FINGER VEIN VERIFICATION TECHNOLOGY FOR

MOBILE APPARATUS

Hideo Sato

FVA Business Department, Sony Corporation, Japan

Keywords: Biometrics, Finger vein, Authentication, Mobile devices, Forge resistance.

Abstract: In this paper we present a new finger vein authentication technology for consumer mobile products. The

finger vein patterns are unique for each individual and do not change over a long time. Since finger veins

exist inside of the body, it is extremely hard to be forged. Very short response time while keeping high-level

security authentication is achieved by the new compact-fast-matching scheme and small-size template. The

data size is nearly as small as the one of the minutiae-based fingerprint authentication systems. The compact

sensor size is realized by the method of reflecting scattering light. These technologies enable the use of

finger vein authentication to the mobile devices and smart cards.

1 INTRODUCTION

Biometrics is expected to become more important as

the core technology for mobile network security. In

spite of such expectations, a lot of issues on the

biometrics are pointed out (Virginia Ruiz-Albacete

et al., 2008, Tsutomu Matsumoto et al., 2002). It is

also standardized for example in ISO and JIS (ISO-

standard on vein, JIS (JP) standard on vein).

Especially following issues need to be improved for

mobile usage of biometrics.

•Forge resistance

The fake physical biometric is serious problem.

Actually fake fingerprint was used to trick biometric

airport fingerprint scan. Latent fingerprint is another

issue (Tsutomu Matsumoto et al., 2002). When you

lost your mobile phone, your fingerprint patterns

remain on the surface and it can be used for forge.

•Disability to enrol some users

All the people should be able to use the biometrics

system. About 4% of the population have poor

quality of fingerprint, especially the elder people and

manual worker (Privacy International, 2005). For

these cases one need to consider other biometrics or

password service.

•Environmentally free

Finger print authentication is affected by skin

condition (dry or wet). Dry skin cause the increase

of the false rejection rate in winter season.

•Limited resource environments

The authentication algorithm should be designed to

be small, efficient, and use resources efficiently for

operating on the resource-limited devices. A very

good performance is necessary to achieve a

comfortable authentication. Of course, small size

low cost sensor is must issue for mobile use.

Finger vein authentication technology is expected to

be more secure, reliable than current biometric

systems. However, the usage has been still limited

due to the above issues to be cleared. The purpose of

our development is to improve the actual

performance of the finger vein authentication for

mobile usage. It is necessary to resolve the following

two major issues for wider and practical use.

The first issue is the improvement of the sensor size.

This is a problem concerning the biometric sensor

device and the image processing.

The second issue is the efficient authentication

scheme for mobile usage. This needs improvements

the authentication algorithm. It is also related to the

data size of the biometric template.

Sato H. (2009).

FINGER VEIN VERIFICATION TECHNOLOGY FOR MOBILE APPARATUS.

In Proceedings of the International Conference on Security and Cryptography, pages 37-41

DOI: 10.5220/0002231300370041

 SciTePress

2 VEIN PATTERN

AUTHENTICATION

Finger vein authentication is expected to be a new

technology that replaces the fingerprint. Like irises

and fingerprints, a person's veins are completely

unique. Even twins don't have identical veins, and

each person's veins differ between their left and right

sides. Furthermore veins are not visible usually since

they are covered by the skin, making them extremely

difficult to be counterfeited or tampered. The pattern

also changes very little as a person ages.

The advantages on finger vein authentication in

comparison with other biometrics and listed as

follows;

•Advantages

(1) Robustness

As veins are hidden inside the body, there is little

risk of forgery or theft.

(2) High accuracy

The authentication accuracy is less than 0.1% for the

FRR (False Rejection Rate), less than 0.0001% for

the FAR (False Acceptance Rate).

(3) Very good performance on Failure to Enrol Rate

(FTE)

FTE of the finger vein authentication is nearly equal

to 0%. All the people can use this method.

(*Finger Print = 4%)

(4) Dependence of the surface conditions of the

fingers

The surface conditions of the hands have no effect

on authentication.

It is already used for biometrics ATM cards in

Japan. This shows the trust to vein authentication

technologies by banking field.

•Issues in the past technologies

On the other hand, some weak points of a past vein

technology still remain to be resolved compared

with the fingerprint.

(1) Size of device

It is too large for mobile apparatus.

(2) Authentication speed

It is slower than a general fingerprint

authentication.

(3) Need the guide to fix hand or finger

All of vein sensors used the assistance of finger or

hand guide. Steady and stable operations are

difficult without it. This is not practical for the use

of mobile environment.

(4) High cost of sensor

It is too expensive for the cellular phone.

(5) Data size

The data size of the template needs to make is

very small.

