BIOMETRIC TECHNOLOGIES AND APPLICATIONS Patrick S. P. Wang Northeastern University Boston, MA, USA email:
[email protected]
Svetlana N. Yanushkevich Biometric Technology Laboratory University of Calgary Calgary, Alberta, Canada email:
[email protected]
ABSTRACT Biometric technologies focus on the verification and identification of humans using their possessed biological (anatomical, physiological and behavioral) properties. This paper overview current advances in biometric technologies, as well as some aspects of system integration, privacy and security and accompanying problems such as testing and evaluation. KEY WORDS Biometrics, intelligent systems, security, image processing, pattern recognition.
corresponding computer models, is somewhat opposite to identification, which require signal processing with the purpose of featuring the models used for recognition. However, we could generate a synthetic face from its corresponding computer model. Such a model could include muscular dynamics to model facial expressions. In Table 1, the basic terminology of synthetic biometrics is given. In particular, the term “face reconstruction” indicates a class of problems aimed at synthesizing a model of a static face. The term “mimics animation” is defined as the process of modeling facial appearance and facial topology, a behavioral characteristic called facial expression, the visible result of synthetic emotion.
1 Introduction Biometric technology may be defined as the automated use of physiological or behavioral characteristics to determine or verify an individual’s identity. The word biometric also refers to any human physiological or behavioral characteristic which possesses the requisite biometric properties [1,2]. They are: Universal (every person should have that characteristic), Unique (no two people should be exactly the same in terms of that characteristic), Permanent (invariant with time), Collectable (can be measured quantitatively), Reliable (must be safe and operate at a satisfactory performance level), Acceptable (non-invasive and socially tolerable), and Non-circumventable (how easily the system is fooled into granting access to impostors). Biometric technologies attempt to generate computer models of the physical and behavioural characteristics of human beings with a view to reliable personal identification. The above characteristics include visual images and other human phenomena such as speech, gait, odour, DNA, and indeed anything at all which might help to uniquely identify the individual. Biometric technologies are usually thought as implementation of pattern recognition algorithms, since it is aimed at identifying humans. It, therefore, adapts methods and algorithms from computer vision, but it is not restricted to visual images. Recently, the other direction of the biometrics, namely biometric synthesis, has obtain more attention. Synthesis, or rendering biometric phenomena from their
2 Direct Biometrics The term direct biometrics refer to traditional human recognition methods. The most recent development and applications of biometric technologies largely rely on the fundamental definition of pattern matching, which require learning (analysis) and recognition (synthesis). The significant progress have been achieved in this area since developing the computer-aided tools for signal (images, audio signal and other) processing, analysis and pattern recognition systems [3]. 2.1 Biometric data analysis and pattern recognition The problem of analysis of biometric information has long been investigated. Many researchers have provided efficient solutions for human authentication based on signature, fingerprints, facial characteristics, hand geometry, keystroke analysis, ear, gait, iris and retina scanning. Active research is being conducted using both traditional and emerging technologies, to find better solutions to the problems of verification where claimants are checked against their specific biometric records in a database, and identification where a biometric database is searched to see if a particular candidate can be matched to any record. Detailed descriptions and analysis of the principles, methods, technologies, and core ideas used in biometric authentication systems can be found in [2,4]. It explains the definition and measurement of performance and examines the factors involved in choosing between different biometrics. It also delves into practical applications and covers the parameters critical for successful system integration:
Table 1. Direct and inverse problems of biometric technology. DIRECT PROBLEM Signature identification Handwriting character recognition face recognition Voice identification and speech recognition Iris and retina identification Fingerprint identification hand geometry identification Infrared identification Gait identification Ear identification
recognition accuracy, total cost of ownership, acquisition and processing speed, intrinsic and system security, privacy and legal requirements, and user acceptance.
