical sensors either in static or dynamic modes [196,197]. Human identification
for data acquisition. Nevertheless, the authenticity of DNA is well through gait requires the analysis of human motion features such as
established whereas identification based on the analysis of heart the shape and dynamics information [198]. Recently, user authenti-
sound and ECG signals need conformity through large scale cation based on gait analysis has been applied on mobile devices
studies. Consequently, DNA being a well-established and most which in turn produced another research arena in gait biometrics.
accurate biometric signature dominates other group members. The potential approaches utilize human motion features acquired
A one dimensional DNA code is usually extracted from small through embedded accelerometers of mobile devices to recognize
varying region of entire code very often termed as short tandem their legal users. Several approaches have been developed so far
by acquiring the samples from blood, hair, ear wax, dental floss thereby the authors claimed to achieve better recognition accuracy
and fingernail clippings [184,185]. Body odor being the composi- [199–203]. A closer analysis reveals that behavioral attributes do
tion of different organic compounds can be acquired through an not provide sufficient discriminatory information because these
array of chemical sensors sensitive to those compounds [1,186]. attributes are usually affected by human emotional states, health
However, such a system to recognize humans based on their body conditions, dietary habits and aging conditions. These said factors
odor is not realized yet, albeit some commercial products are already restrict their usage to verification only. For example; the back-
available in the market known as e-noses which can recognize the ground noise, poor quality of acquisition devices as well as health
particular odor such as gases and vapors [187]. In this connection, a conditions may have severe impact on the accuracy of speaker
British company Mastiff electronic systems and DARPA claimed to verification system. Additionally, the accuracy of signature verifica-
achieve specific milestones in human recognition based on body tion drops significantly due to muscular illness of subjects. Further-
odor [188]. Furthermore, some studies based on the analysis of more, a gait recognition system may provide high false rates due to
cepstral coefficients of heart sound and analysis of fiducial points walking conditions and walking surface dynamics. However, human
from ECG data displayed promising results towards human identifi- motion analysis provides the potential advantage of human identi-
cation [189–192]. Conclusively, intrusive data acquisition procedure, fication in the absence of other biometric characteristics. On the
privacy issues, physical contact with sensors, requirement of skilled contrary, identification based on behavioral biometrics is not
operators, full cooperation from the subjects and dependence over beyond doubt since mimicking human behavior is easy for an
human medical and emotional states are some of the restrictions experienced and skilled impostor. Table 4 provides behavioral
towards large scale deployment of such systems. biometrics information in terms of feature sets, promising recogni-
tion algorithms and public data sources available to researchers. The
3.5. Behavioral biometrics behavioral modalities have been displayed in Fig. 6.
Nevertheless, in terms of accuracy Table 5 presents a rough
Behavioral attributes establish identity based on the analysis comparison between different biometric recognition techniques in
of the way humans do the things. For example, typing styles terms of average accuracy as reported in the literature (refer
(keystroke dynamics), vocal characteristics (voice), signature Tables 1–4). However, an accurate comparison is very difficult
dynamics and the way humans walk (gait). In fact, a keystroke due to variance in datasets.
J.A. Unar et al. / Pattern Recognition 47 (2014) 2673–2688 2679
Fig. 5. Ocular region modalities: (a) iris, (b) retina [142], (c) sclera and vasculature.
Table 4
Summary of behavioral modalities.
Voice Spectrum, Glottal pulse features, pitch, energy, LPCC [204], MFCC [205], VQ [206–208], HMM [209], GMM [210,211], TIMIT, TIDIGIT,
duration, rhythm, temporal features, phones, idiolects, DTW [212,213], ANN [214], ICA [215], SVM [216,217], ACO [195,218] AURORA, YOHO
semantics, accent, pronunciation,
Keystroke Keystroke duration, hold time, keystroke latency, speed, Nearest neighbor [219], SVM [220,221], HMM [222], Manhattan MySQL, GREYC
dynamics pressure, digraph latency, distance [223], GMM [224], Euclidean distance [225], ANN [226–228],
Random forests [229], Fuzzy logic [230], GA [231], Mean & Standard
deviation [232], Bayesian & FLD [233], Time interval histogram [234]
Gait Full subject silhouette, strides, length, cadence, speed, PCA [235], LDA [236], K-nearest neighbor [237], SVM [238], DTW CASIA, CMU Mobo,
singularity of silhouette shape [239,240], HMM [241–244], VHT [245,246], Radon transform [247], UMD, USF
LPP [248,249], DLA [250], Wavelets [251]
Signature Signature shape, Pen position, pressure, pen direction, DTW [252,253], HMM [254,255], ANN [256,257], Bayesian [258], SVM MCYT, SignatureDB,
acceleration, length of strokes, tangential acceleration, [259,260], Fuzzy [261],EPs [262], PCA [263], Regional correlation SUSIG, GAVAB offline
curvature radius, azimuth [264], NCA/PCA [265], DTW-VQ [266] signature database
3.6. Soft biometrics burns cannot destroy them [272]. In general, establishing a
suspect's or victim's identity through tattoos in forensic applica-
The poor recognition accuracy of biometric systems due to tions is usually done by the human experts manually [271,272].
imaging errors convinced the research community to find the ways However, the authors in [270–272] developed an automated CBIR
and means in order to improve the recognition rate or in other system for tattoo recognition and claimed to achieve good recog-
words make the systems robust against the environmental factors. nition accuracy. However, from our stand point, tattoos do not
Parts from this, many applications require the production of recogni- qualify for being a biometric modality. This is due to the fact that
tion results within minimum possible time. As a result, the tattoos negate the basic requirement of “Universality” which
researchers proposed the use of soft biometric attributes along with clearly states that the attribute must be possessed by entire/
the hard biometric modalities (face, iris etc.). However, soft biometric majority of the population [1]. Nevertheless, several studies
characteristics lack distinctiveness and permanence because these [268,269,273,274] claimed to achieve significant improvements
attributes are most common among the humans. For example; in recognition accuracy by combining soft biometrics with hard
gender, ethnicity, height, weight, skin color, eye color, hair color biometric modalities.
and SMT(scars, marks and tattoos) [268,269]. Even though soft
biometrics possess inability to provide sufficient discriminatory 3.7. Application perspectives
information to perform the recognition task, they can be used for
data categorization which ultimately improves the recognition The recent years witnessed mass deployment of biometric
time as well as accuracy thereby narrowing down the search technologies for reliable human identification in public as well
space. Furthermore, these traits can be used for identification as private sector. Nevertheless, the application environments for
either in absence or contaminated hard biometric traits. each technology are summarized in the following table Table 6.
Amongst the soft biometric modalities, the recent years wit- A closer look reveals that most of the techniques in order to
nessed considerable interest in tattoo recognition [270]. However, establish the identity require close contact of the subject with the
tattoo recognition is limited to forensic applications only. This is imaging sensor. However, face recognition and recognition based
due to the prevailing trend of using tattoos in criminal groups on gait has the potential to identify humans at a distance. As a
which in turn helps the agencies to establish gang associations, result, face recognition and gait recognition suits more the video
previous convictions and years spent in jail [271]. On the contrary, surveillance activities such as suspect identification at airports as
the general population uses the tattoos to distinguish them from well as crowd monitoring and human activity recognition at public
others as well as to display their religious beliefs. Generally, the places which is not possible with other biometric techniques such
law enforcement agencies use tattoos to establish identity of a as fingerprint, iris recognition which require dedicated arrange-
suspect/victim in absence or corrupted hard biometric evidences ments for imaging as well as sufficient user cooperation.
[271]. For the purpose, many agencies regularly collect tattoo
images and maintain the same in databases for future use. The
exemplary evidences are identification of victims of 9/11terrorist 4. Multi-modal biometrics
attacks and Asian tsunami in 2004. This has become possible due
to the physical characteristics associated with tattoos that they are Instead of the tremendous advances in biometric technology,
usually carved deep on the human skin that even the severe skin the recognition systems based on the measurement of single
2680 J.A. Unar et al. / Pattern Recognition 47 (2014) 2673–2688
Fig. 6. Behavioral modalities: (a) voice signal [194], (b) typing rhythm [267], (c) gait, (d) signature.
Table 6
Application scenarios for biometric based identification.
Fingerprint Border control, forensics, criminal identification, access control, computer logins, e-commerce, welfare disbursements, missing children
identification, id-cards, passports, user authentication on mobile devices, time and attendance monitoring systems
Palmprint Access control, e-commerce, computer logins, user authentication on mobile devices, time and attendance monitoring systems
Hand geometry Access control, e-commerce, computer logins, user authentication on mobile devices, time and attendance monitoring systems
Hand vein pattern Access control, e-commerce, computer logins, user authentication on mobile devices
Finger knuckle print Access control, e-commerce, computer logins, time and attendance monitoring systems
Face Border control, forensics, criminal identification, access control, computer logins, e-commerce, welfare disbursements, missing children
identification, id-cards, passports, video surveillance, crowd monitoring
Ear shape Access control, e-commerce, computer logins, time and attendance monitoring systems
Tongue print Access control, e-commerce, computer logins
Iris Border control, criminal identification, access control, computer logins, e-commerce, welfare disbursements, missing children identification,
id-cards, passports, time and attendance monitoring systems
Retina Border control, criminal identification, access control in highly sensitive installations
Sclera Access control, e-commerce, computer logins
Voice Access control, e-commerce, computer logins, user authentication on mobile device
Keystroke dynamics Computer logins, user authentication on mobile devices
Gait Video surveillance, forensics, criminal identification from video sequences, user authentication on mobile devices
Signature Access control, e-commerce, computer logins, e-banking
Fig. 8. Fusion levels in multi-modal systems [5]: (a) feature level (b) score level (c) decision level.
