See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/224377208

Fingerprint based authentication application using visual cryptography

methods (Improved ID card)

Conference Paper · December 2008

DOI: 10.1109/TENCON.2008.4766425 · Source: IEEE Xplore

CITATIONS READS

33 1,095

4 authors, including:

Rao Y V S
University of Hyderabad
30 PUBLICATIONS 132 CITATIONS

SEE PROFILE

All content following this page was uploaded by Rao Y V S on 24 November 2014.

The user has requested enhancement of the downloaded file.

 Fingerprint based authentication application
using visual cryptography methods
(Improved ID card)
Mr. Y.V. Subba Rao, Ms. Yulia Sukonkina
Department of Computer and Information Sciences, University of Hyderabad, Gachibowli, Hyderabad, India
[email protected], [email protected]

Dr. Chakravarthy Bhagwati, Mr. Umesh Kumar Singh

Department of Computer and Information Sciences, University of Hyderabad, Gachibowli, Hyderabad, India
[email protected], [email protected]

Keywords: Visual Cryptography, Biometrics, Fingerprints, Authentication.

Abstract: The main idea of this paper is to efficiently apply the Visual Cryptography (VC) techniques onto the
area of authentication using fingerprints. We present an alternative approach of using the
fingerprints, attempting to solve two major problems related to fingerprint based automatic access
control systems which are falsification and the costly maintenance of the large fingerprint database.
In the proposed application we divide an input fingerprint image into two shares with the help of the
basic VC techniques, keeping one with the participant in the form of ID card and saving the other one
in the database. This share kept in the database will be the same for all of the participants. While
accessing, we will stack the corresponding shares together and compare the obtained image with the
provided fresh fingerprint using any modern minutia extraction algorithm.

1 INTRODUCTION administrative are simply superimposed to obtain the

secret fingerprint image. From this image now the
During the studies of multiple enhanced techniques minutiae of the finger can be extracted.
in VC we found an interesting area to implement the On the next step the application requests the
existing ideas. Being aware of the certain issues participant to render the new fingerprint. This image
concerning authentication control such as spoofing is processed by a system and the minutiae are
and "buddy punching" we are trying to introduce an extracted and compared with the minutiae of the
application based on biometrics and an ID card in secret fingerprint image obtained earlier.
order to improve the security and cost of the overall Minutiae extraction and matching can be done with
admission process. the help of any fingerprint scanner. The authentication
Our approach is to enable the completely succeeds only in case if minutiae are matching.
synchronised combination of VC and the fingerprint Overcoming the problems stated above our
scanner. application does not reintroduce the problems
Considering fingerprint as a secret image we associated with the non-biometric possession-based
distribute it among the two shares following one of authentication techniques: that is, forgotten or
the advanced VC methods. First share is being a guessed password, lost or stolen key, etc.
random image of the administrative database, whereas This paper is organized as following: Section 2
the second is the photograph on the ID card of the introduces a term biometrics; Section 3 and 4
participant. We fix the same administrative share summarizes the history of fingerprints and
valid for all of the participants, hence utilising authentication process; Problems with the existing
economically the size of the administrative database. fingerprint readers are discussed in section 5;
Being unbreakable VC assures the security of the Introduction to Visual Cryptography and it’s
both shares of the secret stored with the participants. alternative presented in section 6 and 7; Section 8
As long as VC does not require any computation describes the proposed application; Conclusions and
during the decoding process, once the participant future task are given in section 9.
inserts his ID card and the corresponding share is
extracted from it both shares participant’s and
 Table 1: Comparison of Biometric Technologies.

Biometrics Universality Uniqueness Permanence Collectability Performance Acceptability Circumvention

Face High Low Medium High Low High Low
Fingerprint Medium High High Medium High Medium High
Hand
Medium Medium Medium High Medium Medium Medium
Geometry
Hand Vein Medium Medium Medium Medium Medium Medium High
Iris High High High Medium High Low High
Retinal
High High Medium Low High Low High
Scan
Signature Low Low Low High Low High Low
Voice Print Medium Low Low Medium Low High Low
F. Thermo-
High High Low High Medium High High
grams

