Documents
Resources
Learning Center
Upload
Plans & pricing Sign in
Sign Out

Biometrics In Fingerprint Recognition Using Distance Vector Method

VIEWS: 174 PAGES: 6

									                                                         (IJCSIS) International Journal of Computer Science and Information Security,
                                                         Vol. 9, No. 8, August 2011



      Biometrics in Fingerprint Recognition using Distance Vector Method

                     Sumeet Dhawan                                                             Amit Makkar
            Dept. of Information Technology                                  Dept. of Computer Science and Engineering
           Adesh Institute of Engineering & Technology                       Adesh Institute of Engineering & Technology
                     Faridkot, India                                                        Faridkot, India
                             .                                                                     .


Abstract-A lot of fingerprint features have been defined                feature is matched with the Distance Vector of each
for the representation of the fingerprint. Among the                    fingerprint in the template database. This process increases
features, ridge ending and ridge bifurcation are the most               the reliability of the fingerprint recognition task. In the first
widely used ones. These features are known as minutiae                  stages, image normalization and orientation of the ridges are
points. Utilizing more information other than minutiae is               estimated.
much helpful for large scale fingerprint recognition
application. By considering some predefined features                            II.   FINGERPRNT RECOGNITION AND ITS
one can derive new features for the verification of the                                    MODERN USUAGE
fingerprint. In this thesis, we proposed a new feature for                  In security system a person can be identified based on the
fingerprint images. This new feature is named as                        three approaches: something you have, such as a key;
Distance Vector. A Distance Vector counts the minutiae                  something you know, such as a password and something you
points in each row of a particular fingerprint image. A                 are, for example any biometrical data. In an increasingly
Distance Vector is associated with every fingerprint in                 digital world, reliable personal authentication has become an
the database. At the time of enrollment this feature is                 important human computer interface activity. National
stored with the concerned fingerprint and at the time of                security, e-commerce, and access to computer networks are
matching this feature is matched with the Distance                      some examples where establishing a person’s identity is
Vector of each fingerprint in the template database. This               vital. Methods that use the concept of key or passwords are
process increases the reliability of the fingerprint                    ubiquitous; such methods are not very secure. Tokens such
recognition task. In the first stages, image normalization              as badges and access cards may be shared or stolen.
and orientation of the ridges are estimated.                            Passwords and PIN numbers may be stolen electronically.
                                                                        Furthermore, they cannot differentiate between authorized
Keywords- Biometrics; Optical Character Recognition;                    user and a person having access to the tokens or knowledge.
                                                                        In first part of this paper we have tried to explain what is
DIP-Digital Image Processing; Minutiae; Fingerprint
                                                                        biometrics and its modern day usage. Later it is followed by
Identification Recognition.
                                                                        the classification of biometrics based upon physiological or
                     I.    INTRODUCTION                                 behavioral characteristics. After describing both these types
                                                                        in detail, focus is stressed upon how to explain fingerprint
    Biometrics refers to the automatic identification of a              identification as a biometric device capable of sustaining
person based on his or her physiological or behavioral                  modern day equipments, based on this novel device. In the
characteristics. This identification method is preferred over           next section minutiae of small detail extraction is suitably
traditional methods involving passwords and PINs. PINs and              studied so as to set benchmark identification which is almost
passwords may be forgotten, and token based identification              a revolution in its usage channel. Finger Image enhancement
methods such as passports and driver’s licenses may be                  is also suitably dealt with in this passage of the paper. In the
forged, stolen, or lost. Thus, biometric systems of                     heart of our methodology, proposed Approach: Matching
identification are enjoying a new interest. Various types of            through Distance Vector is studied. Finally, this paper is
biometric systems are being used for real-time                          concluded.
identification.The most popular are based on face recognition
and fingerprint matching; however, other biometric systems
use iris and retinal scans, speech, facial feature comparisons
and facial thermo grams, and hand geometry. In last section
experimental results have been discussed and finally
conclusion has been given.
    Distance Vector: A Distance Vector counts the minutiae
points in each row of a particular fingerprint image. A
Distance Vector is associated with every fingerprint in the
database. At the time of enrollment this feature is stored with
the concerned fingerprint and at the time of matching this                            Figure 1. Basic Approach Used in Biometric.




                                                                  171                                http://sites.google.com/site/ijcsis/
                                                                                                     ISSN 1947-5500
                                                         (IJCSIS) International Journal of Computer Science and Information Security,
                                                         Vol. 9, No. 8, August 2011