INVERSE PROBLEM Signature forgery Handwritten text forgery Face reconstruction and mimics animation Voice and speech synthesis Iris and retina image synthesis Fingerprint imitation Hand geometry imitation infrared image reconstruction Gait modeling Ear-print imitation
teraction which enable input of handwriting and signatures [8,9]. 2.5 Faces
2.2 Multimodal biometrics Through rapid advancements and developments of past years, there have been tremendous achievements have been reach in improvement of reliability of recognition. For example, there have been developed character recognition methods that can reach as high as almost 99.99% accuracy rate [3]. However, real-life environments pose significant difficulties on biometric-based recognition systems. To increase overall reliability, the contemporary biometric systems measure multiple physiological or behavioral traits. This approach is called multimodal biometrics (see Handbook of Multibiometrics [5]. The most often multi-biometric data, employed in the biometric systems, include iris and retina of the eye, fingerprint, geometry and palmprint of the hand, and also face and ears. For example, face geometry is a highly dynamic but reach topological structure (smile, lip, brow, and eye movements). Combining facial images with more static biometric such as fingerprint, is an example of the concept of multibiometrics. The multimodal human recognition is deployed in physical access security systems and other areas such as banking systems [1,5,6]. 2.3 Fingerprints Fingerprint is, perhaps, the oldest type of biometrics, used in ancient world. The most popular utilization example of fingerprints is forensic investigations [7]. Today’s fingerprint readers is the most developed type of biometric sensors. 2.4 Signatures Current interest in signature analysis is motivated by the development of improved devices for human-computer in-
Face recognition systems detect patterns, shapes, and shadows in the face, perform feature extraction and recognition of facial identity. In the broader view, it include all types of facial processing such as tracking, detection, analysis and synthesis. The most popular approach is based on eigenfaces, that represent the differences between the face under recognition and the enrolled ones in the database. The principlecomponent analysis using higher-order statistics is the underlying mathematics for this facial pattern recognition. Many biometric systems are confused when identifying the same person smiling, aged, with various accessories (moustache, glasses), and/or in badly lit conditions. For robustness of recognition, advanced techniques such as morphable models and expression-invariant face representation methods. On the other hand, facial recognition tools can be improved by training on a set of synthetic facial expressions and appearance/environment variations generated from real facial images [10,11]. The most advance on-going research in this area is devoted to understanding of how humans can routinely perform robust face recognition, in order to improve machine recognition of human faces. This research is relevant to computer vision paradigm. A comprehensive references to the current state-of-the-art approaches to face processing can be found in [12]. 2.6 Iris and retina Iris recognition systems scan the surface of the iris to compare patterns. It was Daugman who first introduced socalled Daugman code to represent an iris pattern. The iris biometrics is considered to be the most reliable one, as the plots for comparing the hamming distances for 9.1 million iris comparisons to the Beta-binomial distribution showed that the data fit the distribution remarkably well
[11]. Retina recognition systems scan the surface of the retina and compare nerve patterns, blood vessels and such features. Various model of the iris, retina, and eye have been used to improve recognition, and can be found in [1416]. 2.7 Gait biometrics Gait recognition is defined as the identification of a person through the pattern produced by walking [6,17]. The potential of gait as a biometric was encouraged by the considerable amount of evidence available, especially in biomechanics literature [19,20]. A unique advantage of gait as biometrics is that it offers potential for recognition at a distance or at low resolution, when other biometrics might not be perceivable. As gait is behavioural biometrics there is much potential for within-subject variation [21]. This includes footwear, clothing and apparel. Recognition can be based on the (static) human shape as well as on movement, suggesting a richer recognition cue. Model-based techniques use the shape and dynamics of gait to guide the extraction of a feature vector. Gait signature derives from bulk motion and shape characteristics of the subject, articulated motion estimation using an adaptive model and motion estimation using deformable contours. 2.8 Other biometrics A variety of biometrics such as ear geometry [22], odour, electrocardiogram, keystroke dynamics [1,5] and other other physiological and behavioral characteristics are being investigated to be deployed for human authentication.