The aforementioned report also predicts further revenue The forecasted revenue share for each modality produced in
growth at CAGR of 60.99% from 2009 to 2017. Besides, another [277] is illustrated in Fig. 10.
source predicted the revenue growth at a CAGR of 21.6% from An interpretation shows the dominance of fingerprinting techno-
US$ 4217 million in 2010 to US$ 11,229.3 million by the year logy perhaps due to the preliminary biometric technology as well as
2015 [278]. the low cost of imaging sensors. In comparison to the figures for the
J.A. Unar et al. / Pattern Recognition 47 (2014) 2673–2688 2683
6. Conclusion
performance evaluation indicators. This paper also provides an in- [16] Z. Lin, Z. Lei, D. Zhang, Finger-knuckle-print: a new biometric identifier,
depth overview followed by detailed discussion over different in: Proceedings of the 16th IEEE International Conference on Image Proces-
sing (ICIP), 2009, pp. 1981–1984.
biometric attributes along with their potential advantages and [17] S. Aoyama, K. Ito, T. Aoki, Finger-knuckle-print recognition using BLPOC-
challenges. However, the limitations of mono-modal biometric based local block matching, in: Proceedings of the First Asian Conference on
systems forced the researchers to search for more options. Conse- Pattern Recognition (ACPR), 2011, pp. 525–529.
[18] L. Zhang, L. Zhang, D. Zhang, Z. Guo, Phase congruency induced local features
quently, the multi-modal biometric systems evolved. Multi- for finger-knuckle-print recognition, Pattern Recognit. 45 (2012) 2522–2531.
modality based systems offer better accuracy as compared to [19] D. Bhattacharyya, R. Ranjan, F. Alisherov, M. Choi, Biometric authentication:
mono-modal systems. However, in terms of multiple attributes, a review, Int. J. u- and e-Service, Sci. Technol. 2 (2009) 13–28.
[20] A. Kumar, K.V. Prathyusha, Personal authentication using hand vein trian-
careful selection is the key point to success since selecting the
gulation and knuckle shape, IEEE Trans. Image Process. 18 (2009) 2127–2136.
modalities belonging to one region may not be a good choice [21] X. Wu, K. Wang, D. Zhang, HMMs based palmprint identification, Biom.
because accidental loss of that organ will result in user's inability Authentication (2004) 1–11.
to submit the required signature. Hence, we are of the view that [22] R. Sanchez-Reillo, C. Sanchez-Avila, A. Gonzalez-Marcos, Biometric identifi-
cation through hand geometry measurements, IEEE Trans. Pattern Anal.
the selected multiple modalities must be from different organs. Mach.Intell. 22 (2000) 1168–1171.
Furthermore, we also stress the importance of reliable template [23] M. Deriche, Trends and challenges in mono and multi biometrics, in: First
protection strategies since the development of artificial gummy Workshops on Image Processing Theory, Tools and Applications, 2008. IPTA
2008, 2008, pp. 1–9.
fingers and synthetic irises may lead to questions regarding the [24] A.K. Jain, S. Prabhakar, L. Hong, A multichannel approach to fingerprint
reliability of the most secure biometric signatures so far. classification, IEEE Trans. Pattern Anal. Mach. Intell. 21 (1999) 348–359.
Despite the challenges, the market trend presented in the paper [25] A.P. Fitz, R.J. Green, Fingerprint classification using a hexagonal fast fourier
transform, Pattern Recognit. 29 (1996) 1587–1597.
portrays a keen interest in biometric solutions to a number of [26] M.R. Girgis, A.A. Sewisy, R.F. Mansour, A robust method for partial deformed
identity management tasks. The large deployment of such systems fingerprints verification using genetic algorithm, Expert Syst. Appl. 36 (2009)
in private as well as the stringent security environments is evident 2008–2016.
[27] X. Tan, B. Bhanu, Fingerprint matching by genetic algorithms, Pattern
from the exemplary evidences produced here. Conclusively, the
Recognit. 39 (2006) 465–477.
increasing trend of interest from government and private stake- [28] Z.M. Kovacs-Vajna, A fingerprint verification system based on triangular
holders promises a bright future to the science of biometrics. matching and dynamic time warping,, IEEE Trans. Pattern Anal. Mach. Intell.
22 (2000) 1266–1276.
[29] K. Cao, X. Yang, X. Chen, Y. Zang, J. Liang, J. Tian, A novel ant colony
optimization algorithm for large-distorted fingerprint matching, Pattern
Conflict of interest statement Recognit. 45 (2012) 151–161.
[30] D.K. Isenor, S.G. Zaky, Fingerprint identification using graph matching,
Pattern Recognit. 19 (1986) 113–122.
No conflicts. [31] D. Maio, D. Maltoni, A structural approach to fingerprint classification,
in: Proceedings of the 13th International Conference on Pattern Recognition,
1996, 1996, vol.3, pp. 578–585.
[32] M.S. Pattichis, G. Panayi, A.C. Bovik, H. Shun-Pin, Fingerprint classification
Acknowledgment using an AM–FM model, IEEE Trans. Image Process. 10 (2001) 951–954.
[33] K.A. Nagaty, Fingerprints classification using artificial neural networks:
a combined structural and statistical approach, Neural Netw. 14 (2001)
This work is funded by IPPP research grant from University of 1293–1305.
Malaya, Kuala Lumpur, Malaysia under Project No. PS036-2012A. [34] Y. Yao, G.L. Marcialis, M. Pontil, P. Frasconi, F. Roli, Combining flat and
structured representations for fingerprint classification with recursive neural
networks and support vector machines, Pattern Recognit. 36 (2003) 397–406.
[35] J. Li, W.-Y. Yau, H. Wang, Combining singular points and orientation image
References
information for fingerprint classification, Pattern Recognit. 41 (2008)
353–366.
[1] A.K. Jain, A. Ross, S. Prabhakar, An introduction to biometric recognition, IEEE [36] A. Senior, A combination fingerprint classifier, IEEE Trans. Pattern Anal.
Trans. Circuits Syst. Video Technol. 14 (2004) 4–20. Mach. Intell. 23 (2001) 1165–1174.
[2] A. Jain, A. Kumar, Biometric recognition: an overview, in: E. Mordini, [37] T. Xuejun, B. Bhanu, Y. Lin, Fingerprint classification based on learned features,
D. Tzovaras (Eds.), Second Generation Biometrics: The Ethical, Legal and IEEE Trans. Syst., Man, Cybern., Part C: Appl. Rev. 35 (2005) 287–300.
Social Context, vol. 11, Springer, Netherlands, 2012, pp. 49–79. [38] I. Iancu, N. Constantinescu, Intuitionistic fuzzy system for fingerprints
[3] S. Yan, Z. Xukai, E.Y. Du, Biometrics-based authentication: a new approach, authentication, Appl. Soft Comput. 13 (2013) 2136–2142.
in: Proceedings of 20th International Conference on Computer Communica- [39] C. Xinjian, J. Tian, Y. Xin, A new algorithm for distorted fingerprints matching
tions and Networks (ICCCN),, 2011, pp. 1–6. based on normalized fuzzy similarity measure, IEEE Trans. Image Process. 15
[4] A.K. Jain, S. Pankanti, S. Prabhakar, H. Lin, A. Ross, Biometrics: a grand (2006) 767–776.
challenge, in: Proceedings of the 17th International Conference on Pattern [40] Q. Zhang, H. Yan, Fingerprint classification based on extraction and analysis
Recognition, 2004, ICPR 2004, 2004, pp. 935–942 Vol.2. of singularities and pseudo ridges, Pattern Recognit. 37 (2004) 2233–2243.
[5] A.K. Jain, A. Ross, S. Pankanti, Biometrics: a tool for information security,, [41] J.-M. Guo, Y.-F. Liu, J.-Y. Chang, J.-D. Lee, Fingerprint classification based on
IEEE Trans. Inf. Forensicsd Secur. 1 (2006) 125–143. decision tree from singular points and orientation field, Expert Syst. Appl. 41
[6] M. Liu, Fingerprint classification based on Adaboost learning from singularity (2014) 752–764.
features,, Pattern Recognit. 43 (2010) 1062–1070. [42] X. Wu, K. Wang, D. Zhang, Fuzzy directional element energy feature (FDEEF)
[7] A.K. Jain, F. Jianjiang, Latent fingerprint matching, IEEE Trans. Pattern Anal. based palmprint identification, in: Proceedings of the 16th International
Mach. Intell. 33 (2011) 88–100. Conference on Pattern Recognition, 2002, pp. 95–98.
