Li S.Z., Jain A.K. (eds.) Encyclopedia of Biometrics

Подождите немного. Документ загружается.

the pulse oximetry measurement. This goes beyond

the performance of the liveness algorith m de-

scribed above, because it requires assessment of

spooﬁng approaches that have yet to be replicated

in the database used to test the liveness algorithm.

Summary

In summar y, liveness systems are being suggested to

reduce the vulnerability due to spooﬁng. Liveness

measures have an inherent performance, that is, abil-

ity to separate spoof and live attempts. In addition,

liveness algorithms have other factors and considera-

tions including ease of use, collectability, user accep-

tance, universality, uniqueness, permanence, and

spoof-ability. One factor, which is difﬁcult to mea-

sure is spoof-abilit y, the possibility that the liveness

measure can be spoofed. In this c hapter, the term

liveness is used, fully acknowledging that it is not a

perfect system and that it is not possible to recreate a ll

possible spoof a ttempts for a sy stem. Furthermore,

there may be measurements, which rule out speciﬁc

spoofs b ut cannot be shown to a bsolutel y mea sure

liveness. For example, algorit hms may be designed

which may readily detect silicon, b ut not gelatin,

spoof images. In summary, it is unlikely that any

system w ill perfectly measure liveness and be spoof-

proof.Livenessmaybeboileddowntoanattemptto

stay one step ahead of those intending to defeat the

system through s poof a ttacks. Methods such as live-

ness or antispooﬁng are critical to the security and

credibility of biometric systems to p rotect them from

security vulnerabilities to the degree needed for a

particular application.

Related Entries

▶ Liveness Iris

▶ Security and Liveness

References

1. Matsumoto, T., Matsumoto, H., Yamada, K., Hoshino, S.: Im-

pact of artiﬁcial ‘gummy’ ﬁngers on ﬁngerprint systems. Proc.

SPIE 4677, 275–289 (2002)

2. Schuckers, S.A.C.: Spooﬁng and anti-spooﬁng measures. Inform

Security Tech Rep 7(4), 56–62 (2002)

3. Valencia, V., Horn, C.: Biometric liveness testing. In: Woodward

J.D. Jr., Orlans, N., Higgins M.R.T. (eds.) Biometrics. McGraw-

Hill, Osborne Media, New York (2002)

4. Schuckers, S.A.C., Abhyankar, A.: A wavelet based approach to

detecting liveness in ﬁngerprint scanners. Paper presented at the

Proceedings of Biometric Authentication Workshop, ECCV,

Prague, May 2004

5. Coli, P., Marcialis, G.L., Roli, F.: Power spectrum-based ﬁnger-

print vitality detection. Paper presented at IEEE Workshop on

Automatic Identiﬁcation Advanced Technologies AutoID

pp. 169–173 (2007)

6. Coli, P., Marcialis, G.L., Roli, F.: Vitality detection from ﬁnger-

print images: A critical survey. Adv. Biometrics 4642, 722–731

(2007)

7. Biel, L., Pettersson, O., Philipson, L., Wide, P.: ECG analysis:

A new approach in human identiﬁcation. IEEE Trans. Instrum.

Meas. 50(3), 808–812 (2001)

8. Kallo , P., Kiss, I., Pod manicz ky, A., Talosi, J.: Detector for recognizing

the living character of a ﬁnger in a ﬁngerprint recognizing apparatus.

Dermo Corporation, U .S. Patent No . 6,175,64. 16 Jan 2001

9. Baldisserra, D., Franco, A., Maio, D., and Maltoni, D.: Fake

ﬁngerprint detection by odor analysis. Paper presented at the

proceedings of International Conference on Biometric Authen-

tication (ICBA06), Hong Kong, Jan 2006

10. Nixon, K.A., Rowe, R.K.: Spoof detection using multispectral

ﬁngerprint imaging without enrollment. Paper presented at the

Proceedings of Biometrics Symposium (BSYM2005), Arlington,

VA, 19–21 Sept 2005

11. Chen, Y., Jain, A., Dass, S.: Fingerprint deformation for

spoof detection. Paper presented at the Proceedings of

Biometrics Symposium (BSYM2005), Arlington, VA, 19–21

Sept 2005

12. Antonelli, A., Cappelli, R., Maio, D., Maltoni, D.: Fake ﬁnger

detection by skin distortion analysis. IEEE Trans. Inform. For-

ensics Security 1(3), 360–373 (2006)

