See discussions, stats, and author profiles for this publication at: [Link]

net/publication/224086092

Hand geometry verification system: A review

Conference Paper · November 2009

DOI: 10.1109/ICUMT.2009.5345652 · Source: IEEE Xplore

CITATIONS READS
11 3,202

3 authors:

Amit Kumar Singh Amrit Agrawal

National Institute of Technology Patna Apollo Institute of Technology, Kanpur
197 PUBLICATIONS 6,443 CITATIONS 18 PUBLICATIONS 137 CITATIONS

SEE PROFILE SEE PROFILE

Chandra Bhan Pal

Techanical Education Uttar Pradesh
2 PUBLICATIONS 11 CITATIONS

SEE PROFILE

All content following this page was uploaded by Amit Kumar Singh on 28 December 2015.

The user has requested enhancement of the downloaded file.

 Hand Geometry Verification System: A Review

Amit Kumar Singh, Amrit Kumar Agrawal, Chandra Bhan Pal

Department of Computer Science & Engineering, Jaypee University of Information Technology,
Wakanaghat, Solan, (H.P), India.
amit_245singh@[Link]
agrawal.amrit4@[Link]
[Link]@[Link]

Abstract Hand geometry, as the name suggests, refers to the

geometric structure of hand. This structure includes
Biometric system is becoming increasingly width of the finger at various locations, width of palm,
important, since they provide more reliable and thickness of the palm, length of the finger etc. Hand
efficient means of identity verification. The physical geometry measurement is non intrusive and the
dimensions of a human hand contain information that verification involves a simple processing of the
is capable of authenticating the identity of an resulting features.
individual. This information has been popularly known
as hand or palm geometry; the hand geometry based 2. Why Hand Geometry?
identity verification system is being widely used in
various applications like access control, time and What is the most effective biometric measurement?
attendance, point-of-scale, anti-pass back and There is no ideal biometric measurement; each
interactive kiosks etc. This paper presents biometric biometrics has its strengths and limitations, and
user recognition system based on hand geometry. The accordingly each biometric appeal to a particular
goal of a biometric verification system consists in identification (authentication) application. Suitability
deciding whether two characteristics belong to the of a particular biometric to a specific application
same person or not. depends upon several factors [1]; among these factors,
the user acceptability seems to be the most significant.
Keyword For many access control applications, like
immigration, border control and dormitory meal plan
Hand geometry, Biometrics system, Verification, access, very distinctive biometrics, e.g., fingerprint
Identification. and iris, may not be acceptable for the sake of
protecting an individual's privacy. In such situations, it
is desirable that the given biometric indicator be only
1. Introduction distinctive enough for verification but not for
identification. As hand geometry information is not
The question of locating the identity of a person very distinctive, it is one of the biometrics of choice in
can be categorized in to two fundamentally distinct applications like those mentioned above. Hand
types of problems with different inherent geometry-based authentication is also very effective
complexities: Verification and Identification. for various other reasons. Almost all of the working
Associating an identity with an individual is called population has hands and exception processing for
identification. Verification refers to the problem of people with disabilities could be easily engineered [2].
confirming or denying a person’s claimed identity Hand geometry measurements are easily collectible
(Am I who I claim I am?). Identification (who am I?) due to both the dexterity of the hand and due to a
refers to the problem of establishing a subject’s relatively simple method of sensing which does not
identity. The human hand provides a number of impose undue requirements on the imaging optics.
physiological biometric features; among the most Note that good frictional skin is required by
commonly used are fingerprints, the palm print, the fingerprint imaging systems, and a special illumination
geometry of hand etc.