2 BIOMETRICS Modern fingerprint techniques were initiated in 1684 by

English plant morphologist Nehemiah Grew. Starting from
Biometrics is the detailed measurements of the human 1809, Thomas Bewick began to use his fingerprint as his
body. Biometrics deals with automated methods of trademark, which is believed to be one of the most important
identifying a person or verifying the identity of a person contributions in the early scientific study of fingerprint
based on physiological or behavioural characteristics (S. identification (A. Jain et al., 1997).
Bistarelli et al., 2003). The most popular characteristics are Later fingerprint identification systems were involved in
fingerprints and facial features. A brief comparison of nine criminal identification process. Nowadays these systems are
biometric techniques made by A. Jain et al. in 1997 is widely used in multiple civil areas, such as in prevention of
provided in Table 1. multiple enrollments in an electoral, welfare, custom control,
employee attendance logging, security desk in banks,
3 FINGERPRINTS HISTORY security installations, visitor verification, member
verification in clubs, member organizations etc.
Definition: Fingerprints are graphical flow-like ridges
present on human fingers (see Fig. 1). Their formations 4 FINGERPRINT AUTHENTICATION PROCESS
depend on the initial conditions of the embryonic mesoderm
from which they develop (A. Jain et al., 1997). In 2003 S. Bistarelli, G. Boffi and F. Rossi. In their paper
“Computer Algebra for Fingerprint Matching” proofed that
during fingerprint identification, it is desirable to have, at the
least, a two stage search. The first stage makes use of global
fingerprint characteristics while the second stage is the
minutiae matcher (point pattern matching).
Authors state that typically, automatic fingerprint
identification and authentication systems rely on representing
the two most prominent structures: ridge endings and ridge
bifurcations ((a) and (b) in Fig. 2 correspondingly). These
two structures are background-foreground duals of each
other and pressure variations could convert one type of
structure into the other. Therefore, many common
representation schemes do not distinguish between ridge
Fig. 1: A fingerprint classification schema of six categories: (a) arch, (b) endings and bifurcations. Both the structures are treated
tented arch, (c) right loop, (d) left loop, (e) whorl, and (f) twin loop; critical equivalently and are collectively called minutiae. The
points in a fingerprint, called core and delta, are marked on image (c).
simplest of the minutiae-based representations constitute a
list of points defined by their spatial coordinates with respect
to a fixed image-centric coordinate system. Typically,
though, these minimal minutiae-based representations are
further enhanced by tagging each minutia (or each algorithms is infallible. However, the rates of false negatives
combination of minutiae subset, e.g., pairs, triplets) with and false positives have markedly improved. One of the
additional features. For instance, each minutia could be significant problems with fingerprint readers, for instance, is
associated with the orientation of the ridge at that minutia; or that they couldn't distinguish between an actual fingerprint
each pair of the minutiae could be associated with the ridge and the image of one. Some fingerprint scanners can be
count: the number of ridges visited during the linear traversal spoofed with nothing more than a breath of hot air, which
between the two minutiae. reactivates latent prints left on the scanner.
The solution of these problems is provided by the higher-
end fingerprint readers, which are expansive. The latest
fingerprint readers are incorporating more advanced features,
such as making sure the finger is a certain temperature, a
pulse and pressure. Such sophistication, however, has its
drawbacks. Authorized users may find themselves locked out
even when the devices are working properly, because of tiny
Fig. 2: An example of ridge endings b) and bifurcations a). changes, due to accidents or injuries, which can change a
biometrics profile, rendering it effectively obsolete.
A feature extractor finds the ridge endings and ridge Biometric authorization techniques are no longer so
bifurcations from the input fingerprint images. If ridges can leading edge that they are difficult to marry with traditional
be perfectly located in an input fingerprint image, then security safeguards. A strong authentication system is what
minutiae extraction is just a trivial task of extracting singular we want to focus on and biometrics can be a part of it, but
points in a thinned ridge map. However, in practice, it is not the user should still have to memorize something or have a
always possible to obtain a perfect ridge map. token, and we need to make sure that polices and the
Finely, the researches have listed the stages required for management structure relating to it are firmly in place.
the biometric authentication as following: Such a solution we find using Visual Cryptography
Capture: A raw biometric sample is captured by a methods along with Fingerprint Based Access Control
sensing device, such as a fingerprint scanner or video System.
camera.
Process: The distinguishing characteristics are extracted 6 INTRODUCTION TO VISUAL CRYPTOGRAPHY
from the raw biometric sample and converted into a
processed biometric identifier record (sometimes called In 1995 Naor and Shamir have suggested for the first
biometric sample or biometric template). time to solve the secret sharing problem by the means of new
Enrol: The biometric template is stored in a storage cryptographic structure called Visual Cryptography (VC). In
medium for comparison during an authentication phase. the proposed approach the secret is divided into two shares,
Notice that the original biometric sample cannot be which are printed onto the two transparencies (shares) and
reconstructed from this identifier. given to the participants. Only these two participants who
Verification: In this mode (\1 to 1 matching"), a newly possess the transparencies can reconstruct the secret by
captured/processed biometric sample taken for instance superposition of shares. One can not recover a secret without
during a login, is compared against a previously enrolled the other one.
sample to address the question \Are you the person you claim In the visual threshold scheme, the shares are images
to be?". represented on transparencies consisting of black and white
Identification: In this mode (\1 to N matching"), the (transparent, actually) pixels. The visual systems perform a
individual does not claim an identity. The individual presents Boolean OR operation, which is easy to visualise using the
a biometric sample and the system tries to identify the (2, 2) Visual Threshold Scheme shown in Fig. 3.
individual from a database of stored biometric samples.