13. Jia, J., Lianhong, C., Kaifu, Z., Dawei, C.: A new approach to fake

ﬁnger detection based on skin elasticity analysis. Adv. Biometrics

4642, 309–318 (2007)

14. Zhang, Y., Tian, J., Chen, X., Yang, X., Shi, P.: Fake ﬁnger

detection based on thin-plate spline distortion model. Adv.

Biometrics 4642, 742–749 (2007)

15. Derakhshani, R., Schuckers, S., Hornak, L., O’Gorman, L.:

Determination of vitality from a non-invasive biomedical mea-

surement for use in ﬁngerprint scanners. Pattern Recogn. 17(2),

383–396 (2003)

16. Schuckers, S.A.C, Derakhshani, R., Parthasardhi, S., Hornak,

L.A.: Liveness detection in biometric devices. In: Electrical

Engineering Handbook, Chapter 26, 3rd edn. CRC, Boca

Raton (2006)

17. Parthasaradhi, S., Derakhshani, R., Hornak, L., Schuckers,

S.A.C.: Time-series detection of perspiration as a liveness test

in ﬁngerprint devices. IEEE Trans. Systems Man Cybern C Appl

Rev 35, 335–343 (2005)

18. Tan, B., Schuckers, S.: Liveness detection using an intensity

based approach in ﬁngerprint scanner. Paper presented at Pro-

ceedings of Biometrics Symposium (BSYM2005), Arlington, VA,

19–21 Sept. 2005

930

Liveness Detection: Fingerprint

19. Moon, Y.S., Chen, J.S., Chan, K.C., So, K., Woo, K.C.: Wavelet

based ﬁngerprint liveness detection. Electron. Lett. 41(20),

1112–1113 (2005)

20. Jin, C., Kim, H., Elliott, S., Liveness detection of ﬁngerprint

based on band-selective fourier spectrum. Information Security

Cryptol –ICISC, vol. 4817, pp. 168–179. Springer, Berlin (2007)

21. Tan, B., Schuckers, S.: A new approach for liveness detection in

ﬁngerprint scanners based on valley noise analysis. J. Electron.

Imag. 17(1), 011009-1–011009-9 (2008)

22. Jia, J., Lianhong, C.: Fake ﬁnger detection based on time-series

ﬁngerprint image analysis. In: Advanced Intelligent Computing

Theories and Applications. With Aspects of Theoretical and

Methodological Issues, vol. 4681, pp. 1140–1150. Springer,

Berlin (2007)

23. Abhyankar, A., Schuckers, S.A.C.: Characterization, similarity

score, and uniqueness of ﬁngerprint perspiration patterns.

In: Kanade et al. (eds.) et al.: Proceedings of Audio- and

Video-Based Biometric Person Authentication: 5th International

Conference. Lecture Notes in Computer Science, vol. 3546,

pp. 860–868. Springer, Berlin (2005)

Liveness Detection: Iris

BORI TOTH

Deloitte & Touche LLP, London, UK

Synonyms

Anti-spooﬁng; Spooﬁng countermeasures; Spoof -

resistance; Vitality tests

Definition

Iris liveness detection techniques are designed to coun-

teract physical

▶ spooﬁng attack s launched against iris

recognition systems. Such attacks include the use of

photographs, video recordings, printed contact lenses

etc. Iris liveness detection mechanisms aim to ascertain

that iris images acquired were acquired from a live and

authorized user present at the time of transaction.

Introduction

Iris recognition systems are among the most accurate

biometric systems available today. Additionally, the iris

is an internal organ which makes it more robust to

spooﬁng attacks when compared to some of the other

biometric technologies, especially ﬁngerprinting. This

is mainly due to two reasons. Firs t, unlike DNA and

ﬁngerprints, individuals do not leave traces of their

irises behind which could be lifted and copied. Second,

it is more difﬁcult to manipulate an internal organ

than to di sguise an external body part such as

the face. Nevertheless, all technologies have inherent

weaknesses which can be exploited, including iris

recognition.