3 Inverse Biometrics Synthetic biometrics offer the ways to improve reliability of the biometric-based systems. It can be accomplished by providing variations on biometric data mimicking unavailable or hard to access data (for example, modeling of badly lit faces, noisy iris images, “wet” fingerprints etc.) Such an artificial biometric data are understood as biologically meaningful data for existing biometric systems [23,24]. Synthetic data can also be used for “spoofing” biometric devices with “forged” data. Summarizing, synthetic biometric can: Improve data the performance of existing identification systems. This can be accomplished by using automatically generated biometric data to create statistically meaningful sets of data variations (appearance, environmental, and others, including “forgeries”). Improve the robustness of biometric devices by modeling the strategies and tactics of forgery. Improve the efficiency of training systems by providing the user-in-training with the tools to model various conditions of biometric data acquisition (non-contact such
as passive surveillance, contact, cooperative, noncooperative), environmental factors (light, smog, temperature), appearance (aging, camouflage). Therefore, synthetic biometric data plays an important role in enhancing the security of biometric systems. Traditionally, security strategies (security levels, tools, etc.) are designed based on assumptions about a hypothetical robber or forger. Properly created artificial biometric data provides for detailed and controlled modeling of a wide range of training skills, strategies and tactics, thus enabling a better approach to enhancing the system’s performance. This study aims to develop new approaches for the detection of attacks on security systems. The tools for analysis and recognition (direct biometrics) of human characteristics have their analogs in the synthesis (Figure 1. RECOGNITION AND IDENTIFICATION OF BIOMETRIC DATA
SYNTHETIC BIOMETRIC DATA GENERATION
BEHAVIORAL PHYSICAL MODELS
VISIBAL BAND IR BAND ACOUSTIC BAND
DISTANCE NEAR DISTANCE CONTACT MODELS
KNOWLEDGE DOMAIN SIGNAL DOMAIN REPRESENTATION
NONCORRELATED CORRELATED DATA
Figure 1. Analysis-by-synthesis approach in facial image. The crucial point of modeling in biometrics is the analysis-by-synthesis paradigm. This paradigm states that synthesis of biometric data can verify the perceptual equivalence between original and synthetic biometric data, i.e., synthesis based feedback control. For example, facial analysis can be formulated as deriving a symbolic description of a real facial image. The aim of face synthesis is to produce a realistic facial image from a symbolic facial expression model [25]. 3.1 Synthetic fingerprints Albert Wehde was the first to “forge” fingerprints in the 1920’s. Wehde “designed” and manipulated the topology of synthetic fingerprints at the physical level. The forgeries
were of such high quality that professionals could not recognize them. Today’s interest in automatic fingerprint synthesis addresses the urgent problems of testing fingerprint identification systems, training security personnel, biometric database security, and protecting intellectual property. Cappelli et al. [26] developed a commercially available synthetic fingerprint generator called SFinGe. Kuecken [27] developed a method for synthetic fingerprint generation based on natural fingerprint formation and modeling based on state-of-the-art dermatoglyphics.
Many biometric systems are confused when identifying the same person smiling, aged, with various accessories (moustache, glasses), and/or in badly lit conditions (Fig. 2). Facial recognition tools can be improved by training on a set of synthetic facial expressions and appearance/environment variations generated from real facial images.
3.2 Synthetic signatures The simplest method of generating synthetic signatures is based on geometrical models [28–30]. Spline methods and Bezier curves are used for curve approximation, given some control points. The following evaluation properties are distinguished for synthetic signatures [31]: statistical, kinematical (pressure, speed of writing, etc.), geometric, also called topological, and uncertainty (generated images can be intensively “infected” by noise) properties. A model based on combining shapes and physical models in synthetic handwriting generation has been developed in [32]. The so-called delta-log normal model was developed in [9]. This model can produce smooth connections between characters, but can also ensure that the deformed characters are consistent with the models. In [33], it was proposed to generate character shapes by Bayesian networks. By collecting handwriting examples from a writer, a system learns the writers’ writing style. 3.3 Synthetic retina and iris images Iris pattern painting has been used by ocularists in manufacturing glass eyes or contact lenses for sometime. The ocularist’s approach to iris synthesis is based on the composition of painted primitives, and utilized layered semitransparent textures built from topological and optic models [34]. These methods are widely used by today’s ocularists: vanity contact lenses are available with fake iris patterns printed onto them (designed for people who want to change eye colors). Other approaches include image processing and synthesis techniques such as PCA combined with super-resolution [35], and random Markov field [36]. In [37], a cancelable iris image design is proposed for the problem as follows. The iris image is intentionally distorted to yield a new version. For example, a simple permutation procedure is used for generating a synthetic iris. An alternative approach is based on generation of iris layer patterns [31].
Figure 2. Modeling of facial accessories, aging, drunk, and a badly lit faces.