5 PROBLEMS WITH THE EXISTING FINGERPRINT

READERS

The performance of currently available minutiae

extraction algorithms depends heavily on the quality of input
fingerprint images. Due to a number of factors (aberrant
formations of epidermal ridges of fingerprints, postnatal
marks, occupational marks, problems with acquisition
devices, humidity and ageing of the finger etc.) fingerprint
images may not always have well-defined ridge structures
(A. Jain et al., 1997). For example, paper of A. Jain et al.
shows the development of a new algorithm for fingerprint Fig. 3: (2, 2) Visual Threshold Scheme.
matching able to take care of translations, rotations and other
transformations. The algorithm is also able to match Later in 2001 the engineers from Taiwan in their paper
fingerprints when some minutiae are missed or when some (C.S. Tsai et al., 2001) have claimed that during the encoding
unreal minutia is detected. But unfortunately none of these process shares are generated in such a way that they contain
random dots to create a chaos for preventing intruders of
random guesswork. This leads to a trouble of stacking the
dots with each other. However the problems of disorderliness
and unstackability can be eliminated using a multiple secret
sharing method, which adds a small amount of computation.
They propose two algorithms for secret sharing and secret
recovery derived from the least significant bit substitution
method. They convert the secret into many bit planes and Fig. 4: Bit plane extraction.
modify the so called cover images based on these bit planes.
Each modified cover image is called stego-image. Permuted bits of the bit plane should satisfy the
Thus generation of shares could be also done using so following equation and reproduce si:
called cover images. In the next section we will explain this
alternative algorithm for creating shares used by Tsai, Chang si= C’xi[perx(k)+16] XOR C’yj[pery(k)+16] (2)
and Chen.
where 1 ≤ k ≤ t, C’xi , C’yj are the i-th and j-th bit planes
7 ALTERNATIVE WAY OF SHARE GENERATION of the C’x and C’y modified cover images correspondently.
INTRODUCED BY TSAI ET AL. Hence during recovery of the secret bits we have to
obtain both modified cover images, scan them to find and
Assume there are two participants {n1} and {n2} and two extract the bit planes used for the secret sharing. Then the
digital grayscale cover images C1 and C2. Suppose S = s1 s2 Equation 2 has to be applied on each bit of the bit plane one
s3 … st is the selected secret image. Here si is the i-th bit of S after the other stepwise reconstructing the secret.
and the bit length of S is t, where: Next section explains the idea of embedding these secret
sharing and recovery algorithms into the Automated
t ≤ min (C1 size, C2 size) – 16 (1) Fingerprint Authentication System.