[8] A.W.K. Kong, D. Zhang, G. Lu, A study of identical twins' palmprints for [43] X. Wu, K. Wang, D. Zhang, Line feature extraction and matching in palmprint,
personal verification, Pattern Recognit. 39 (2006) 2149–2156. in: Proceedings of the Second International Conference on Image and
[9] A. Kong, D. Zhang, M. Kamel, A survey of palmprint recognition, Pattern Graphics, 2002, pp. 583–590.
Recognit. 42 (2009) 1408–1418. [44] W. Boles, S. Chu, Personal identification using images of the human palm,
[10] D.S. Huang, W. Jia, D. Zhang, Palmprint verification based on principal lines, in: TENCON'97. IEEE Region 10 Annual Conference. Speech and Image
Pattern Recognit. 41 (2008) 1316–1328. Technologies for Computing and Telecommunications, Proceedings of IEEE,
[11] L. Shang, D.S. Huang, J.X. Du, C.H. Zheng, Palmprint recognition using FastICA 1997, pp. 295–298.
algorithm and radial basis probabilistic neural network, Neurocomputing 69 [45] D.-S. Huang, W. Jia, D. Zhang, Palmprint verification based on principal lines,
(2006) 1782–1786. Pattern Recognit. 41 (2008) 1316–1328.
[12] J. You, W.K. Kong, D. Zhang, K.H. Cheung, On hierarchical palmprint coding [46] M.K. Leung, A.C.M. Fong, S.C. Hui, Palmprint verification for controlling
with multiple features for personal identification in large databases, IEEE access to shared computing resources, Pervasive Comput., IEEE 6 (2007)
Trans. Circuits Syst. Video Technol. 14 (2004) 234–243. 40–47.
[13] J.S. Noh, K.H. Rhee, Palmprint identification algorithm using Hu invariant [47] G. Lu, D. Zhang, K. Wang, Palmprint recognition using eigenpalms features,
moments and Otsu binarization, 2005, pp. 94–99. Pattern Recognit. Lett. 24 (2003) 1463–1467.
[14] D.L. Woodard, P.J. Flynn, Finger surface as a biometric identifier, Comput. Vis. [48] T. Connie, A.T.B. Jin, M.G.K. Ong, D.N.C. Ling, An automated palmprint
Image Underst. 100 (2005) 357–384. recognition system,, Image Vis. Comput. 23 (2005) 501–515.
[15] M. Adán, A. Adán, A.S. Vázquez, R. Torres, Biometric verification/identifica- [49] X. Wu, D. Zhang, K. Wang, Fisherpalms based palmprint recognition, Pattern
tion based on hands natural layout, Image Vis. Comput. 26 (2008) 451–465. Recognit. Lett. 24 (2003) 2829–2838.
J.A. Unar et al. / Pattern Recognition 47 (2014) 2673–2688 2685
[50] R. Chu, Z. Lei, Y. Han, R. He, S. Li, Learning Gabor magnitude features for [79] W. Yi-Ding, Y. Qing-Yu, L. Ke-Feng, Hand vein recognition based on multi-
palmprint recognition, in: Y. Yagi, et al., (Eds.), Computer vision – ACCV 2007, scale LBP and wavelet, in: Proceeings of the 2011 International Conference
vol. 4844, Springer, Berlin Heidelberg, 2007, pp. 22–31. on Wavelet Analysis and Pattern Recognition (ICWAPR), 2011, pp. 214–218.
[51] L. Shang, D.-S. Huang, J.-X. Du, Z.-K. Huang, Palmprint recognition using ICA [80] W. Shangqing, G. Xiaodong, A method for hand vein recognition based on
based on winner-take-all network and radial basis probabilistic neural Curvelet transform phase feature, in: Proceeings of the 2011 International
network, in: J. Wang, et al., (Eds.), Advances in Neural Networks - ISNN Conference on Transportation, Mechanical, and Electrical Engineering
2006, vol. 3972, Springer, Berlin Heidelberg, 2006, pp. 216–221. (TMEE), 2011, pp. 1693–1696.
[52] X.-Y. Jing, D. Zhang, A face and palmprint recognition approach based on [81] A. Kumar, C. Ravikanth, Personal authentication using finger knuckle surface,
discriminant DCT feature extraction,, IEEE Trans. Syst., Man, Cybern., Part B: IEEE Trans. Inf. Forensics Secur. 4 (2009) 98–110.
Cybern. 34 (2004) 2405–2415. [82] L. Zhang, L. Zhang, D. Zhang, H. Zhu, Online finger-knuckle-print verification
[53] Y. Li, K. Wang, D. Zhang, Palmprint recognition based on translation invariant for personal authentication, Pattern. Recognit. 43 (2010) 2560–2571.
zernike moments and modular neural network, in: J. Wang, et al., (Eds.), [83] A. Morales, C. Travieso, M. Ferrer, J. Alonso, Improved finger-knuckle-print
Advances in Neural Networks – ISNN 2005, vol. 3497, Springer, Berlin authentication based on orientation enhancement, Electron Lett 47 (2011)
Heidelberg, 2005, pp. 177–182. 380–381.
[54] J.S. Noh, K.H. Rhee, Palmprint identification algorithm using Hu invariant [84] A.F. Abate, M. Nappi, D. Riccio, G. Sabatino, 2D and 3D face recognition:
moments and Otsu binarization, in: Fourth Annual ACIS International a survey, Pattern Recognit. Lett. 28 (2007) 1885–1906.
Conference on Computer and Information Science, 2005, pp. 94–99. [85] X. Zhang, Y. Gao, Face recognition across pose: a review, Pattern. Recognit. 42
[55] C. Poon, D.M. Wong, H. Shen, Personal identification and verification: fusion (2009) 2876–2896.
of palmprint representations, in: D. Zhang, A. Jain (Eds.), Biometric Authen- [86] F. J. Prokoski, R. B. Riedel, J. S. Coffin. Identification of individuals by means of
tication, vol. 3072, Springer, Berlin Heidelberg, 2004, pp. 782–788. facial thermography. in: Proceeding of the IEEE International Carnahan
[56] A. Kumar, H.C. Shen, Palmprint identification using palmcodes, in: Proceed- Conference on Security Technology, Crime Countermeasures. Institute of
ings of the Third International Conference on Image and Graphics, 2004, Electrical and Electronics Engineers, October 1992, pp. 120–125.
pp. 258–261. [87] S.G. Kong, J. Heo, B.R. Abidi, J. Paik, M.A. Abidi, Recent advances in visual and
[57] Q. Dai, N. Bi, D. Huang, D. Zhang, F. Li, M-band wavelets application to infrared face recognition—a review,, Comput. Vis. Image Underst. 97 (2005)
palmprint recognition based on texture features, in: 2004 International 103–135.
Conference on Image Processing, 2004. ICIP'04, 2004, pp. 893–896. [88] P. Buddharaju, I.T. Pavlidis, P. Tsiamyrtzis, M. Bazakos, Physiology-based face
[58] G. Chen, T.D. Bui, A. Krzyzak, Palmprint classification using dual-tree recognition in the thermal infrared spectrum,, IEEE Trans. Pattern Anal.
complex wavelets, in: 2006 IEEE International Conference on Image Proces- Mach. Intell. 29 (2007) 613–626.
sing, 2006, pp. 2645–2648. [89] K.P. Ramesh, K.N. Rao, Pattern extraction methods for ear biometrics—a
[59] C.-C. Han, A hand-based personal authentication using a coarse-to-fine survey, in: World Congress on Nature & Biologically Inspired Computing,
strategy, Image Vis. Comput. 22 (2004) 909–918. 2009, NaBIC 2009, 2009, pp. 1657–1660.
[60] L. Zhang, Z. Guo, Z. Wang, D. Zhang, Palmprint verification using complex [90] L. Yuan, Z.c. Mu, Ear recognition based on local information fusion, Pattern
wavelet transform, in: IEEE International Conference on Image Processing, Recognit. Lett. 33 (2012) 182–190.
2007, ICIP 2007, 2007, pp. II-417–II-420. [91] A. H. Akkermans, T. A. Kevenaar, D. W., Schobben. Acoustic ear recognition
[61] X. Wang, H. Gong, H. Zhang, B. Li, Z. Zhuang, Palmprint identification using for person identification, in: Fourth IEEE Workshop on Automatic Identifica-
boosting local binary pattern, in: Proceedings of the 18th International tion Advanced Technologies, IEEE, October 2005, pp. 219–223.
Conference on Pattern Recognition, 2006, ICPR 2006, 2006, pp. 503–506. [92] R. Purkait, P. Singh, A test of individuality of human external ear pattern:
[62] X. Zhou, Y. Peng, M. Yang, Palmprint recognition using wavelet and support Its application in the field of personal identification, Forensic Sci. Int. 178
vector machines, in: Q. Yang, G. Webb (Eds.), PRICAI 2006: Trends in (2008) 112–118.