9781-4244-3941-6/09/$25.00 ©2009 IEEE

Authorized licensed use limited to: JAYPEE UNIVERSITY OF INFORMATION TECHNOLOGY. Downloaded on January 4, 2010 at 00:16 from IEEE Xplore. Restrictions apply.
 setup is needed by iris or retina-based identification performance of multiple biometric devices, including
systems. Further, hand geometry is ideally suited for hand geometry [4].
integration with other biometrics, in particular,
fingerprints. For instance, an identification/verification Table1. Comparison between Biometrics
system may use fingerprints for (infrequent)
identification and use hand geometry for (frequent) Technology Fingerpri Iris Facial Hand
verification. It is easy to conceptualize a sensing characteristi nt
system which can simultaneously capture both c
fingerprints and hand geometry. How it works Capture Capture Capture Measures
and and and and
compares compar compar compares
2.1. Advantages fingertip es iris es dimensio
patterns patterns facial ns of
• Acquisition convenience and good patterns hand and
verification performance fingers
Cost of Low High Modera Moderate
• Suitable for medium and low security device te
applications. Enrollment About 3 About 2 About 3 About 1
• Ease of integration. Time minutes, minutes minutes minutes
• Currently being used for functions such as 30 , 15
access control, employee time recording and seconds seconds
point of sale applications. Transaction 9-19 12 10 6-10
• Reasonably high acceptance among users and time seconds seconds seconds seconds
False non .2%-36% 1.9%- 3.3%- 0%-5%
it is opt-in.
match rate 6% 70%
• Works in challenging environments. False match 0%-8% <1% .3%-5% 0%-2.1%
• Low template size, which reduces storage rate
needs. Variability Stable Stable Affecte Stable
with ages d by
2.2. Disadvantages aging
Commercial 1970s 1997s 1990s 1970s
Availability
• Large size of hand geometry device. since
• Limiting the applications of hand geometry
system to verification task only.
• Single hand use only. 2.4. Standards Overview
• It is not highly unique.
• Weather, temperature and medical conditions Standards development efforts focusing on hand
such as pregnancy or certain medication can geometry technology, on both the national and
affect hand size. international levels, are intended to accelerate the
• Hand size and geometry changes over time, development of interoperable authentication-based
especially in the very young and very old. security solutions. ANSI INCITS 396-2005 Hand
Geometry Interchange format defines the data
2.3. United States Government Evaluations interchange format for storing, recording, and
transmitting hand geometry information collected
The US government has sponsored two evaluation from the hand silhouette. It defines both content and
of hand geometry technology. The 1996 Evaluation of format of the data for exchange as well as the units
the INSPASS Hand Geometry Data determined the used for measurement of the hand geometry data [5].
effect of a threshold on system operation, established This national standard corresponds to ISO/IEC CD
false accept and false reject rates as a function of the (Committee Draft) 19794-10 Biometric Interchange
threshold, and presented and estimate of the Receiver Format – Part10, Hand Geometry Silhouette Data on
Operating Characteristics (ROC) curve for the the international standards level (ISO/IEC). The
INSPASS system [3]. The evaluators noted that an international standard is still in draft format and has
estimate was the best that could be achieved with the not yet been approved as an official standard.
available data. A 1991 Performance Evaluation of
Biometric Identification Devices evaluated the relative 3. System Operations

Authorized licensed use limited to: JAYPEE UNIVERSITY OF INFORMATION TECHNOLOGY. Downloaded on January 4, 2010 at 00:16 from IEEE Xplore. Restrictions apply.
 Each human hand is unique. Finger length, width, Different platen heights change the relative position of
thickness, curvatures and relative location of these the body and hand, and affect the hand shape.
features distinguish every human being from every
other person (Figure 1). The hand geometry scanner 3.2. Verification Process
uses a charge coupled device (CCD) camera, infrared
light emitting diodes (LEDs) with mirrors and This process involves matching a given hand to a
reflectors to capture black and white images of the person previously enrolled in the system. Two
human hand silhouetted against a thirty-two thousand snapshots of the hand are taken and the “average”
pixel field. The scanner records no surface details, feature vector is computed. The given feature vector is
ignoring fingerprints, lines, scars and color. The then compared with the feature vector stored in the
process is much like placing a hand on a beaded database associated with the claimed identity. Let X=
projector screen. The hand scanner reads the hand ( x1 ; x 2 ; x3 ;:::::::::::::::::::::::::: ; x d ) represent the d-
shape by recording the silhouette of the hand. In
combination with a side mirror and reflector, the dimensional feature vector in the database associated
optics produces two distinct images, one from the top with the claimed identity and Y =
and one from the side. This method is known as ( y1 ; y 2 ; y 3 ;::::::::::::::::; y d ) be the feature vector of
orthographic scanning. the hand whose identity has to be verified. The
verification is positive if the distance between X and Y
is less than a threshold value. Four distance metrics,
absolute, weighted absolute, Euclidean, and weighted
Euclidean, corresponding to the following four
equations were explored:
p

¦
K =1
Ň y k - x k Ň< £ A (1)

¦
K =1
(Ň y k - x k Ň) / ı k < £ WA (2)

¦
K =1
( yk - xk ) 2 < £ E (3)

Figure1. Typical hand geometry measurements [6] ¦

K =1
( y k - x k ) 2 / ı 2k < £ WE (4)

2 th
3.1. Enrollment Process Where ı k is the feature variance of the K features
and £ A , £ WA , £ E , £ WE , are the threshold value for
This process involves one of the following tasks:
each respective distance metrics.
• Add new user to the database.
• Update a current user’s feature vector.
There are two important types of errors associated
with biometric system.
3.1.1. False Accept Rate (FAR). The FAR is the
probability of wrongfully accepting and
imposter user.