Usage of databases of synthetic faces in a facial recognition context has been reported in [38]. 3.5 Usage of synthetic biometrics The commercially available synthetic fingerprints generator [26] has been used, in particular, in the Fingerprint Verification Test competition since 2003. An example of a tool used to create databases for fingerprints is SFinGe, developed at the University of Bologna (http://bias.csr.unibo.it/research/biolab/sfinge.html). The generated databases were entered in the Fingerprint Verification Competition FVC2004 and performed just as well as real fingerprints [39]. 3.6 Cancelable biometrics The issue of protecting privacy in biometric systems has inspired the area of so-called cancelable biometrics. It was first initiated by the Exploratory Computer Vision Group at IBM T.J. Watson Research Center and published in [37]. Cancelable biometrics aim to enhance the security and privacy of biometric authentication through generation of “deformed” biometric data, i.e. synthetic biometrics. Instead of using a true object (finger, face), the fingerprint or face image is intentionally distorted in a repeatable manner, and this new print or image is used. If, for some reason, the old print or image is “stolen”, an essentially “new” print can be issued by simply changing the parameters of the distortion process. This also results in enhanced privacy for the user since his true print is never used anywhere, and different distortions can be used for different types of accounts.
3.4 Synthetic faces 3.7 Biometric data model validation Face recognition systems detect patterns, shapes, and shadows in the face. The reverse process - face reconstruction is a classical problem of criminology.
Data generated by various models are classified as acceptable or unacceptable for further processing and use in var-
ious applications. The application-specific criteria must provide a reasonable level of acceptability. Acceptability is defined as a set of characteristics which distinguish original and synthetic data. A model that approximates original data at reasonable levels of accuracy for the purpose of analysis is not considered a generator of synthetic biometric information. Artificial biometric data must be verified for their meaningfulness. Statistical model verification is accomplished by solving the equations that describe physicsbased models, and obtaining the correct values. Model validation must prove if the equations that describe the model are right. Comparing the statistical distributions of real biometrics to the statistical distributions from empirical and physics-based models for a wide range of operational conditions validates these models for the range of conditions provided by the real biometric samples [40]. A simple method for validating these distributions is via visual comparison of overlapped distributions. The MITRE research project [23] used synthetically generated faces to better understand the performance of face recognition systems. If a person’s photo in the system’s database was taken 10 years ago, is it possible to identify the person today? A pose experiment was also conducted with synthetic data to isolate and measure the effect of camera angle in one-degree increments. The modeling technique will provide an effective, more structured basis for risk management in a large biometric system. This will help users choose the most effective systems to meet their needs in the future. 3.8 Ethical and social aspects of inverse biometrics Ethical and social aspects of inverse biometrics include several problems, in particular, the prevention of undesirable side-effects, and targeting of areas of social concern in biometrics. Prevention of undesirable side effects aims at studying the potential negative impacts of biometrics, as far as important segments of society are concerned, and how can these be prevented. The undesirable ethical and social effects of the solutions of inverse biometrics have not been studied yet. However, it is possible to predict some of them. The particular examples of negative impact of synthetic biometrics are asinformation follows: can be used not only for Synthetic biometric improving the characteristics of biometric devices and systems, but also can be used by forgers to discover new strategies of attack. Synthetic biometric information can be used for generating multiple copies of original biometric information.
4 Conclusions The concept of inverse biometrics arose from the analysisby-synthesis paradigm, and has become an integral part of the modeling and simulation of human biometrics in many applications. Data generated by various models are used as
databases (for example, databases of synthetic fingerprint) of synthetic biometrics for testing biometric hardware and software. The other application is biometric-based decision making support systems for security, banking, and forensic applications. A generator of synthetic biometric information (for example, an aging or surgically changed face), is a vital component of such systems. Yet another recently emerging application is the creation of simulators for training highly qualified personnel in biometric-based physical access control systems such as airport gates, hospital registration, and others. The ability to increase the reliability and accuracy of these systems while probing their vulnerabilities under numerous environmental conditions is critical, as biometrics becomes an essential part of law enforcement and security communities.
Acknowledgment This paper was prepared while Dr. P.S.P. Wang was the iCore Visiting Professor in the University of Calgary, Canada. His visit was supported by iCore, the Informatics Circle of Research Excellence, the province of Alberta, Canada. Dr. Yanushkevich acknowledges the Canadian Foundation for Innovations (CFI) and the National Science and Engineering Research Council of Canada, for funding the equipment and researchers in the Biometric Technologies Laboratory (http://btlab.enel.ucalgary.ca), University of Calgary.