Artificial Intelligence, vol. 4099, Springer, Berlin Heidelberg, 2006, [93] D. Zhang, Z. Liu, J.-Q. Yan, Dynamic tongueprint: a novel biometric identifier,
pp. 385–393. Pattern Recognit. 43 (2010) 1071–1082.
[63] A. Kumar, D. Wong, H.C. Shen, A.K. Jain, Personal authentication using hand [94] T. Theoharis, G. Passalis, G. Toderici, I.A. Kakadiaris, Unified 3D face and ear
images, Pattern Recognit. Lett. 27 (2006) 1478–1486. recognition using wavelets on geometry images, Pattern Recognit. 41 (2008)
[64] N. Covavisaruch, P. Prateepamornkul, P. Ruchikachorn, P. Taksaphan, Personal 796–804.
verification and identification using hand geometry, ECTI Trans. Comput. Inf. [95] M. Kirby, L. Sirovich, Application of the Karhunen-Loeve procedure for the
Technol. 1 (2005) 134–139. characterization of human faces,, IEEE Trans. Pattern Anal. Mach. Intell. 12
[65] E. Yoruk, E. Konukoglu, B. Sankur, J. Darbon, Shape-based hand recognition, (1990) 103–108.
IEEE Trans. Image Process. 15 (2006) 1803–1815. [96] M. A. Turk, A. P. Pentland. Face recognition using eigenfaces, in: Proceedings
[66] A.K. Jain, A. Ross, S. Pankanti, A Prototype Hand Geometry-Based Verification CVPR'91., Computer Society Conference on Computer Vision and Pattern
System, 1999, pp. 166-171. Recognition. IEEE, June 1991. pp. 586–591.
[67] N. Pavešić, S. Ribarić, D. Ribarić, Personal authentication using hand- [97] P.N. Belhumeur, J.P. Hespanha, D.J. Kriegman, Eigenfaces vs. fisherfaces:
geometry and palmprint features—the state of the art, Hand 11 (2004) 12. Recognition using class specific linear projection, IEEE Trans. Pattern Anal.
[68] G. Amayeh, G. Bebis, A. Erol, M. Nicolescu, Peg-free hand shape verification Mach. Intell. 19 (1997) 711–720.
using high order zernike moments, 2006, p. 40. [98] S. Lawrence, C.L. Giles, A.C. Tsoi, A.D. Back, Face recognition: A convolutional
[69] M. Golfarelli, D. Maio, D. Malton, On the error-reject trade-off in biometric neural-network approach,, IEEE Trans. Neural Netw. 8 (1997) 98–113.
verification systems,, IEEE Trans. Pattern Anal. Mach. Intell. 19 (1997) 786–796. [99] R. Brunelli, T. Poggio, Face recognition: Features versus templates, IEEE Trans.
[70] A.K. Jain, N. Duta, Deformable matching of hand shapes for user verification Pattern Anal. Mach. Intell. 15 (1993) 1042–1052.
in: Proceedingsof the 1999 International Conference on Image Processing, [100] Y. Gao, M.K.H. Leung, Face recognition using line edge map, IEEE Trans.
1999, ICIP 99, 1999, pp. 857–861. Pattern Anal. Mach. Intell. 24 (2002) 764–779.
[71] A.L. Wong, P. Shi, Peg-Free Hand Geometry Recognition Using Hierarchical [101] L. Wiskott, J.M. Fellous, N. Kuiger, C. von der Malsburg, Face recognition by
Geomrtry and Shape Matching, Mach. Vis. Appl. (2002) 281–284. elastic bunch graph matching, IEEE Trans. Pattern Anal. Mach. Intell. 19
[72] A. Morales, M.A. Ferrer, F. Díaz, J. Alonso, C. Travieso, Contact-free hand (1997) 775–779.
biometric system for real environments, in: Proceedings of the 16th [102] Y. Gao, Y. Qi, Robust visual similarity retrieval in single model face databases,
European Signal Processing Conference (EUSIPCO), 2008. Pattern Recognit. 38 (2005) 1009–1020.
[73] Y. Ding, D. Zhuang, K. Wang, A study of hand vein recognition method, [103] T. Ahonen, A. Hadid, M. Pietikainen, Face description with local binary
in: 2005 IEEE International Conference Mechatronics and Automation, 2005, patterns: application to face recognition, IEEE Trans. Pattern Anal. Mach.
pp. 2106–2110. Intell. 28 (2006) 2037–2041.
[74] K.-A. Toh, H.-L. Eng, Y.-S. Choo, Y.-L. Cha, W.-Y. Yau, K.-S. Low, Identity [104] H. Nguyen, L. Bai, Cosine similarity metric learning for face verification, in:
verification through palm vein and crease texture, in: D. Zhang, A. Jain (Eds.), R. Kimmel, et al., (Eds.), Computer Vision – ACCV, vol. 6493, Springer, Berlin
Advances in Biometrics, vol. 3832, Springer, Berlin Heidelberg, 2005, Heidelberg, 2010, pp. 709–720.
pp. 546–553. [105] N. Kumar, A.C. Berg, P.N. Belhumeur, S.K. Nayar, Attribute and simile
[75] M. Heenaye-Mamode Khan, R.K. Subramanian, N. Ali MamodeKhan, Repre- classifiers for face verification, in: Proceedings of the 2009 IEEE 12th
sentation of hand dorsal vein features using a low dimensional representation International Conference on Computer Vision, 2009, pp. 365–372.
integrating Cholesky decomposition, in: Proceedings of the 2nd International [106] V. Nair, G.E. Hinton, Rectified linear units improve restricted boltzmann
Congress on Image and Signal Processing, 2009, CISP'09, 2009, pp. 1–6. machines, in: Proceedings of the 27th International Conference on Machine
[76] L. Wenjing, Q. Xinjun, L. Chenghang, Fusion of palm dorsal vein and hand Learning (ICML-10), 2010, pp. 807–814.
geometry for personal identification based on linear discriminant analysis, [107] M. Ranzato, J. Susskind, V. Mnih, G. Hinton, On deep generative models with
in: Procedings of the 2010 Fifth International Conference on Frontier of applications to recognition, in: Proceedings of the 2011 IEEE Conference on
Computer Science and Technology (FCST), 2010, pp. 532–536. Computer Vision and Pattern Recognition (CVPR), 2011, pp. 2857–2864.
[77] W. Hua-Bin, T. Liang, Z. Jian, Novel algorithm for hand vein feature extraction [108] G.B. Huang, H. Lee, E. Learned-Miller, Learning hierarchical representations
and recognition based on vertical integral projection and wavelet decomposi- for face verification with convolutional deep belief networks, in: Proceedings
tion, in: Proceedings of the 2012 2nd International Conference on Consumer of the 2012 IEEE Conference on Computer Vision and Pattern Recognition
Electronics, Communications and Networks (CECNet), 2012, pp. 1928–1931. (CVPR), 2012, pp. 2518–2525.
[78] W. Yiding, F. Yun, L. Weiping, L. Kefeng, L. Shark, M.R. Varley, Hand vein [109] H.J. Oh, K.M. Lee, S.U. Lee, Occlusion invariant face recognition using selective
recognition based on multiple keypoints sets, in: Proceedings of the 2012 local non-negative matrix factorization basis images, Image Vis. Comput. 26
5th IAPR International Conference on Biometrics (ICB), 2012, pp. 367–371. (2008) 1515–1523.
2686 J.A. Unar et al. / Pattern Recognition 47 (2014) 2673–2688
[110] J. Seo, H. Park, Robust recognition of face with partial variations using local [141] R. Hill, Retina Identification, Biometrics: Personal Identification in Networked
features and statistical learning, Neurocomputing 129 (2014) 41–48. Society (The International Series in Engineering and Computer Science), vol.
[111] T. Barbu, Unsupervised SIFT-based face recognition using an automatic hier- 256, Kluwer Academic Publishers, Dordrecht (1999) 67–72.
archical agglomerative clustering solution, Proc. Comput. Sci. 22 (2013) 385–394. [142] M. Sofka, C.V. Stewart, Retinal Vessel Centerline Extraction Using Multiscale
[112] F. Samaria, S. Young, HMM-based architecture for face identification, Image Matched Filters, Confidence and Edge Measures,, IEEE Trans. Med. Imaging
Vis. Comput. 12 (1994) 537–543. 25 (2006) 1531–1546.
[113] A. Albiol, D. Monzo, A. Martin, J. Sastre, A. Albiol, Face recognition using [143] J. Xu, O. Chutatape, E. Sung, C. Zheng, P. Chew Tec Kuan, Optic disk feature
HOG–EBGM, Pattern Recognit. Lett. 29 (2008) 1537–1543. extraction via modified deformable model technique for glaucoma analysis,
[114] M. Burge, W. Burger, Ear biometrics, in: A. Jain, et al., (Eds.), Biometrics, Pattern Recognit. 40 (2007) 2063–2076.