3.1.2. False Reject Rate (FRR). The FRR is the

probability of wrongfully rejecting a genuine
user.
System decisions (i.e. Accept/Reject) is based on so –
called thresholds. By changing the threshold value,
Figure 2. Sandia Labs evaluation results for ID3D
one can produce various pair of (FAR/FRR)
System for authenticating identity based on hand
The quality of enrollment affects the false reject rate
geometry [7].
(FRR), especially during the first few weeks of use.

Authorized licensed use limited to: JAYPEE UNIVERSITY OF INFORMATION TECHNOLOGY. Downloaded on January 4, 2010 at 00:16 from IEEE Xplore. Restrictions apply.
 The FRR/FAR crossover error rates for Recognition Hand scanner interfaces emulating a card reader
System’s hand geometry scanner was 0.1% at the follows three common protocols viz. Wiegand,
manufacturer’s default threshold as documented by magnetic stripe and barcode. The hand scanner
Sandia Laboratories, in 1991 [7] (Figure 2). The connects to a third party access control terminal and
results were based on a two-try false reject rate. The act as a card-reader. When the user enters a PIN
one-try false reject rate was 0.2%. Field results taken number or presents a card, the scanner reads the ID
from actual sites with over 100,000 events confirm the filled from the wiegand data stream and recalls the
controlled test false reject rate. stored user’s template for verification. The third party
access control terminal receives the card number
4. Feature Extraction including the facility code and parity bits in wiegand
format and decide to admit or deny the user.
The hand geometry-based authentication system
relies on geometric invariants of a human hand. 5.2. Stand Alone Access Control
Typical features include length and width of the
fingers, aspect ratio of the palm or fingers, thickness A hand geometry scanner, when configured as an
of the hand, etc. [8]. To our knowledge, the existing unlocking device will give two control outputs viz. A
commercial systems do not take advantage of any non- lock output and an auxiliary output which control
geometric attributes of the hand, e.g., color of the skin. other devices. If there are no time or location
Figure 3 shows the 14 axes along which the various restrictions in the user’s database file, the lock output
features mentioned above have been measured. The gets energized on successful verification, coupled with
five pegs on the image serve as control points and a built in a timer for a preprogrammed period.
assist in choosing these axes. The hand is represented
as a vector of the measurements selected above. Since 5.3. Privacy Issues
the positions of the five pegs are fixed in the image, no
attempt is made to remove these pegs in the acquired Hand geometry based authentication protects
images. privacy of the users better than other biometrics but
the template can not be accurately reverse engineered
who identify users.

5.4. Operation by Disabled People

Even blind can enter PIN number on keys raised

and arranged in the typical ‘telephone’ format. Thus
hand scanner can be enabled to give an audio feedback
to guide the blind user.

5.5. Outdoor Conditions

Hand scanner electronics and scanning computers

can be made to work in extreme weather conditions.
Cold can hamper imaging optics when the moisture of
human body forms a vapor coating around the human
hand in contact with a cold surface. The vapor may
condense making a dense “halo” on the platen surface
which the scanner’s optics may interpret it as being
part of the land and is likely to reject it. Use of integral
platen and LCO heaters will prove to be helpful in
Figure.3. The fourteen axes along which feature values
sub-freezing environments.
are computed [9].
6. Application
5. Implementation Issues
There are some emerging applications of hand
5.1. Card Reader Emulation geometry based authentication systems. [6]:

Authorized licensed use limited to: JAYPEE UNIVERSITY OF INFORMATION TECHNOLOGY. Downloaded on January 4, 2010 at 00:16 from IEEE Xplore. Restrictions apply.
 6.1. Parking Lot Application Hand scanners have found broad applications in the
interactive kiosks. A host computer maintains user
Hand scanner used for access control in parking lots files and interacts with the user through a touch screen
will be a welcome by the users as they don’t need to monitor or keyboard. It checks that the user is valid or
carry cards. However, prevalent hand scanners are not, if yes, monitor displays a menu of choices from
designed for use with the right hand making it difficult which the user may select. The interactive kiosk
for left hand driven automobiles or by sports utility communicates with the user after ID entry and
vehicles and sports cars which may require platen verification. An automated border crossing is a
height to vary by as much as 0.9 meter. popular application of the interactive biometric kiosk.