It is necessary to improve these issues for vein

technologies to be used widely in the mobile area.

3 THE PROPOSED FINGER VEIN

TECHNOLOGY

3.1 Biometric sensor

The first issue is the improvement of the sensor size.

•Open Type Capability

Generally other vein authentication systems such as

palms and back of the hands use the reflection of the

body. On the other hand, as near-infrared light is

transmitted through the finger, and strongly scattered

in the inside of the body, it is difficult to capture the

finger vein image by this method. Small finger veins

are masked by the reflection light of the epidermis.

Finger vein are much smaller than other vein such as

palms and back of the hands. It is more difficult to

forge the finger vein patterns than other portions.

This is a big advantage of finger vein. On the other

hand, we had to use the transmission type optical

system for the finger vein image usually. This is the

reason why the size of sensor device is large.

Figure 1: The reflecting scattering light method.

•The reflecting scattering light method

We developed a new finger vein sensor by the

reflecting scattering light method as shown in Figure

1. This uses the scattering light in the inside of

finger. The LED array emits near-infrared light, and

it is strongly diffused. By the deep investigation of

these characteristics we developed a new structure of

SECRYPT 2009 - International Conference on Security and Cryptography

the vein scanner for finger. This method can be

achieved only by lighting from the direction of

single side, which enabled open type's sensor.

It shows the relation between the sensor and LED

unit. For transmission type, the angle is 180 degrees.

The angle between camera and incident light affect

the image quality of vein. The maximum efficiency

and the image quality are obtained by about 120

degrees through investigation. The influence of the

image of the dermis and the epidermis appears by

about 90 degrees, and the vein image quality is

deteriorated. It increases the failure to enrol rate

(FTE).This method (about 120 degrees) can obtain

the high performance equal with a transmission type.

3.2 Feature Extraction Algorithm for

Vein Blood Flow

•Noise Removal

Conventionally CCD sensor is used for vein

authentication systems. CMOS sensor is one of the

solutions for cost saving and size down. CMOS

Image sensors are widely used by cellular phone

because of low cost and low power. But there are

some shortcomings for using CMOS small sensor

for finger vein authentication. The disadvantage of

CMOS imagers is the noise of the image. We

analyzed the characteristics of this noise and tested

various kind of noise filter. Based on these

investigations, we developed noise reduction

algorithm for finger vein image. This enabled to use

low cost and small size CMOS sensor whish has the

same performance as CCD sensor for finger vein

authentication.

•The extraction algorithm for vein patterns

It is possible to classify it into two types roughly

though there are various techniques in the extraction

algorithm for vein patterns.

Type1. Path search algorithms

Type2. Zero crossing methods

There is a report on Type 1 with the line tracking

method for vein pattern (Naoto Miura et al., 2005).

Our technology is based on Type2.

The zero-crossing method is known as a method that

is appropriate for such a pattern extraction. The

Laplacian filter is one of the generalized method for

zero- crossing detection. It is defined as follows.

L(I) = Ixx + Iyy (1)

However, because the character of the vein image is

greatly different according to the race, sex, the age,

and the health condition it was very difficult to

achieve a steady vein pattern extraction by this

method.

We resolved this problem by developing a new filter

with special characteristics that is appropriate for the

vein image noise and the vein continuous patterns. It

has higher-speed than Path search algorithms.

•Centreline detecting of the vein

This method improves the robustness for the

variation of the climate and the health condition.

Figure 2: The change of the vein width.

It is said that the vein patterns changes very little as

a person ages. But it does not mean the vein itself

will not change in any case. Actually the thickness

of vein changes by the physical condition, climate,

and temperature. This two pictures show the change

of the vein width by the height of hand to heart (See

the Figure 2).

Our approach extracts the centreline of the vein

accurately (see Figure 3). This method improves

stability enormously by removing such uncertainties.

Furthermore, this approach is very fast and very

compact.

Figure 3: Feature Extraction Approach.

FINGER VEIN VERIFICATION TECHNOLOGY FOR MOBILE APPARATUS

3.3 New Compact-fast-matching

Algorithm

Another issue is the comfortable authentication

algorithm for mobile usage.

Conventionally it is necessary to use a guide place it

accurately. It is quite difficult to install the finger

guide used for ATM in mobile use.

And it cannot be expected from the user to a certain

forced adjustment on finger position.

Figure 4: New compact-fast-matching scheme.

The common problem of the vein pattern recognition

is repeatability of the taking picture images using the

camera. On the other hand, the algorithm

corresponding to the change including the zooming

and the rotation generally requires big calculation

power. This is a reason why conventional vein

authentication systems need the assistance of finger

or hand guide (BioGuard 2008, Chris Roberts 2006).