On the one hand, privacy experts argue that bio-

metric information is private. Additionally, the grow-

ing uptake of large-scale biometric systems worldwide

intensiﬁes the fear of hackers stealing biometric infor-

mation from centralized databases. On the other hand,

it is a fact that the acquisition of biometric information

is much easier than breaking datab ases. Facial images

are being taken via public and private CCTV systems

day and night: think of ATMs, petrol stations, banks,

and airports just to name a few places were camera

surveillance systems are usually in 24/7 operation. The

situation is similar for most other biometric traits: we

leave our ﬁngerprints and DNA behind on every sur-

face we touch and phone-based service providers usu-

ally record our voices during phone calls [1]. In the

case of iris recognition, a high resolution image of

someone’s eye can be sufﬁcient to make the technology

work. CCTV image quality might often be too poor

to extract iris images of sufﬁcient quality but this is

merely a hardware question.

We have to accept that our facial and eye images,

voice patterns, ﬁngerprints, and DNA etc. are publicly

available. It has already been shown, through various

experiments that many if not all biometric technolo -

gies including iris recognition are susceptible to spoof-

ing attacks: biometric identiﬁers can be copied

and used to create some

▶ artifacts that can deceive

several biometric devices available today (

▶ Security

and Liveness, Overview).

Therefore, the question is not whether biometrics

can be copied and forged but rather whether devices

can perform accurate liveness testing. The aim of live-

ness testing is to determine if the biometric data

is being captured from a legitimate, live user who is

physically present at the point of acquisition. This

is especially cruc ial for remote authentication services

performed over open networks where neither the end

user’s terminal nor the data transmission channels can

Liveness Detection: Iris

931

be controlled by system operators. An increasing num-

ber of vendors are implementing liveness testing into

their biometric devices, to guard against the threat

of spooﬁng attacks. However, spooﬁng-related issues

remain unknown to many [2].

Risks of Biometric Spoofing

In order to understand security mechanisms such as

liveness detection, it is important to analyze the inher-

ent risks and weaknesses ﬁrst. Like any other security

technolog y, biometrics also have inherent weaknesses

that can potentially lead to security breaches. Suscepti-

bility to spooﬁng attacks is just one possible weakness

inherent to biometric readers (

▶ Biometrics Vulner-

abilities: Overview).

Biometric spooﬁng attacks can be either digital or

physical. Digital attacks are defended against by

authenticating the biometric reader sending the data

and eliminating vulnerable data paths; liveness testing

methods are not applicable in this case. Hence this type

of vulnerability is not being discussed any further in this

article. Physical spooﬁng of a biometric credential refers

to the attack whereby an adversary copies a legitimate

biometric to generate a fake artifact and tries to gain

access to the system using this artifact. Spooﬁng attacks

may be undertaken with the cooperation of the legiti-

mate user, in an effort to delegate access rights, or

without user knowledge by collecting iris pictures from

iris recognition systems and infrared cameras, or facial

images from camera and surveillance equipment [2].

A biometric system can be used in an access control

scenario, either logical or physica l, or as a watch list

application to detect and identify particular ‘‘wanted’’

individuals. In each case, the purpose of the system and

the risks of spooﬁng are different:

1. In an access control scenario, the system keeps a

system is the voluntary, fully-automated immigra-

tion control system, IRIS, which operates at several

major airports in the UK. In such a scenario, fake

iris artifacts could be used to:

a. Mount attacks against existing enrolments in

order to gain unauthorized access – either logi-

cal or physical – to the resources protected by

the iris recognition system and/or to fraudu-

lently associate an audit trail with an unwitting

individual

b. Enroll into the iris system and then delegate

these artifacts across multiple individuals,

undermining the integrity of the system

c. Additionally, a legitimate user could try to re-

pudiate transacti ons associated with his account

or enrolment, claiming instead that they are the

result of attacks, due to the inability of the

biometric system to ensure liveness

2. In a watch list application, the system keeps a

record of people who are being sought by the

authorities or who are to be unequivocally denied

access to the assets or facilities protected by the

biometric device. An examp le of such a system is

the iris-based border control system operated by

United Arab Emirates’ authorities. In such a setup,

it is of course preferable from a registered person’s

point of view not to be recognized by the system. In

connection to the watch lists , an iris artifact could

be used to:

a. A bogus enrolment record can be created so that

the wanted/unauthorized person could contin-

ue to use the system with his real irises without

being detected

b. A wanted person’s iris patterns can be imitated

to lead authorities to think (even just temporar-

ily) that someone from the watch list had been

caught

c. A wanted person’s iris patterns can be disguised

(which had been previously registered in the

system) avoid identiﬁcation and/or gain unau-

thorized access.

Note that the goals of the

▶ impostor are different

in these two scenarios: for access lists, it’s about imper-

sonating a legitimate user, while for the watch lists, it’s

about disguising one’s identity. The latter is believed to

be a much easier task, i.e., it is easier to disguise one’s

iris patterns so that they cannot be recognized anymore

than to imitate someone else’s iris patterns so closely

that a match is achieved.

Spoof-resistance Testing

Governments, academics and an increasing number of

industry players are active in the space of testing the

resilience of biometric systems against spooﬁng. While

vendors and governments tend to keep their results

secret, several test methods and results were published

in recent years by the academics and consultants.

932

Liveness Detection: Iris

In 2002, a report was published by the German

Fraunhofer Research Institute, detailing results of

much earlie r spooﬁng experiments on face, ﬁnger,

and iris systems which they carried out in cooperation

with t he German Federal Institute for Information

security (BSI) [3]. This report was among the ﬁrst to

raise a few eyebrows about the security of biometric

systems available at that time.

A 2002 issue of c’t magazine [4] followed by the

description of further attacks. Alongside several other

biometric technologies, the authors were able to spoof

a low-cost iris recognition device using high-resolution

eye images with cut out holes for the pupil which they

placed in front of a real eye.

A well-known authority in spooﬁng, Professor Tsu-

tomu Matsumoto of Yokohama National University in

Japan, has published the results of two iris spooﬁng

rounds so far. In 2004, he spoofed three iris recognition

cameras with high-resolution photographs with cut

out holes for the pupil placed in front of a real eye

[5]. Only one of the iris systems did not accept the

fake iris for enrolment, but all devices could be spoofed

during veriﬁcation. In 2007, Professor Matsumoto

presented another spooﬁng method using metallic riv-

ets with shiny black rounds heads and printed iris

images [6, 7].

Much research has gone into analyzing the effec-

tiveness of spooﬁng of contact lenses with printed or

hand-painted iris patterns as well. In addition to

printed photographs and printed/painted contact

lenses, other artifacts which could be used to physically

spoof iris devices include screen images, video record-

ings, and artiﬁcial eyes (glass, plastic etc.).

Liveness Detection Mechanisms

Biometric experts have been actively researching meth-

ods to counter the threat of physical spooﬁng of bio-

metric samples for more than a decade now. In

particular, various liveness detection methods have

been conceived and indeed implemented in some

devices. However, as every man-made solution can be

defeated, efforts are ongoing in this area.

System supervision is the ﬁrst line of defense

against spooﬁng. The use of several types of spooﬁng

artifacts becomes inconvenient if not impossible,

if a human supervisor is present at the point of iris

image acquisition. Such examples include the use of

photographs or video recordings. Human operators

can also detect printed or painted lenses but the costs

and inconvenience of such a process make its day-to-

day use prohibitive. Additionally, human performance

is affected by many factors including tiredness, moti-

vation, sickness etc.

Automate d liveness detection can be performed in

a biometric device either at the acquisition or proces-

sing stage. It usually involves enhancements to soft-

ware and/or hardware. The presented set of method s is

not an exclusive list; however, examples are given for

each iris liveness detection category. Automated live-

ness detection techniques measure and ana lyze one of

the following three characteristics [8].