6.2. Cash Vault Applications 7. A Brief review of work

A cash vault mantrap has two door and entry and an 7.1. Golfarelli et al. [10] formulated the identity
exit, and a hand scanner inside verifies the entrants. verification problem as a statistical pattern recognition
Number of people entering from the public side of problem and thus evolved a method for performance
mantrap is recorded by a personnel counter and a evaluation in biometric verification system. He
programmable logic controller reports the count to the derived explicit expression by defining optimum
hand scanner which must match the number of people Bayesian decision rule and established relation
using the scanner. between FRR (t) and FAR (t).

6.3. Dual Custody Applications 7.1.1. Advantages

In dual custody access control, two different people • Estimating the performance of biometric
must verify before the scanner sends an output. Dual systems without collecting large databases of
custody concept common in physical security, has biometric characteristics.
several variations and can be translated easily to hand • Dimensionality reduction applications.
scanner electronic access control. • Choosing the optimum features subset in a
biometric verification system.
6.4. Anti-pass back
7.1.2. Disadvantages
A common-access control function in which a user
is prevented from passing a card to an accomplice. • Partial discrepancies between the theoretical
Anti-pass back seems redundant for hand scanner and real performance
applications as its difficult at best to pass back a hand. • Less efficient model in presence of “noise” in
the normal distributions.
6.5. Time and Attendance • Less efficient model the real biometric
system performance.
The first hand geometry time and attendance
installations used hand scanners connected to a
printer or access control software to record users’
7.2. Jain et al. [11] describe the prototype of hand
geometry verification system. Features like length and
arrival and departure. This required manual sorting of
width of the fingers and thickness of the hand.
the event data, though some ‘‘computer savvy”
Equations (1) to (4) were used for the verifying
managers exported event data files to spreadsheet
whether the feature vector of hand matches with the
programs where they could sort and calculate the data.
feature vector stored in the database. When he
compared feature vector of same hand and different
6.6. Point of Sale Applications hand, he found a genuine matching score and imposter
matching score respectively. The result showed false
For the purpose of identity verification, point of
acceptance rate of 2% and false reject rate are 15%.
scale application is used, like debit systems are
becoming more common in our everyday lives as we
7.2.1. Advantages: The system performance could be
move toward being a cashless society.
significantly improved by
• Having habituated users.
6.7. Interactive Kiosks

Authorized licensed use limited to: JAYPEE UNIVERSITY OF INFORMATION TECHNOLOGY. Downloaded on January 4, 2010 at 00:16 from IEEE Xplore. Restrictions apply.
 • Better registration of hand geometry polynomials coefficients that do not change after a
measurements coordinate transformation.
7.2.2 Disadvantages 7.4.2. Disadvantage: More accurate 6th degree
• Not explore the effect of infra-red imaging on polynomials fit not higher and geometric invariants
the system performance. have not been tested.
• The existing feature set should be extended to
include 2-D feature of the hand, but not for, 7.5. Raul Somchez Reillo[14] used a database of 200
use deformable modules for a robust images from 20 people and implemented hand
representation of the hand. geometry verification system. They used Gaussian
• A more extensive system performance on Mixture Modeling, a statistical pattern recognition
larger database collected over a period of technique, with morphological features extracted from
time is necessary. images. This technique, usually applied to speaker
identification, can be applied to hand geometry
7.3. Jain and Duta [12] proposed an authentication biometrics and get improved results. Out of their
method based on deformable matching of hand shapes. database, they reported 97% success in identification
A more reliable criterion of shape distance was used and error values below 10% in verification.
for verification which is computed by aligning the
hand shape before extracting a feature set. This 7.5.1 Advantages: Previous approaches to hand
approach found 96.5% genuine accept rate against a biometrics have been developed by author using
2% false accept rate. This performance further different metrics [15], [16]. In order to obtain better
improved by learning an enrollment template shape for result, the authors determined to study the possibility
each user. of using GMMs for the Recognition Block.