In our device, it is enough only by the correction of

a little rotation and a little zooming. We developed a

new match method to correct these errors on polar

coordinates as shown in Figure 4. The shortcoming

of the conventional finger guide was resolved by our

approach of combining the match in the direction of

X-Y and the match in polar coordinates. The rotation

axially of the finger is corrected by the selection

from three templates or more.

The purpose of this approach is to improve the

convenience of the finger vein authentication. Our

approach resolves these uncertainties of height, the

rotation, the angle, and the distance. An efficient

authentication was achieved without a special guide

of the finger by this approach. The test environment

and the result are shown in Table 1 as follows.

Table 1: The performance of the algorithm.

Example column 1 Average operation time

Windows PC 15ms

The authentication performance of this algorithm

was tested by the reference finger vein sensor by the

reflecting scattering light method.

The result is also shown in the following Table 2.

Table 2: FAR and FRR of new algorithm.

Test item FAR ( % ) FRR ( % )

performance 0.0001 0. 1

3.4 Vein Pattern Data Compression

with Accuracy

Our compression scheme transforms shown in

Figure 5 vein pattern into geometric Information. It

compresses the centre lined data of the vein with

accuracy compensation as small as to 112Byte. This

data size is nearly equal to one of the minutiae-based

fingerprint authentication systems. The small size

template is suitable for smart card storing and

mobile phone. It achieves quick authentication

operation.

Figure 5: Vein Pattern Data Compression.

4 PROTOTYPE PHONE

We made the prototype for cellular phone. This

prototype phone adopted the reflecting scattering

light method for the vein sensor. New approach is

coded in C language and suitable for mobile OS. We

investigated the performance of the combination of

new sensor and algorithm, under mobile

environments. The following are the results in the

enrolment and verification of 100 people. This

actual test was as shown in Table 1 shows the

SECRYPT 2009 - International Conference on Security and Cryptography

comfortableness and without any failure. Although

this data is not enough estimating the FTE, FAR and

FRR, yet it shows practicality for mobile usage.

Table 3: The results in the enrolment and verification.

Test item False number Test number

Enrolment

0 100

False Acceptance 0 2000

False Rejection 0 2000

The prototype for cellular phone structure is

explained in Figure 6. The Infra-red LED Unit is

placed at the head of display and sensor is at the

bottom of display. Finger vein authentication

software runs on the OS of cellular phone. The

distance between sensor and finger is about 60mm.

The stability of the authentication performance

depends on this enough distance. It is necessary for

the steady reproducibility of the finger vein image.

Figure 6: The image of Prototype Phone.

The average time of authentication is 250ms with

ARM CPU (150MHz) as shown in Table 4. It

achieved higher speed than a general fingerprint

authentication of a cellular phone.

Table 4: The average time of authentication.

Example column 1 Average time

Mobile Phone

@ARM9 150MHz

250ms

5 CONCLUSIONS

In this paper, we focus on our achievement of our

newly developed finger vein authentication

technologies. The sensor size is improved by the

reflecting of scattering light and the new feature

extraction scheme. By the well combination of these

technologies we realized the use of the low cost

small sized CMOS sensor for the finger vein

authentication. Furthermore, the new compact-fast-

matching and compression scheme contribute to the

efficient authentication. This new Finger vein

technology is expected to be the promising

technology which can improve the shortcomings of

the fingerprint. We expect that our new technologies

will expand the application range of the vein pattern

recognition, especially for mobile usage.

REFERENCES

Virginia Ruiz-Albacete, Pedro Tome-Gonzalez, Fernando

Alonso-Fernandez, Javier Globally, Julian Fierrez, and

Javier Ortega-Garcia 2008, Direct attacks using fake

images in iris verification

Tsutomu Matsumoto, Hiroyuki Matsumoto, Koji Yamada,

Satoshi Hoshino, 2002. Impact of Artificial "Gummy"

Fingers on Fingerprint Systems

Privacy International, 2005, UK Identity Cards and Social

Exclusion

Naoto Miura, Akio Nagasaka, Takafumi Miyatake

2005.Extraction of Finger-Vein Patterns Using

Maximum Curvature Points in Image Profiles

(Hitachi)

BioGuard 2008. Palm Vein Authentication Technology

Chris Roberts 2006, Biometric Technologies - Palm and

Hand

ISO-standard on vein. Biometric data interchange formats

- Part 9: Vascular image data (ISO/IEC 19794-9).

Conformance testing methodology for biometric data

interchange formats - Part 9: Vascular image data

(ISO/IEC 29109-9)

JIS (JP) standard on vein. Evaluation Method for

Accuracy of Vein Authentication Systems (JIS-TR

X0079)

FINGER VEIN VERIFICATION TECHNOLOGY FOR MOBILE APPARATUS