Intrinsic properties of a living body

Methods belonging in this category, analyze static

characteristics of the material presented to the biomet-

ric reader. Such characteristics include densit y and

elasticity (physical properties), capacitance, resistance

and permittivity (electrical properties), reﬂectance and

absorbance (spectral properties), color and opacity

(visual properties), and chemical content analysis

in ﬂuids.

Spectrographic properties of tissue, fat, blood, and

pigment can be used to test for liveness in irises.

Figure 1 shows that different components of living

tissue have distinctive spectrographic signatures. Com-

paring the fractions of light reﬂected in 300–1000 nm

bandwidth can reveal these spectrographic signatures

[9]. If the iris presented to the system is a glass eye, a

photograph, or dead tissue, spectrographic analyses

could help in detecting the spooﬁng attack. In fact,

ink and paper used to create photographic printouts

are often completely ineffective in near-infrared

light, which is used during the acquisition of iris

images [11].

Retinal light reﬂections commonly known as the

‘‘red-eye effect’’ can also be used to detect liveness of

the eye. Essentially, light entering the eye is reﬂected

back to the light source by the retina; this effect can be

captured by a camera. Functional eye cavity optics

make the eye to appear red of the pigment (cal led

retinal or visual purple) in the photoreceptors

of the retina (Fig. 2). Red-eye effect will occur if the

angle between light source, eye and camera is smaller

than 2.5



[10].

Liveness Detection: Iris

933

The iris is a relatively ﬂat internal organ, located

behind the cornea and in front of the lens. When a

printed or hand-painted contact lens is placed over the

eye, the fake ‘‘iris’’ is ﬂoating over an external, round

surface, the cornea (Fig. 3). Therefore, another intrinsic

property for which iris devices can scan is the 3D nature

of the layer containing the iris patterns acquired.

In a natural eye, four optical surfaces reﬂect light:

the front and back surfaces of the cornea as well as the

front and back surfaces of the lens (Fig. 4). These

reﬂections are also referred to as Purkinje reﬂections

or images, named after a Czech physiologist. The front

surface of the cornea produces the brightest reﬂection

while the back of the lens produces the weakest one.

The position of the reﬂected light determines the posi-

tion of the reﬂections – another intrinsic property

which can be used to distinguish between natural

Liveness Detection: Iris. Figure 1 Light-reflecting properties of different components of living tissue can help to detect

iris-spoofing attempts [2, 9, 10].

Liveness Detection: Iris. Figure 2 The ‘‘red-eye effect’’ – the retina reflects the light entering the eye back to the light

source [10].

Liveness Detection: Iris. Figure 3 Contact lenses with

fake iris patterns float over the curved external surface of

the eye whereas the iris is lying in an internal plane inside

the eye.

934

Liveness Detection: Iris

eyes and fake artifacts. A change in the location of the

light source should therefore even screen out photo-

graphs displaying Purkinje reﬂections [11]. It might be

difﬁcult to capture all four Purkinje reﬂections at all

times due to their varying strength; however, it could

be sufﬁcient to analyze the strongest reﬂections coming

from the outer layer of the cornea. Varying positions of

near-infrared lig ht diodes used during image acquisi-

tion could also be used to analyze this property of the

living eye.

John Daugman, the inventor of iris recognition,

has also pointed out that the printing process itself

can leave detectable traces on spooﬁng lenses [11].

A 2D Fourier analysis of the acquired image can show

off traces of printing, as demonstrated in Fig. 5. Pacut

and Czajka have developed automated methods

to analyze artiﬁcial frequencies in printed iris images.

One great advantage of this method is that, it does

not require any additional hardware; it merely analyzes

the iris image already captured. However, according

to Shanon’s theory, the method has a drawback:

it fails once the resolution of the printing device

is higher than twice the resolution of the analysis

camera [12].

Involuntary signals of a living body

Living tissue involuntarily displays dynamic signals

which are measurable. These signals can be attributed

Liveness Detection: Iris. Figure 4 A picture of a natural

eye displaying Purkinje reflections.