7.3.1. Advantages: Classification system is very 7.5.2 Disadvantages:

accurate i.e., for a threshold T=1.80 the genuine accept • Less better figures modify.
rate is 96.5% for a 2% false accept rate. • Test the system with small database not for
large database.
7.3.2. Disadvantages: Although the 2% false accept
rate may seem high, in practice, it is much smaller, 8. Conclusion
since a user of the system does not know the identity
of which user he can assume such that their hand This paper presents an overview of biometric hand
shapes match. geometry recognition. The major advantage of hand
geometry verification system is to ease of image
7.4. Oden, Ercil and Buke [13] Implicit polynomials acquisition compared to fingerprint and iris image
were used for identification and verification in the acquisition. The systems are widely implemented for
system proposed. When they combined their method their ease of use, public acceptance, and integration
with geometric feature, the result was success 95% for capabilities. Hand geometry can be used for
identification and 99% for verification. Different identification by using implicit polynomials and
results were obtained when combining the features of Gaussian Mixture Modeling (GMM). The limitation of
both methods as in table II. hand geometry can be reduced by designing robust
algorithms and including extra features.
Table 2. Comparison of result using different method

Geom Geom IP IP Geom.+ REFERENCES

etric(1 etric( (15) (12) IP(16) [1] A. Jain, L. Hong, S. Pankanti, and R. Bolle, \On-line identity-
6) 12) authentication system using fingerprints," Proceedings of
Identificati 80% 88% 73% 85% 95% IEEE, vol. 85, pp. 1365 -1388, September 1997.
[2] R. Zunkel, \Hand Geometry Based Authentication" in
on
\Biometrics: Personal Identification in Networked Society",
Verificatio 89% 98% 84% 90% 99% A. Jain, R. Bolle, and S. Pankanti (Eds.), Kluwer Academic
n Publishers.
[3] James Wayman, ed., “National Biometric Test Center
7.4.1. Advantages: The main advantages of implicit Collected Works,” San Jose State University, August 2000.
http:// [Link]/biometrics/[Link].
polynomials for recognition are the existence of [4] James Holmes, Larry Wright, and Russell Maxwell,” A
algebraic invariants, which are function of the Performance Evaluation of Biometric Identification Devices,”

Authorized licensed use limited to: JAYPEE UNIVERSITY OF INFORMATION TECHNOLOGY. Downloaded on January 4, 2010 at 00:16 from IEEE Xplore. Restrictions apply.
 Sandia National Laboratories, June1991.
[Link]
1991/ [Link].
[5] “Information Technology - Hand Geometry Format for Data
Interchange,” ANSI INCITS 396-2005.
[Link]
[6] Jain, A.K., Bolle, R., and Pankanti, S. (Eds.): ‘Biometrics:
personal identification in networked society’ (Kluwer
Academic Publishers, 1999).
[7] A Performance Evaluation of Riometnc Identification
Devices, Technical Report SAND9 1-0276, UC-906, Sandia
National Laboratories, Albuquerque, NM and Livermore, CA
for the United States Department of Energy under Contract
DE-AC04-76DP00789, 1991.
[8] HaSIS - A Hand Shape Identi_cation System",
[Link]
[9] Ross, A.: ‘A prototype hand geometry-based verification
system’. MS project, Department of Computer Science and
Engineering, Michigan State University, USA, 1999.
[10] Golfarelli, M., Maio, D., and Maltoni, D.: ‘On the error-reject
trade-off in biometric verification systems’, IEEE Trans.
Pattern Anal. Mach. Intell., 1997, 19, (7), pp. 786–796.
[11] Jain, A.K., Ross, A., and Pankanti, S.: ‘A prototype hand
geometrybased verification system’. Proc. 2nd Int. Conf. on
Audio- and videobased personal authentication (AVBPA),
Washington, USA, March 1999, pp. 166–171.
[12] Jain, A.K., and Duta, N.: ‘Deformable matching of hand
shapes for verification’. Proc. IEEE Int. Conf. on Image
processing, Kobe, Japan, October 1999, pages 5.
[13] C. Oden, A. Ercil, V.T. Yildiz, H. Kirmizitas, B. Biuke,
"Hand recognition using implicit polynomials and geometric
features," 2001.
[14] R.S. Reillo, C.S. Avila, "Biometric identification through
hand geometry measurements," IEEE Trans. Pattern Analysis
and Machine Intelligence, vol. 22, no. 10, pp. 1168-1171,
Oct. 2000.
[15] Hand Geometry verification and smart [Link] Sanchez-
Rcillo and Ana Gonzalez-Marcos. E.T.S.I.
Telecommunication. IEEE AES system Magazine, February
2000.
[16] Sanchez-Reillo R., Sanchez-Avila C., Gonzalez-Marcos A.
“Multiresolution Analysis and Geometric Measure for
Biometric Identification”. CQRE Secure Networking.
Diisseldorf, Germany, 1999.

AuthorizedView
licensed use
publication statslimited to: JAYPEE UNIVERSITY OF INFORMATION TECHNOLOGY. Downloaded on January 4, 2010 at 00:16 from IEEE Xplore. Restrictions apply.