Springer, US, 1996, pp. 273–285. [144] J. Daugman, How iris recognition works,, IEEE Trans. Circuits Syst. Video
[115] L. Yuan, Z.-c. Mu, Ear recognition based on 2D images, in: First IEEE Technol. 14 (2004) 21–30.
International Conference on Biometrics: Theory, Applications, and Systems, [145] K.W. Bowyer, K. Hollingsworth, P.J. Flynn, Image understanding for iris
BTAS 2007, 2007, pp. 1–5. biometrics: A survey,, Comput. Vis. Image Underst. 110 (2008) 281–307.
[116] D.J. Hurley, M.S. Nixon, J.N. Carter, Force field energy functionals for image [146] R. Derakhshani, A. Ross. A texture-based neural network classifier for
feature extraction,, Image Vis. Comput. 20 (2002) 311–317. biometric identification using ocular surface vasculature. in: International
[117] J. Dong, Z. Mu, Multi-pose ear recognition based on force field transforma- Joint Conference on Neural Networks. IJCNN 2007. IEEE, August 2007. pp.
tion, in: Second International Symposium on Intelligent Information Tech- 2982–2987.
nology Application, 2008, IITA'08, 2008, pp. 771–775. [147] S. Crihalmeanu, A. Ross, On the use of multispectral conjunctival vasculature
[118] T. Yuizono, Y. Wang, K. Satoh, S. Nakayama, Study on individual recognition as a soft biometric, in: IEEE Workshop on Applications of Computer Vision
for ear images by using genetic local search, in: Proceedings of the 2002 (WACV), 2011, pp. 204–211.
Congress on Evolutionary Computation, CEC'02, 2002, pp. 237–242. [148] J. Matey, D. Ackerman, J. Bergen, M. Tinker, Iris recognition in less con-
[119] L. Nanni, A. Lumini, Fusion of color spaces for ear authentication, Pattern strained environments, Adv. Biom. (2008) 107–131.
Recognit. 42 (2009) 1906–1913. [149] T. Tan, Z. He, Z. Sun, Efficient and robust segmentation of noisy iris images for
[120] K. Chang, K.W. Bowyer, S. Sarkar, B. Victor, Comparison and combination of non-cooperative iris recognition, Image Vis. Comput. 28 (2010) 223–230.
ear and face images in appearance-based biometrics, IEEE Trans. Pattern [150] H. Proença, L.A. Alexandre, Introduction to the special issue on the segmen-
Anal. Mach. Intell. 25 (2003) 1160–1165. tation of visible wavelength iris images captured at-a-distance and on-the-
[121] L. Lu, Z. Xiaoxun, Z. Youdong, J. Yunde, Ear recognition based on statistical shape move, Image Vis. Comput. 28 (2010) 213–214.
model, in: Proceedings of the First International Conference on Innovative [151] H. Proença, G. Santos, Fusing color and shape descriptors in the recognition of
Computing, Information and Control, ICICIC'06, 2006, pp. 353–356. degraded iris images acquired at visible wavelengths, Comput. Vis. Image
[122] L. Yuan, Z.-c. Mu, Y. Zhang, K. Liu, Ear recognition using improved non- Underst. 116 (2012) 167–178.
negative matrix factorization, in: Proceedings of the 18th International [152] H. Proença, L.A. Alexandre, Introduction to the special issue on the recogni-
Conference on Pattern Recognition, ICPR 2006, 2006, pp. 501–504. tion of visible wavelength iris images captured at-a-distance and on-the-
[123] L. Nanni, A. Lumini, A multi-matcher for ear authentication, Pattern Recognit. move,, Pattern Recognit. Lett. 33 (2012) 963–964.
Lett. 28 (2007) 2219–2226. [153] BBC NEWS, Black Hat: Iris Scanners ‘Can be Tricked’ by Hackers. Available:
[124] B. Arbab-Zavar, M.S. Nixon, Robust log-gabor filter for ear biometrics, 〈http://www.bbc.com/news/technology-18997580〉 (visited 03.08.12).
in: Proceedings of the 19th International Conference on Pattern Recognition, [154] P. Li, H. Ma, Iris recognition in non-ideal imaging conditions,, Pattern
ICPR 2008, 2008, pp. 1–4. Recognit. Lett. 33 (2012) 1012–1018.
[125] A. Kumar, T.-S.T. Chan, Robust ear identification using sparse representation [155] L. Ma, Y. Wang, T. Tan, Iris recognition based on multichannel Gabor filtering,
of local texture descriptors,, Pattern. Recognit. 46 (2013) 73–85. in: Proceedings of the Fifth Asian Conference on Computer Vision, 2002,
[126] H.-J. Zhang, Z.-C. Mu, W. Qu, L.-M. Liu, C.-Y. Zhang, A novel approach for ear pp. 279–283.
recognition based on ICA and RBF network, in: Proceedings of 2005 [156] M. Li, T. Tieniu, W. Yunhong, Z. Dexin, Efficient iris recognition by character-
International Conference on Machine Learning and Cybernetics 2005, izing key local variations, IEEE Trans. Image Process. 13 (2004) 739–750.
pp. 4511–4515. [157] C. Sanchez-Avila, R. Sanchez-Reillo, Two different approaches for iris recog-
[127] X. Wang, W. Yuan, Gabor wavelets and General Discriminant analysis for ear nition using Gabor filters and multiscale zero-crossing representation,,
recognition, in: Proceedings of the 8th World Congress on Intelligent Control Pattern Recognit. 38 (2005) 231–240.
and Automation (WCICA), 2010, pp. 6305–6308. [158] R. Szewczyk, K. Grabowski, M. Napieralska, W. Sankowski, M. Zubert,
[128] J.D. Bustard, M.S. Nixon, Toward unconstrained ear recognition from two- A. Napieralski, A reliable iris recognition algorithm based on reverse
dimensional images, IEEE Trans. Syst., Man and Cybern., Part A: Syst. Hum. 40 biorthogonal wavelet transform, Pattern Recognit. Lett. 33 (2012) 1019–1026.
(2010) 486–494. [159] W.W. Boles, B. Boashash, A human identification technique using images of the
[129] A. Kumar, M. Hanmandlu, M. Kuldeep, H. Gupta, Automatic ear detection for iris and wavelet transform, IEEE Trans. Signal Process. 46 (1998) 1185–1188.
online biometric applications, in: Proceedings of the Third National Con- [160] R.P. Wildes, Iris recognition: an emerging biometric technology, Proc. IEEE 85
ference on Computer Vision, Pattern Recognition, Image Processing and (1997) 1348–1363.
Graphics (NCVPRIPG), 2011, pp. 146–149. [161] D.M. Monro, S. Rakshit, Z. Dexin, DCT-Based Iris Recognition,, IEEE Trans.
[130] S. Prakash, P. Gupta, An efficient ear recognition technique invariant to Pattern Anal. Mach. Intell. 29 (2007) 586–595.
illumination and pose,, Telecommun. Syst. (2011) 1–14. [162] T. Tan, X. Zhang, Z. Sun, H. Zhang, Noisy iris image matching by using
[131] Y. Wang, Z.-c. Mu, H. Zeng, Block-based and multi-resolution methods for ear multiple cues,, Pattern Recognit. Lett. 33 (2012) 970–977.
recognition using wavelet transform and uniform local binary patterns, in: [163] S. Zhenan, T. Tieniu, Ordinal measures for Iris recognition, IEEE Trans. Pattern
19th International Conference on Pattern Recognition, ICPR 2008, 2008, Anal. Mach. Intell. 31 (2009) 2211–2226.
pp. 1–4. [164] S.A. Schuckers, N.A. Schmid, A. Abhyankar, V. Dorairaj, C.K. Boyce, L.A. Hornak,
[132] Y. Guo, Z. Xu, Ear recognition using a new local matching approach, in: On techniques for angle compensation in nonideal iris recognition,, IEEE Trans.
Proceedings of the 15th IEEE International Conference on Image Processing, Syst., Man, Cybern., Part B: Cybern. 37 (2007) 1176–1190.
2008, ICIP 2008, 2008, pp. 289–292. [165] D.R.S. Kumar, K.B. Raja, N. Nuthan, B. Sindhuja, P. Supriya, R.K. Chhotaray, S.
[133] W. Xiaoyun, Y. Weiqi, Human ear recognition based on block segmentation, Pattnaik, Iris Recognition Based on DWT and PCA, in: International Con-
in: International Conference on Cyber-Enabled Distributed Computing and ference on Computational Intelligence and Communication Networks (CICN),
Knowledge Discovery, 2009, CyberC'09, 2009, pp. 262–266. 2011, pp. 489–493.
[134] M. Yaqubi, K. Faez, S. Motamed, Ear recognition using features inspired by [166] C. Wen-Shiung, C. Chi-An, S. Sheng-Wen, and C. Shun-Hsun, Iris recognition
visual cortex and support vector machine technique, in: International using 2D-LDA þ 2D-PCA, in: IEEE International Conference on Acoustics,
Conference on Computer and Communication Engineering, ICCCE 2008, Speech and Signal Processing,ICASSP 2009, 2009, pp. 869–872.