Liveness Detection: Iris. Figure 5 2D Fourier analysis extracts remnants of the printing process on a contact lens.

Liveness Detection: Iris

935

to the nervous system and include pulse, blood pres-

sure, pupillary unrest (hippus), perspiration, blood

ﬂow, brain wave signals (EEG), and electrical heart

signals (ECG or EKG).

Daugman mentioned the idea of using involuntary

signals of the body to measure liveness detection in

iris recognition schemes in several of his papers.

A very interesting yet little known involuntary signal

generated by the human body is the hippus, which is

a pupillary steady-state oscillation at about 0.5 Hz,

occurring in eyes without any changes in illuminat-

ion. The coefﬁcient of variation is at least 3% [11],

although it declines with advancing age. This liven-

ess detection technique can effectively be used

to screen out prosthetic eyes, high-resolution

photographs, or dead tissue. The upper graph (with

round dots) in Fig. 6 shows involuntary changes in

pupil size.

▶ Iris recognition algorithms need to track the

inner and outer boundaries of the iris anyway as part

of the extraction process so tracking the changes of

pupil size as well as eyelid movements are relatively

easy liveness detection methods to implement.

Bodily responses to external stimuli

Finally, it is possible to measure dynamic bodily

responses to external stimuli. These liveness detection

methods are challenge-response techniques that either

look for voluntary (behavioral) responses or involun-

tar y (reﬂexive) ones.

Behavioral challenge-response methods require

user cooperation. As an example, the spooﬁng resis-

tance of iris recognition products can be enhanced by

prompting the user to blink or look left and right, and

up and down. If the signal presented to the system is a

photograph or video recording, the system is likely to

recognize these as fakes.

For iris recognition, an involuntary reﬂex of the body

can be easily triggered by changing illumination levels. The

pupil can be driven larger or smaller by changes in lighting

conditions, with a response time constant of about

250 ms for constriction and about 400 ms for dilation

[11]. The lower graph (with square dots) in Fig. 6 shows

the pupillary reﬂex as a diode is switched on and off.

Another interesting effect which can be observed

when the pupil size changes is the nonelastic distortion

Liveness Detection: Iris. Figure 6 Spontaneous pupil size variations with (square dotted line) and without (round dotted

line) any changes in lighting levels [2, 12].

936

Liveness Detection: Iris

of the iris tissue itself. Contact lenses or photographs

won’t be able to imitate this process.

The Effectiveness of Liveness Testing

Methods

Some of the above presented methods have been tested

independently with very promising results [3]. How-

ever, there is a need for a consistent testing framework

to assess the effectiveness of liveness testing methods

and market iris products on an ongoing basis. Schemes

have been proposed by both the academia and industry

[3, 7, 13, 14] but there is still no consensus over

an in ternationally standardized spoof-resistance test-

ing methodology.

The Trade-off between Security and

Convenience

Biometric devices should only be spoof-protected to a

level corresponding to the nature of operations (i.e.,

depending on whether operations are mainly security-

or convenience-focussed) due to the following limita-

tions of liveness detection methods [2]:



Firstly, there is a conjecture that for all biometrics,

the problem of conﬁrming the vitality of a sample

(‘‘liveness testing’’) is more difﬁcult than to make

decisions about matches between templates. The

two distributions of similarity generated by ‘‘genu-

ine’’ and ‘‘spoof’’ samples for the same person are

likely to be closer and to have more overlap in the

vitality test than the two distributions that are

generated in a template matching test by ‘‘same’’

and ‘‘different’’ persons without any spooﬁng effort

[9]. In other words, liveness testing can adversely

affect recognition performance (‘‘Security and

Liveness, Overview’’).



Secondly, liveness tests have the propensity to in-

crease the time to acquire the biometric sample,

thus reducing user convenience.



Finally, the incorporation of liveness tests into a

device usually also means increasing hardware/

software costs.