References [1] A. Jain, R. Bolle, and S. Pankanti, Eds., Biometrics: Personal Identification in a Networked Society, Kluwer, 1999 [2] R. Bolle, J. Connell, S. Pankanti, N. Ratha, and A. Senior, Guide to Biometrics, Springer, 2004 [3] C. Chen, and P. Wang, Eds. Handbook of Pattern Recognition and Computer Vision, World Scientific Publishers, 2005 [4] J. Wayman, A. Jain, D. Maltoni, and D. Maio, Eds. Biometric Systems Technology, Design and Performance Evaluation, Spinger, 2005. [5] A. Ross, K. Nandakumar, and A. Jain Handbook of Multibiometrics, International Series on Biometrics Springer, 6, 2006. [6] V. Shmerko, Phil Phillips, G. Kukharev, W. Rogers, and S. Yanushkevich, Biometric Technologies, Proc. Int. Conf. The Biometrics: Fraud Prevention, Enchanced Service, Las Vegas, Nevada, 270–286, 1997 [7] S. A. Cole, Suspect Identities – A History of Fingerprinting and Criminal Identification, Harvard University Press, 2001 [8] J. M. Hollerbach, An oscilation theory of handwriting, Biological Cybernetics, 39, 139–156, 1981. [9] R. Plamondon, and W. Guerfali, The generation of handwriting with delta-lognormal synergies, Bilogical Cybernetics, 78, pp. 119–132, 1998
[10] P. Ekman, and E. L. Rosenberg, Eds. What the Face Reveals: Basic and Applied Studues of Spontaneouse Expression Using the Facial Action Coding System (FACS) Oxford University Press, 1997 [11] M. Pantic, and L. J. M. Rothkrantz, Automatic analysis of facial expressions: the state-of-the-art, IEEE Trans. Pattern Analysis and Machine Intelligence, 22, 12, pp. 1424–1445, 2000 [12] W. Zhao, R. Chelappa, Eds., Face Processing: Advanced Modeling and Methods, Elsevier, 2006 [13] J. Daugman, The importance of being random: Statistical principles of iris recognition, Pattern Recognition, 36(2), pp. 279–291, 2003 [14] A. Can, C.V. Steward, B. Roysam, and H. L. Tanenbaum, A feature-based, robust, hierarchical algorithm for registering pairs of images of the curved human retina, IEEE Trans. Analysis and Machine Intelligence, 24(3), 347–364, 2002 [15] T. Moriyama, J. Xiao, T. Kanade, and J. F. Cohn, Meticulously Detailed Eye Model and its Application to Analysis of Facial Image. Proc. IEEE Int. Conf. Systems, Man, and Cybernetics, 629–634, 2004 [16] C. Sanchez-Avila, and R. Sanchez-Reillo, Irisbased biometric recognition using dyadic wavelet transform, IEEE Aerospace and Electronic Systems Magazine, October, 3–6, 2002 [17] D. Cunado, M.S. Nixon, and J.N. Carter, Automatic extraction and description of human gait models for recognition purposes, Computer Vision and Image Understanding, 9(1), 1–14, 2003 [18] M.S. Nixon, J.N. Carter, M.G. Grant, L. Gordon, and J.B. Hayfron-Acquah, Automatic recognition by gait: progress and prospects, Sensor Review, 23(4), 323-331, 2003 [19] L. Sloman, M. Berridge, S. Homatidis, D. Dunter, and T. Duck, Gait patterns of depressed patients and normal subjects, American Journal of Psyhiatry, 139(1), 94–97, 1982 [20] I.P.I. Pappas, M.R. Popovic, T. Keller, V. Dietz, and M. Morari, A reliable gait phase detection system, IEEE Transaction on Neural System Rehabilitation Engineering, June, 9(2), 113–125, 2001 [21] N. V. Boulgouris, D. Hatzinakos, and K. N. Plataniotis, Gait recognition: A challening signal processing technology for biometric identification, IEEE Signal Processing Magazine, November, 78–90, 2005. [22] D. J. Hurley, M. S. Nixon, and J. N. Carter, Force field feature