2008, pp. 533–537. [167] L. Ma, T. Tan, Y. Wang, D. Zhang, Personal identification based on iris texture
[135] P. Yan, K.W. Bowyer, Biometric recognition using 3D ear shape, IEEE Trans. analysis,, IEEE Trans. Pattern Anal. Mach. Intell. 25 (2003) 1519–1533.
Pattern Anal. Mach. Intell. 29 (2007) 1297–1308. [168] G. Santos, E. Hoyle, A fusion approach to unconstrained iris recognition,
[136] H. Liu, J. Yan, Multi-view ear shape feature extraction and reconstruction, in: Pattern Recognit. Lett. 33 (2012) 984–990.
Proceedings of the Third International IEEE Conference on Signal-Image [169] M. De Marsico, M. Nappi, D. Riccio, Noisy Iris recognition integrated scheme,
Technologies and Internet-Based System, SITIS'07, 2007, pp. 652–658. Pattern Recognition Letters 33 (2012) 1006–1011.
[137] H. Chen, B. Bhanu, Human ear recognition in 3D, IEEE Trans. Pattern Anal. [170] P. Li, X. Liu, N. Zhao, Weighted co-occurrence phase histogram for iris
Mach. Intell. 29 (2007) 718–737. recognition, Pattern Recognit. Lett. 33 (2012) 1000–1005.
[138] H. Zeng, J.-Y. Dong, Z.-C. Mu, Y. Guo, Ear recognition based on 3D keypoint [171] R.M. Sundaram, B.C. Dhara, Neural network based Iris recognition system
matching, in: Proceedings of the 2010 IEEE 10th International Conference on using Haralick features, in: Proceedings of the 3rd International Conference
Signal Processing (ICSP), 2010, pp. 1694–1697. on Electronics Computer Technology (ICECT), 2011, pp. 19–23.
[139] D Zhang, Z Liu, J.-q. Yan, P.-f. Shi, Tongue-print:anovelbiometricspattern, in: [172] R.M. Farouk, R. Kumar, K.A. Riad, Iris matching using multi-dimensional
S.-W Lee, S Li (Eds.), Advances in Biometrics, vol. 4642, Springer, Berlin artificial neural network, Comput. Vis., IET 5 (2011) 178–184.
Heidelberg, 2007, pp. 1174–1183. [173] G. Hong-ying, Z. Yue-ting, P. Yun-he, An iris recognition method based on multi-
[140] H. Oinonen, H. Forsvik, P. Ruusuvuori, O. Yli-Harja, V. Voipio, H. Huttunen, orientation features and non-symmetrical SVM, J. Zhejiang Univ. Sci. A 6 (2005)
Identity verification based on vessel matching from fundus images, in: 428–432.
Proceedings of the 17th IEEE International Conference on Image Processing [174] C. Belcher, Y. Du, Region-based sift approach to iris recognition, Opt. Lasers
(ICIP), 2010, pp. 4089–4092. Eng. 47 (2009) 139–147.
J.A. Unar et al. / Pattern Recognition 47 (2014) 2673–2688 2687
[175] Q. Wang, X. Zhang, M. Li, X. Dong, Q. Zhou, Y. Yin, Adaboost and multi- [207] V. Hautamaki, T. Kinnunen, I. Karkkainen, J. Saastamoinen, M. Tuononen,
orientation 2D Gabor-based noisy iris recognition,, Pattern Recognit. Lett. 33 P. Franti, Maximum a posteriori adaptation of the centroid model for speaker
(2012) 978–983. verification, Signal Process. Lett., IEEE 15 (2008) 162–165.
[176] D. Wenbo, S. Zhenan, T. Tieniu, Iris Matching Based on Personalized Weight [208] J. He, L. Liu, G. Palm, A discriminative training algorithm for VQ-based
Map,, IEEE Trans. Pattern Anal. Mach. Intell. 33 (2011) 1744–1757. speaker identification, IEEE Trans. Speech Audio Process. 7 (1999) 353–356.
[177] V. Velisavljevic, Low-complexity Iris coding and recognition based on [209] M.F. BenZeghiba, H. Bourlard, User-customized password speaker verifica-
directionlets, IEEE Trans. Inf. Forensics Secur. 4 (2009) 410–417. tion using multiple reference and background models, Speech Commun. 48
[178] K. Roy, P. Bhattacharya, C.Y. Suen, Towards nonideal iris recognition based on (2006) 1200–1213.
level set method, genetic algorithms and adaptive asymmetrical SVMs, Eng. [210] D.A. Reynolds, T.F. Quatieri, R.B. Dunn, Speaker verification using adapted
Appl. Artif. Intell. 24 (2011) 458–475. Gaussian mixture models, Digit. Signal Process. 10 (2000) 19–41.
[179] C.V. Stewart, T. Chia-Ling, B. Roysam, The dual-bootstrap iterative closest [211] D.A. Reynolds, R.C. Rose, Robust text-independent speaker identification
point algorithm with application to retinal image registration, IEEE Trans. using Gaussian mixture speaker models,, IEEE Trans. Speech Audio Process.
Med. Imaging 22 (2003) 1379–1394. 3 (1995) 72–83.
[180] R.S. Choras, Retina recognition for biometrics, in: Proceedings of the Seventh [212] S. Furui, Cepstral analysis technique for automatic speaker verification,, IEEE
International Conference on Digital Information Management (ICDIM), 2012, Trans. Acoust., Speech Signal Process. 29 (1981) 254–272.
pp. 177–180. [213] L. Bahler, J. Porter, A. Higgins, Improved voice identification using a nearest-
[181] M.Z.C. Azemin, D.K. Kumar, L. Sugavaneswaran, S. Krishnan, Supervised neighbor distance measure, in: IEEE International Conference on Acoustics,
retinal biometrics in different lighting conditions, in: Annual International Speech, and Signal Processing, 1994. ICASSP-94, vol. 1, 1994, pp. I/321–I/323.
Conference of the IEEE, Engineering in Medicine and Biology Society, EMBC, [214] K.R. Farrell, R.J. Mammone, K.T. Assaleh, Speaker recognition using neural
2011, pp. 3971–3974. networks and conventional classifiers, IEEE Trans. Speech Audio Process. 2
[182] J. Zheng, J. Tian, K. Deng, X. Dai, X. Zhang, M. Xu, Salient feature region: a new (1994) 194–205.
method for retinal image registration, IEEE Trans. Inf. Technol. Biomed. 15 [215] G.-J. Jang, T.-W. Lee, Y.-H. Oh, Learning statistically efficient features for
(2011) 221–232. speaker recognition, Neurocomputing 49 (2002) 329–348.
[183] S. Crihalmeanu, A. Ross, Multispectral scleral patterns for ocular biometric [216] W.M. Campbell, J.P. Campbell, D.A. Reynolds, E. Singer, P.A. Torres-
recognition,, Pattern Recognit. Lett. 33 (2012) 1860–1869. Carrasquillo, Support vector machines for speaker and language recognition,
[184] Shannon Soltysiak, Hamed Valizadegan. DNA as a Biometric Identifier. Comput. Speech Lang. 20 (2006) 210–229.
Available: 〈http://www.cse.msu.edu/ cse891/Sect601/CaseStudy/DNABiome [217] W.M. Campbell, D.E. Sturim, D.A. Reynolds, Support vector machines using
tricIdentifier.pdf〉 (accessed 18.05.12). GMM supervectors for speaker verification, Signal Process. Lett., IEEE 13
[185] IBG. Is DNA a Biometric? Available: 〈http://www.ibgweb.com/products/ (2006) 308–311.
reports/free/dna-biometric〉 (accessed 18.05.12). [218] S. Nemati, R. Boostani, M. Jazi, Retracted: a novel text-independent speaker
[186] M.D. Gibbs, Biometrics: body odor authentication perception and acceptance, verification system using ant colony optimization algorithm, in: A. Elmoataz,
SIGCAS Comput. Soc. 40 (2010) 16. et al., (Eds.), Image and Signal Processing, vol. 5099, Springer, Berlin
[187] A.P. Pons, P. Polak, Understanding user perspectives on biometric technology,, Heidelberg, 2008, pp. 421–429.
Commun. ACM 51 (2008) 115–118. [219] F. Monrose, A.D. Rubin, Keystroke dynamics as a biometric for authentication,
[188] Z. Korotkaya, Biometric person authentication: Odor, Inner report in Depart- Future Gener. Comput. Syst. 16 (2000) 351–359.
ment of Information Technology, Laboratory of Applied Mathematics, [220] R. Giot, M. El-Abed, B. Hemery, C. Rosenberger, Unconstrained keystroke
Lappeenranta University of Technology, in Advanced Topics in Information dynamics authentication with shared secret, Comput. Secur. 30 (2011)
Processing: Biometric Person Authentication, 2003. 427–445.