If this condition is satisfied, next goal is to find two bit 8 PROPOSED APPLICATION
planes in cover images and apply them to share the
secret.The bit planes are generated as depicted in Fig. 4. The Assume we require assembling the Automated
maximum number of bit planes in one cover image equals to Fingerprint Identification System on the entrance of the
eight (i.e. 8 bits in a byte). secret building. The administrator will collect the
According to Fig. 4 Ci is i-th bit plane of cover image C. fingerprints of the people eligible to enter the building. She
Ci [j] is the j-th bit of the i-th bit plane of C. will consider each fingerprint as a secret image. Randomly a
Each bit plane can be applied to share two secret unique dummy share will be created and saved securely in
messages with two different participants. This indicates an the database. The shares of the participants will be created
economical utilization of bit planes, implying that a small from this dummy share and the fingerprint images with the
number of bit planes may keep a great number of secrets to help of standard VC methods or any other VC sharing
be shared. algorithms. For the better security we will embed the share
Each first 16 bits of the bit plane are used for the of the participant into her cover image, which is a
identification purpose and help to manipulate the bit planes photograph on her ID card. There are multiple papers
efficiently. Here ai is the 1st eight bits of each bit plane Ci and proposed which are referring to how to use the cover images
bi is the next eight bits. Initially ai and bi are set to be zero, during the secret sharing process. For example, in section 7
which indicates that Ci have not yet been applied to share we have described one of them.
any secret. There are nodes identification numbers N1 and The participant share will be permuted using the random
N2, which are set to be any number greater then zero.
permutation, which will be unique for each participant. For
To find two bit planes all indicators in both C should be
generating the random permutation the Random Number
scanned to find min value of i, s.t. ai is not equal to zero and
bi is equal to zero. This indicates that this bit plane is Generator is required. On the other hand we “recall that
occupied to share secret with one participant, its’ contents computers can’t create real random numbers, just streams of
was used and can’t be modified any more. If this condition numbers that appear random to the outside observer” (P.
doesn’t exits in C then value of i is set to be -1 and the Johnson et al., 2000). But random numbers are often
selection process continues. initialized using a computer's real time clock as the seed.
Note that each secret bit si corresponds with one bit of These functions may provide enough randomness for the
the selected bit plane. certain tasks. Much higher quality random number sources
After selecting the bit planes from the cover images, two are available on most operating systems; for example
random permutations are generated. The size of these /dev/random on various BSD flavors, Linux, Mac OS X,
permutations is the same size as the secret. The bits of the bit IRIX, and Solaris, or CryptGenRandom for Microsoft
plane starting from 16th are permuted with the original secret Windows. Random number generators are very useful as
according to the generated permutation. debugging is facilitated by the ability to run the same
sequence of random numbers again by starting from the
same seed. They are also used in cryptography so long as the
seed is secret. Sender and receiver can generate the same set
of numbers automatically to use as keys.
 In our application the administrative database will store • There is no need for the administrator to maintain a
the integer seed which will be used to generate the set of the large data base of the fingerprints.
required random permutations. Thus we will avoid the
problem of storing large sequences of random numbers in We overcome the problems stated above without
our database. Moreover the shares of the participants will be reintroducing the problems associated with the non-biometric
stored in their ID cards and the administrator will have to authentication techniques. These security problems inherent
maintain the database where only the dummy share and the in the knowledge- and possession-based techniques: that is, a
integer seed will be stored. password can be forgotten or guessed, a key may be lost or
For entrance the participant will provide her share in the stolen, and both can be shared (D. Maltoni, 2003).
form of ID card, which will be met by the system. Using the We affirm that the cost difference of the techniques
reverse permutation, dummy share and applying the VC involved in our application is negligible as in compare with
techniques system will generate the image of the fingerprint techniques used by the existing Fingerprint Based
provided by the participant during the registration. This Authentication Systems. Moreover it is luckily slightly
image will be compared with the newly provided fingerprint reduced. The statistical cost analysis is our future task.
using any of the modern minutiae extraction algorithms. If
the results of the comparison will match, entrance will be REFERENCES
allowed, as shown in Fig. 5.
[1] Noar M., Shamir A., 1995. Visual cryptography. Advances in
Cryptography. Eurocrypt’94, Lecture Notes in Computer Science, vol.
950, Springer-Verlag. 1 – 12.
[2] Tsai C.S., Chang C.C., Chen T.S., 2001. Sharing multiple secrets in
digital images. Department of Computer Science and Information
Engineering, Taiwan. 1 – 8.
[3] Subba Rao Y.V., 2007. Presentation on Visual Cryptography and Its
Applications. Department of Computer and Information Sciences,
University of Hyderabad, India. 1 – 42.
[4] Jain A., Hong L., Pankanti S., Bolle R., 1997. An Identity Authentication
System Using Fingerprints. Department of Computer Science, Michigan
State University, USA. 1- 66.
[5] Bistarelli S., Boffi G., Rossi F., 2003. Computer Algebra for Fingerprint
Matching. Universita “G. d'Annunzio”, Dipartimento di Scienze,
Pescara, Italy. 1 – 10.
[6] Davide Maltoni 2003. Handbook of Fingerprint Recognition. 1 – 366.
[7] Swarm Development Group, Johnson P., Lancaster A., 2000. Swarm
User Guide. 1 – 176.
Fig. 5: Registration and authentication process. [8] Stinson D.R., Tavares S., 2000. The Pseudo-Random Number. Selected
Areas in Cryptography. 7th Annual International Workshop, Waterloo,
Ontario, Canada. 100 - 101.
[9] http://www.ravirajtech.com/biometrics_news.html
9 CONCLUSIONS AND FUTURE PLANS [10] http://www.bioenable.com/usb_fingerprint_time_attendance_india.htm
[11]http://www.webopedia.com/DidYouKnow/Computer_Science/2006/bio
metrics_security.asp
Looking back to the existing problems with fingerprint [12]http://en.wikipedia.org/wiki/Automated_Fingerprint_Identification_Syst
readers we achieve the following results: em
[13] http://en.wikipedia.org/wiki/Random_number_generation
• There is no more problem with the falsification of the
finger, because the entrance will succeed only in case if the
participant will provide the ID card.

View publication stats