extraction for ear biometrics, Computer Vision and Image Understanding, 98, 491–512, 2005 [23] N. M. Orlans, D.J. Buettner, and J. Marques, A Survey of Synthetic Biometrics: Capabilities and Benefits, Proc. Int. Conf. Artificial Intelligence, CSREA Press, vol. I, 499– 505, 2004 [24] S. N. Yanushkevich, A. Stoica, S. N. Srihari, V. P. Shmerko, and M. L. Gavrilova, Simulation of Biometric Information: The New Generation of Biometric Systems, In Proc. Int. Workshop Modeling and Simulation in Biometric Technology, Calgary, Canada, 87–98, 2004 [25] Y. Luo, M. L. Gavrilova, M. C. Sousa, J. Pivovarov,
and S. Yanushkevich, Morphing Facial Expressions from Artistic Drawings, In T. Simos, G. Maroulis, Eds., Advanced in Computational Methods in Sciences and Engineering. Lecture Series on Computer and Computational Sciences, Brill Academic Publishers, The Netherlands, Vol. 4, 1507–1511, 2005 [26] R. Cappelli, Synthetic fingerprint generation, In D. Maltoni, D. Maio, A. K. Jain, and S. Prabhakar, Eds., Handbook of Fingerprint Recognition, 203–232, Springer, 2003 [27] M. U. Kuecken, and A. C. Newell, A model for fingerprint formation, Europhysics Letters 68(1), 141–146, 2004 [28] J. J. Brault, and R. Plamondon, A complexity measure of handwritten curves: modelling of dynamic signature forgery, IEEE Trans. Systems, Man and Cybernetics, 23, 400–413, 1993 [29] I. Guyon, Handwriting synthesis from handwritten glyphs, Proc. 5th Int. Workshop Frontiers of Handwriting Recognition, Colchester, UK, 309-312, 1996 [30] C. Oliveira, C. Kaestner, F. Bortolozzi, and R. Sabourin, Generation of signatures by deformation, Proc. the BSDIA97, 283–298, Curitiba, Brazil, 1997 [31] S. N. Yanushkevich, A. Stoica, V. P. Shmerko, and D.V. Popel, Biometric Inverse Problems, Taylor & Francis/ CRC Press, Boca Raton, FL., 2005 [32] J. Wang, C. Wu, Y. Q. Xu, H. Y. Shum, and L. Li, Learning Based Cursive Handwriting Synthesis, Proc. 8th Int. Workshop Frontiers in Handwriting Recognition, Ontario, Canada, 157–162, 2002. [33] H. Choi, S. J. Cho, and J. H. Jin Kim, Generation of handwritten characters with bayesian network based on-line handwriting recognizers, In Proc. 17th Int. Conf. Document Analysis and Recognition, Edinburgh, Scotland, 995–999, 2003 [34] A. Lefohn, B. Budge, P. Shirley, R. Caruso, and E. Reinhard, An ocularists approach to human iris synthesis, Computer Graphics and Applications, IEEE Magazine, 23(6), 70–75, 2003 [35] J. Cui, Y. Wang, J. Huang, T. Tan, Z. Sun, and L. Ma, An iris image synthesis method based on PCA and superresolution, Proc. Int. Conf. Pattern Recognition, 2004 [36] S. Makthal, and A. Ross, Synthesis of iris images using Markov random fields, Proc. 13th European Signal Processing Conf., Antalya, Turkey, 2005 [37] N. Ratha, J. Connell, and R. Bolle, Enhancing security and privacy in biometrics-based authentication systems, IBM Systems J. 40(3), 614–634, 2001 [38] K. Sumi, C. Liu, and T. Matsuyama, Study on Synthetic Face Database for Performance Evaluation, Proc. Int. Conf. Biometric Authentication, 598–604, LNCS3832, Springer, 2006 [39] The Fingerprint Verification Competition FVC2004 http://bias.csr.unibo.it/fvc2004/databases.asp [40] Y. Ma, M. Schuckers, and B. Cukic, Guidelines for appropriate use of simulated data for bio-authentication research, Proc. 4th IEEE Workshop Automatic Identification Advanced Technologies, Buffalo, New York, 251–256, 2005