[189] S.A. Israel, J.M. Irvine, A. Cheng, M.D. Wiederhold, B.K. Wiederhold, ECG to [221] R. Giot, M. El-Abed, and C. Rosenberger, Keystroke dynamics with low
identify individuals, Pattern Recognit. 38 (2005) 133–142. constraints svm based passphrase enrollment, in: Proceedings of the IEEE
[190] G. Wübbeler, M. Stavridis, D. Kreiseler, R.-D. Bousseljot, C. Elster, Verification 3rd International Conference on Biometrics: Theory, Applications, and
of humans using the electrocardiogram, Pattern Recognit. Lett. 28 (2007) Systems, 2009, BTAS'09, 2009, pp. 1–6.
1172–1175. [222] R.N. Rodrigues, G.F. Yared, C.R.d.N. Costa, J.B. Yabu-Uti, F. Violaro, L.L. Ling,
[191] J.M. Irvine, S.A. Israel, W. Todd Scruggs, W.J. Worek, eigenPulse: Robust Biometric access control through numerical keyboards based on keystroke
human identification from cardiovascular function,, Pattern Recognit. 41 dynamics, Advances in Biometrics, Springer (2005) 640–646.
(2008) 3427–3435. [223] M. Rybnik, M. Tabedzki, K. Saeed. A keystroke dynamics based system for user
[192] K. Phua, J. Chen, T.H. Dat, L. Shue, Heart sound as a biometric, Pattern identification. in: Preceedings of the 7th Computer Information Systems and
Recognit. 41 (2008) 906–919. Industrial Management Applications. CISIM'08, IEEE, June 2008, pp. 225–230.
[193] M. Karnan, M. Akila, N. Krishnaraj, Biometric personal authentication using [224] D. Hosseinzadeh, S. Krishnan, Gaussian mixture modeling of keystroke
keystroke dynamics: a review, Appl. Soft Comput. 11 (2011) 1565–1573. patterns for biometric applications,, IEEE Trans. Syst., Man, and Cybern., Part
[194] A. Fazel, S. Chakrabartty, An overview of statistical pattern recognition C: Appl. Rev. 38 (2008) 816–826.
techniques for speaker verification, Circuits Syst. Mag., IEEE 11 (2011) 62–81. [225] F. Monrose, A. Rubin, Authentication via keystroke dynamics, in: Proceedings
[195] S. Nemati, M.E. Basiri, Text-independent speaker verification using ant colony of the 4th ACM conference on Computer and communications security, 1997,
optimization-based selected features,, Expert Syst. Appl. 38 (2011) 620–630. pp. 48–56.
[196] J.D. Woodward Jr, Biometrics: a Look at Facial Recognition, DTIC Document [226] M. Obaidat, D. Macchairolo, A multilayer neural network system for compu-
2003. ter access security, IEEE Trans. Syst., Man Cybern. 24 (1994) 806–813.
[197] C. Vivaracho-Pascual, M. Faundez-Zanuy, J.M. Pascual, An efficient low cost [227] S. Yong, W.K. Lai, G. Goghill, Weightless neural networks for typing
approach for on-line signature recognition based on length normalization biometrics authentication, Knowl.-based Intell. Inf. Eng. Syst. (2004)
and fractional distances,, Pattern Recognit. 42 (2009) 183–193. 284–293.
[198] L. Zongyi, S. Sarkar, Improved gait recognition by gait dynamics normal- [228] S. Cho, C. Han, D.H. Han, H.-I. Kim, Web-based keystroke dynamics identity
ization,, IEEE Trans. Pattern Anal. Mach. Intell. 28 (2006) 863–876. verification using neural network, J. Organ. Comput. Electron. Commer. 10
[199] M. Derawi, P. Bours, Gait and activity recognition using commercial phones, (2000) 295–307.
Comput. Secur. (2013). [229] N. Bartlow, B. Cukic, Evaluating the reliability of credential hardening
[200] M.R. Hestbek, C. Nickel, C. Busch, Biometric gait recognition for mobile through keystroke dynamics, in: Proceeding of the 17th International
devices using wavelet transform and support vector machines, in: Proceed- Symposium on Software Reliability Engineering, 2006, ISSRE'06, 2006,
ings of the 19th International Conference on Systems, Signals and Image pp. 117–126.
Processing (IWSSIP), 2012, pp. 205–210. [230] W.G. De Ru, J.H. Eloff, Enhanced password authentication through fuzzy
[201] J.R. Kwapisz, G.M. Weiss, S.A. Moore, Cell phone-based biometric identifica- logic, IEEE Expert 12 (1997) 38–45.
tion, in: Proceedings of the Fourth IEEE International Conference on [231] E. Yu, S. Cho, Keystroke dynamics identity verification—its problems and
Biometrics: Theory Applications and Systems (BTAS), 2010, pp. 1–7. practical solutions, Comput. Secur. 23 (2004) 428–440.
[202] M.O. Derawi, C. Nickel, P. Bours, C. Busch, Unobtrusive User-Authentication [232] D. Umphress, G. Williams, Identity verification through keyboard character-
on Mobile Phones Using Biometric Gait Recognition, in: Proceedings of the istics,, Int. J. Man Mach. Stud. 23 (1985) 263–273.
Sixth International Conference on Intelligent Information Hiding and Multi- [233] S. Bleha, C. Slivinsky, B. Hussien, Computer-access security systems using
media Signal Processing (IIH-MSP), 2010, pp. 306–311. keystroke dynamics, IEEE Trans. Pattern Anal. Mach. Intell. 12 (1990)
[203] M.O. Derawi, D. Gafurov, R. Larsen, C. Busch, P. Bours, Fusion of gait and 1217–1222.
fingerprint for user authentication on mobile devices, in: Proceedings of the [234] J.R. Montalvão Filho, E.O. Freire, On the equalization of keystroke timing
2nd International Workshop on Security and Communication Networks histograms, Pattern Recognit. Lett. 27 (2006) 1440–1446.
(IWSCN), 2010, pp. 1–6. [235] L. Wang, T. Tan, H. Ning, W. Hu, Silhouette analysis-based gait recognition for
[204] L. Rabiner, B.-H. Juang, Fundamentals of speech recognition. 1993, Prenctice human identification, IEEE Trans. Pattern Anal. Mach. Intell. 25 (2003)
Hall, Englewood Cliffs, NJ, 2001. 1505–1518.
[205] S. Davis, P. Mermelstein, Comparison of parametric representations for [236] T. Dacheng, L. Xuelong, W. Xindong, S.J. Maybank, General tensor discrimi-
monosyllabic word recognition in continuously spoken sentences, IEEE nant analysis and gabor features for gait recognition, IEEE Trans. Pattern
Trans. Acoust., Speech Signal Process. 28 (1980) 357–366. Anal. Mach. Intell. 29 (2007) 1700–1715.
[206] F. Soong, A. Rosenberg, L. Rabiner, B. Juang, A vector quantization approach to [237] R.T. Collins, R. Gross, J. Shi, Silhouette-based human identification from body
speaker recognition, in: IEEE International Conference on Acoustics, Speech, shape and gait, in: Proceedings of the Fifth IEEE International Conference on
and Signal Processing, ICASSP'85, 1985, pp. 387–390. Automatic Face and Gesture Recognition, 2002, pp. 366–371.
2688 J.A. Unar et al. / Pattern Recognition 47 (2014) 2673–2688
[238] Z. Xue, D. Ming, W. Song, B. Wan, S. Jin, Infrared gait recognition based on [262] H. Feng, C.C. Wah, Online signature verification using a new extreme points
wavelet transform and support vector machine, Pattern Recognit. 43 (2010) warping technique, Pattern Recognit. Lett. 24 (2003) 2943–2951.
2904–2910. [263] B. Li, K. Wang, D. Zhang, On-line signature verification based on pca
[239] I.R. Vega, S. Sarkar, Statistical motion model based on the change of feature (principal component analysis) and mca (minor component analysis), Bio-
relationships: human gait-based recognition, IEEE Trans. Pattern Anal. Mach. metric Authentication, Springer, Berlin Heidelberg (2004) 540–546.
Intell. 25 (2003) 1323–1328. [264] M. Parizeau, R. Plamondon, A comparative analysis of regional correlation,
[240] A. Veeraraghavan, A.K. Roy-Chowdhury, R. Chellappa, Matching shape dynamic time warping, and skeletal tree matching for signature verification,,
sequences in video with applications in human movement analysis, IEEE IEEE Trans. Pattern Anal. Mach. Intell. 12 (1990) 710–717.
Trans. Pattern Anal. Mach. Intell. 27 (2005) 1896–1909. [265] B. Li, D. Zhang, K. Wang, Online signature verification based on null
[241] C. Chen, J. Liang, H. Zhao, H. Hu, J. Tian, Factorial HMM and parallel HMM for component analysis and principal component analysis, Pattern Anal. Appl.
gait recognition, IEEE Trans. Syst., Man, Cybern., Part C: Appl. Rev. 39 (2009) 8 (2006) 345–356 (2006/02/01).
114–123. [266] M. Faundez-Zanuy, On-line signature recognition based on VQ-DTW,, Pattern
[242] M.-H. Cheng, M.-F. Ho, C.-L. Huang, Gait analysis for human identification Recognit. 40 (2007) 981–992.
through manifold learning and HMM, Pattern Recognit. 41 (2008) [267] DEEPNETSECURITY, TYPESENSE, Available: 〈http://www.deepnetsecurity.