Summary

Spooﬁng is a real concern with regards to the secur ity

of biometric systems. More and more successful spoof-

ing attempts are being published and even though the

sophistication of these attacks is on the rise, spooﬁng is

still in its infancy. In particular, contact lenses with

hand-painted and printed iris patterns are expected

Liveness Detection: Iris. Table 1 Overview of discussed liveness detection methods for iris recognition with an

indication of their effectiveness

Category Countermeasure Targeted artifacts

Extraction of intrinsic

properties (static)

Spectrographic analysis All

Near-infrared illumination All involving inks and dyes

Red-eye effect All except patterned contact lenses

3D curvature of iris surface Patterned contact lenses

Purkinje reflections All except patterned contact lenses

Frequency spectrum analysis Printed artifacts

Analysis of involuntary

signals (dynamic)

Pupillary unrest (hippus) All except patterned contact lenses which only

party cover the real iris

Eyelid movements All except patterned contact lenses

Challenge-response

methods (dynamic)

Eye movements (blinking, looking in

various directions)

All except patterned contact lenses

Pupillary light reflex All except patterned contact lenses which only

party cover the real iris

Nonelastic distortion of iris tissue All

Liveness Detection: Iris

937

to pose an in creasing threat due to enhancements of

ink quality and printing technologies. Furthermore,

patterned lenses are relatively di fﬁcult to detect when

compared to some of the other spooﬁng methods.

Both the industry and academia are focusing their

efforts to make biometric dev ices more robust but

every countermeasure can eventually be circumvented.

Thus research and development efforts must be

ongoing.

This article illustrates that it is possible to combat

physical spooﬁng attacks with liveness testi ng (Table 1)

but all of these countermeasures come at a certain

price, often affecting user convenience, system prices,

or matching accuracy. Therefore, it is crucial to select a

device that incorporates spooﬁng countermeasures to

a level of sophistication and effectiveness that matches

the requirements of the application.

As spooﬁng techniques are swiftly evolving and

countermeasures have only a limited life cycle, in addi-

tion to the necessary research and development efforts

it is of great importance to perform standardized,

vendor-independent tests of robustness and to assess

on a regular basis the overall level of security provided

by biometric systems.

Related Entries

▶ Anti-spooﬁng

▶ Liveness Detection

▶ Spooﬁng; Biometric Vulnerabilities

References

1. Toth, B.: Biometric ID card debates. ‘‘Biometrie’’ Newsletter of

ISACA Chapter Switzerland, Germany and Austria (2005)

2. Toth, B.: Biometric Liveness Detection. Information Security

Bulletin, ISB1008, CHI Publishing 291–297 (2005)

3. Czajka, A., Strzelczyk, P., Pacut, A.: Making iris recognition more

reliable and spoof resistant. SPIE Newsroom, doi: 10.1117/

2.1200706.0614 (2007)

4. Thalheim, L., Krissler, J., Ziegler, P-M.: Body Check. Biometric

Access Protection Devices and their Programs Put to the Test, c’t

Magazine (2002)

5. Matsumoto, T.: Artiﬁcial ﬁngers and irises: Importance of vul-

nerability analysis. In: Proceedings of the 7th International

Biometrics Conference, London. Elsevier (2004)

6. Matsumoto, T., Kusuda, T.: IECIE Technical Report on

Biometrics Security (2007)

7. Matsumoto, T.: Assessing the security of advanced biometric

systems: Finger, vein and iris. In: Proceedings of the 10th Inter-

national Biometrics Conference, London. Elsevier (2007)

8. Woodward, J.D., Orlans, J.M., Higgins, P.T.: Biometrics. Identity

Assurance in the Information Age, Osborne McGraw-Hill

Osborne Media Berkeley, California, USA (2003)

9. Daugman, J.G.: Iris recognition and anti-spooﬁng countermea-

sures. In: Proceedings of the 7th International Biometrics Con-

ference, London. Elsevier (2004)

10. Toth, B.: Liveness detection for iris recognition, In: Proceedings

of Biometrics and E-Authentication over Open Networks, NIST

Gaithersburg (2005)

11. Daugman, J.G.: Recognizing persons by their iris patterns:

Countermeasures against subterfuge. In: Jain, A.K., et al. (eds.)