2541–2553. com/tokens/bio/typesense/〉 (accessed 08.12.13).
[243] Z. Liu, S. Sarkar, Improved gait recognition by gait dynamics normalization, [268] P. Unsang, A.K. Jain, Face matching and retrieval using soft biometrics, IEEE
IEEE Trans. Pattern Anal. Mach. Intell. 28 (2006) 863–876. Trans. Inf. Forensics Secur. 5 (2010) 406–415.
[244] C. Chen, J. Liang, H. Zhao, H. Hu, J. Tian, Frame difference energy image for [269] J.R. Lyle, P.E. Miller, S.J. Pundlik, D.L. Woodard, Soft biometric classification
gait recognition with incomplete silhouettes, Pattern Recognit. Lett. 30 using periocular region features, in: Proceedings of the Fourth IEEE Interna-
(2009) 977–984. tional Conference on Biometrics: Theory Applications and Systems (BTAS),
[245] J.M. Nash, J.N. Carter, M.S. Nixon, Dynamic feature extraction via the velocity 2010, pp. 1–7.
Hough transform, Pattern Recognit. Lett. 18 (1997) 1035–1047. [270] H. Han, A.K. Jain, Tattoo Based Identification: Sketch to Image Matching.
[246] D. Cunado, M.S. Nixon, J.N. Carter, Automatic extraction and description of [271] A.K. Jain, J.E. Lee, R. Jin, Tattoo-ID: automatic tattoo image retrieval for
human gait models for recognition purposes, Comput. Vis. Image Underst. 90 suspect and victim identification, Advances in Multimedia Information
(2003) 1–41. Processing–PCM, 2007, Springer, Berlin Heidelberg (2007) 256–265.
[247] N.V. Boulgouris, Z.X. Chi, Gait recognition using radon transform and linear [272] L. Jung-Eun, J. Rong, A.K. Jain, W. Tong, Image Retrieval in Forensics: Tattoo
discriminant analysis, IEEE Trans. Image Process. 16 (2007) 731–740. Image Database Application,, MultiMedia, IEEE 19 (2012) 40–49.
[248] L. Wang, D. Suter, Analyzing human movements from silhouettes using [273] D.L. Woodard, S. Pundlik, P. Miller, R. Jillela, A. Ross, On the Fusion of
manifold learning, in: IEEE International Conference on Video and Signal Periocular and Iris Biometrics in Non-ideal Imagery, in: Proceedings of the
Based Surveillance, 2006, AVSS'06, 2006, pp. 7–7. 20th International Conference on Pattern Recognition (ICPR), 2010,
[249] E. Zhang, Y. Zhao, W. Xiong, Active energy image plus 2DLPP for gait pp. 2010204.
recognition, Signal Process. 90 (2010) 2295–2302. [274] S. Bharadwaj, H.S. Bhatt, M. Vatsa, R. Singh, Periocular biometrics: When iris
[250] Y. Mu, D. Tao, Biologically inspired feature manifold for gait recognition, recognition fails, in Proceedings of the Fourth IEEE International Conference
Neurocomputing 73 (2010) 895–902. on Biometrics: Theory Applications and Systems (BTAS), 2010, pp. 1–6.
[251] F. Dadashi, B.N. Araabi, and H. Soltanian-Zadeh, Gait recognition using [275] L.P. Rodriguez, A.G. Crespo, M. Lara, B.R. Mezcua, Study of Different Fusion
wavelet packet silhouette representation and transductive support vector Techniques for Multimodal Biometric Authentication, in: IEEE International
machines, in: Proceedings of the 2nd International Congress on Image and Conference on Wireless and Mobile Computing Networking and Commu-
Signal Processing, CISP'09, 2009, pp. 1–5. nications, 2008, WIMOB'08, 2008, pp. 666–671.
[252] K. Huang, H. Yan, Stability and style-variation modeling for on-line signature [276] M. Gudavalli, A. V. Babu, S. V. Raju, D. S. Kumar. Multimodal biometrics—
verification,, Pattern Recognit. 36 (2003) 2253–2270. sources, architecture and fusion techniques: an overview. in: Preceedings of
[253] M. Wirotius, J.-Y. Ramel, and N. Vincent, Selection of points for on-line the 2012 International Symposium on Biometrics and Security Technologies
signature comparison, in: Proceedings of the Ninth International Workshop (ISBAST), IEEE, March 2012. pp. 27–34).
on Frontiers in Handwriting Recognition, WFHR-9 2004, 2004, pp. 503–508. [277] ACUITY: Market Intelligence, Biometrics Market Development: Mega Trends
[254] M.A. Ferrer, J.B. Alonso, C.M. Travieso, Offline geometric parameters for and Meta Drivers, 2007. Available: 〈http://www.acuity-mi.com/hdfsjosg/
automatic signature verification using fixed-point arithmetic,, IEEE Trans. euyotjtub/Biometrics%202007%20London.pdf〉 (accessed 02.08.12).
Pattern Anal. Mach. Intell. 27 (2005) 993–997. [278] Market set for 21% growth, Biometric Technology Today, vol. 2011, 2011, p. 2.
[255] J. Fierrez, J. Ortega-Garcia, D. Ramos, J. Gonzalez-Rodriguez, HMM-based on- [279] IBG. BMIR 2009-2014. Available: 〈http://www.ibgweb.com/products/reports/
line signature verification: feature extraction and signature modeling, bmir-2009-2014〉 (visited on 02.08.12).
Pattern Recognit. Lett. 28 (2007) 2325–2334. [280] Analyst finds thriving local markets in Asia Pacific, Biometric Technology
[256] H. Baltzakis, N. Papamarkos, A new signature verification technique based on Today, vol. 2011,2011, p. 2.
a two-stage neural network classifier,, Eng. Appl. Artif. Intell. 14 (2001) [281] NADRA. National Database and Registration Authority Pakistan, Available:
95–103. 〈http://nadra.gov.pk/〉 (visited 02.08.12).
[257] C. Şenol, T. Yıldırım, Signature verification using conic section function [282] big think, World's Biggest Biometrics ID Scheme, Available: 〈http://bigthink.
neural network, Computer and Information Sciences—ISCIS 2005, Springer com/ideafeed/worlds-biggest-biometrics-id-scheme〉 (visited 02.08.12).
(2005) 524–532. [283] IEEE Spectrum. Fast Start for world's Biggest Biometrics ID Project Available:
[258] D. Muramatsu, M. Kondo, M. Sasaki, S. Tachibana, T. Matsumoto, A Markov 〈http://spectrum.ieee.org/computing/it/fast-start-for-worlds-biggest-biome
chain Monte Carlo algorithm for Bayesian dynamic signature verification, trics-id-project/〉 (visited on 02.08.12).
IEEE Transactions on Information Forensics and Security 1 (1) (2006) 22–34. [284] Precise Biometrics technology to back Nigerian national ID and bank cards,
[259] E.J. Justino, F. Bortolozzi, R. Sabourin, A comparison of SVM and HMM Biometric Technology Today, vol. 2011, 2011, p. 12.
classifiers in the off-line signature verification, Pattern Recognit. Lett. 26 [285] African countries Embrace biometric e-id, Biometric Technology Today, vol.
(2005) 1377–1385. 2011, 2011, pp. 1–2, .
[260] A. Kholmatov, B. Yanikoglu, Identity authentication using improved online [286] J. Ennis, Swapping PINs for palms – the potential of biometric technology in
signature verification method, Pattern Recognit. Lett. 26 (2005) 2400–2408. retail banking, Biom. Technol. Today 2012 (2012) 8–9.
[261] M. Hanmandlu, M.H.M. Yusof, V.K. Madasu, Off-line signature verification [287] find BIOMETRICS. LaserCard Provides National ID Card Program to Saudi
and forgery detection using fuzzy modeling, Pattern Recognit. 38 (2005) Arabia Receives – Contract Worth $2.1 Million – 16th March, 2011. Available:
341–356. 〈http://www.findbiometrics.com/industry-news/i/8707/〉. (visited 02.08.12).
Jawed Akhtar Unar received his B. Engg. (Bachelors) in Computer Systems Engineering in 1994 and M. Engg. (Masters) in 2005 from Mehran University of Engineering &
Technology, Jamshoro, Sindh, Pakistan. He joined Quaid-e-Awam University of Engineering, Science & Technology University in 2006 and still affiliated with the same University.
Currently, he is pursuing Ph.D. from University of Malaya, Kuala Lumpur, Malaysia. His field of interest is Image Processing, Pattern Recognition, Biometrics and Computer Vision.
Woo Chaw Seng is a senior lecturer at the Faculty of Computer Science and Information Technology, University of Malaya. His research interests include image processing and
mobile applications.
Almas Abbasi is a Research Scholar at the Faculty of Computer Science and Information Technology, University of Malaya. Her research interests include image processing and
information hiding.