Biometrics: Personal Identiﬁcation in a Networked Society, Kluwer

Academic Publishers, Dordrecht, The Netherlands(1999)

12. Pacut, A., Czajka, A.: Liveness detection for iris biometrics. In:

Proceedings of 40th IEEE International Carnahan Conference

on Security Technology, Lexington, KY. IEEE (2006)

13. Dunstone, T.: Current Status of Biometrics in Australia and

New-Zealand including Vulnerability Assessments. In: Proceed-

ings of the 10th International Biometrics Conference, London.

Elsevier (2007)

14. Nanavati, S.: Spooﬁng Biometrics: Deploying Imperfect

Biometrics. In: Proceedings of the 10th International Biometrics

Conference, London. Elsevier (2007)

Live-Scan Furrow Device

It refers to a device, able to read the ridge-valley pat-

tern present on ﬁnger tips, palms, and foot soles. It can

be considered as a generic name grouping all the ﬁn-

gerprint, palmprint, handprint, and soleprint devices.

▶ Fingerprint, Palmprint, Handprint and Soleprint

Sensor

Live-Scan Sensor

A live-scan sensor is a sensor that allows to capture and

digitize biometric data in real time. As opposed to live-

scan acquisition, in off-line acquisition, data is not

digitized in real time (e.g., a ﬁngerprint image is ﬁrst

obtained by smearing ink on the ﬁngertip and creating

938

Live-Scan Furrow Device

an inked impression on paper, and the inked impres-

sion is digitized by scanning the paper).

▶ Fingerprint Databases and Evaluation

Local Adaptive Thresholding

Local adaptive thresholding is used to convert an

image consisting of gray scale pixels to just blac k and

white scale pixels. Usually a pixel value of 0 represents

white and the value 255 represents black with the

numbers from 1 to 254 representing different gray

levels. Unlike the global thresholding technique, local

adaptive thresholding chooses different threshold

values for every pixel in the image based on an analysis

of its neighboring pixels. This is to allow images with

varying contrast levels where a g lobal thresholding

technique will not work satisfactorily. There are a

number of different forms of adaptive thresholding

algorithm reported in the image processing literature.

▶ Hand Vein

Local Descriptors

▶ Local Image Features

Local Fusion

Local fusion in the framework or multi-biometric

score fusion refers to user-speciﬁc score fusion t echni-

ques in which each fusion function is trained based

exclusively on data associated with the claimed user

(both genuine and impostor scores).

▶ Fusion, User-Specific

Local Image Features

KRY S T I A N MIKOLAJCZYK

,TINNE TUYTEL AARS

School of Electronics and Physical Sciences,

University of Surrey, Guildford, Surrey, UK

Department of Electrical Engine ering, Katholieke

Universiteit Leuven, Kasteelpark Arenberg 10, Leuven,

Belgium

Synonyms

Interest points; Keypoints; Local descriptors

Definition

A ▶ local feature is an image pattern which differs from

its immediate neighborhood. It is usually associated

with a change of an image property or several proper-

ties simultaneously, though it is not necessarily loca-

lized exactly on this change. The image properties

commonly considered are intensity, color, and texture.

Figure 1 shows some examples of local features in a

contour image (left) as well as in a grayvalue image

(right). Local features can be points, but also edgels or

small image patches. Typically, some measurements are

taken from a

▶ region centered on a local feature and

converted into descriptors. The descriptors can then be

used for various applications. Three broad categories

of feature

▶ detectors can be distinguished based on

their possible usage. It is not exhaustive or the only way

of categorizing the features but it emphasizes different

properties required by the usage scenarios. First, one

might be interested in a speciﬁc type of local features,

as they may have a speciﬁc semantic interpretation in

the limited context of a certain application. For in-

stance, edges detected in aerial images often corre-

spond to roads; blob detection can be used to

identify impurities in some inspection task; etc. These

were the ﬁrst applications for which local feature

detectors have been proposed. Second, one might be

interested in local features since they provide a limited

set of well localized and individually identiﬁable an-

chor points. What the features actually represent is not

really relevant, as long as their location can be deter-

mined accurately and in a stable manner over time.

This is for instance the situation in most matching

or tracking applications, and especially for camera

Local Image Features

939