               III.    WHAT IS BIOMETRICS?                              2.   Behavioral are related to the behavior of a person.
     The term biometrics is derived from the Greek words bio                 Examples include, but are not limited to typing rhythm,
(life) and metric (to measure). Among the first known                        gait, and voice. Some researchers have coined the term
examples of practiced biometrics was a form of member                        behavioral biometrics for this class of biometrics.
printing used in China in the 14th century, as reported by the              So it is imperative to say that biometrics is the science of
Portuguese historian Joao De Barros. The Chinese merchants              verifying the identity of an individual through physiological
were stamping children’s palm and footprints on paper with              measurements or behavioral traits. Since biometric identifiers
ink to distinguish the babies from one another. In the 1890s,           are associated permanently with the user, they are more
an anthropologist and police desk clerk in Paris named                  reliable than token or knowledge based authentication
Alphonse Bertillon sought to fix the problem of identifying             methods. Biometrics refers to methods for uniquely
convicted criminals and turned biometrics into a distinct field         recognizing humans based upon one or more intrinsic
of study. The problem of resolving the identity of a person             physical or behavioral traits. In particular, biometrics is used
can be categorized into two fundamentally distinct types of             as a form of identity access management and control.
problems with different inherent complexities i.e.
verification and recognition (more popularly known as                       While biometrics is primarily considered as application
identification. Verification (authentication) refers to the             of pattern recognition techniques, it has several outstanding
problem of confirming or denying a person's claimed identity            differences from conventional classification problems as
(Am I who I claim I am?). Identification (Who am I?) refers             enumerated below:
to the problem of establishing a subject's identity - either            1.   In a conventional pattern classification problem such as
from a set of already known identities (closed identification                Optical Character Recognition (OCR) recognition, the
problem) or otherwise (open identification problem).                         number of patterns to classify is small (A-Z) compared
                                                                             to the number of samples available for each class.
                                                                             However in case of biometric recognition, the number of
                                                                             classes is as large as the set of individuals in the
        IV.   CLASSIFICATION OF BIOMETRICS                                   database. Moreover, it is very common that only a single
                  CLASSIFCATION                                              template is registered per user.
    Biometrics refers to the automatic identification of a
person based on his or her physiological or behavioral
characteristics.This identification method is preferred over
traditional methods involving passwords and PINs. The most
popular are based on face recognition and fingerprint
matching; however, other biometric systems use iris and
retinal scans, speech, facial feature comparisons and facial
thermo grams, and hand geometry. Methods that use the
concept of key or passwords are ubiquitous; such methods
are not very secure. Tokens such as badges and access cards
may be shared or stolen. Passwords and PIN numbers may
be stolen electronically. Furthermore, they cannot                                      Figure 2. Optical Character Recognizer.
differentiate between authorized user and a person having
access to the tokens or knowledge. By replacing PINs,                   2.   The primary task in biometric recognition is that of
biometric techniques can potentially prevent unauthorized                    choosing a proper feature representation. Once the
access to or fraudulent use of the following:                                features are carefully chosen, the act of performing
                                                                             verification is fairly straightforward and commonly
1.   ATMs                                                                    employs simple metrics such as Euclidean Distance.
2.   Cellular phones                                                         Hence the most challenging aspects of biometric
                                                                             identification involve signal and image processing for
3.   Smart cards                                                             feature extraction.
4.   Desktop PCs                                                        3.   Biometric templates represent personally identifiable
5.   Workstations                                                            information of individuals, security and privacy of the
                                                                             data is of particular importance unlike other applications
6.   Computer networks                                                       of pattern recognition.
Hence biometric characteristics can be divided in two main              4.   Modalities such as fingerprints, where the template is
classes:                                                                     expressed as an unordered point set (minutiae) do not
1.   Physiological are related to the shape of the body.                     fall under the category of traditional multi-
     Examples include, but are not limited to fingerprint, face              variant/vectorial features commonly used in pattern
     recognition, DNA, hand and palm geometry, iris                          recognition.
     recognition, which has largely replaced retina, and
     odor/scent.



                                                                  172                                http://sites.google.com/site/ijcsis/
                                                                                                     ISSN 1947-5500
                                                           (IJCSIS) International Journal of Computer Science and Information Security,
                                                           Vol. 9, No. 8, August 2011



             V. FINGERPRINT AS A BIOMETRIC
       Fingerprint identification is the process of comparing
questioned and known friction skin ridge impressions fingers
or even toes to determine if the impressions are from the
same finger. Fingerprints were accepted formally as valid
personal identifier in the early twentieth century and have
since then become a de-facto authentication technique in
law-enforcement agencies worldwide. The flexibility of
friction ridge skin means that no two finger prints are ever
exactly alike even two impressions recorded immediately
after each other. Fingerprint identification occurs when an
expert determines that two friction ridge impressions
originated from the same finger to the exclusion of all others.
Fingerprints have several advantages over other biometrics,                    Figure 3. (a) Local Features: Minutiae (b) Global Features: Core and
such as the following:                                                                                        Delta.

1.   High universality: A large majority of the human
     population has legible fingerprints and can therefore be                        VI. FINGERPRINT IMAGE ENHANCEMENT
     easily authenticated. This exceeds the extent of the
                                                                              The performance of a fingerprint feature extraction and
     population who possess passports, ID cards or any other
                                                                          matching algorithm depend heavily upon the quality of the
     form of tokens.
                                                                          input fingerprint image. While the ’quality’ of a fingerprint
2.   High distinctiveness: Even identical twins who share the             image cannot be objectively measured, it roughly
     same DNA have been shown to have different                           corresponds to the clarity of the ridge structure in the
     fingerprints, since the ridge structure on the finger is not         fingerprint image. A ’good’ quality fingerprint image has
     encoded in the genes of an individual. Thus, fingerprints            high contrast and well defined ridges and valleys. A ’poor’
     represent a stronger authentication mechanism than                   quality fingerprint is marked by low contrast and ill-defined
     DNA. Furthermore, there has been no evidence of                      boundaries between the ridges and valleys. There are several
     identical fingerprints in more than a century of forensic            reasons that may degrade the quality of the fingerprint
     practice.                                                            image.
3.   High permanence: The ridge patterns on the surface of                1.     The ridges are broken by presence of creases, bruises or
     the finger are formed in the womb and remain invariant                      wounds on the fingerprint surface.
     until death except in the case of severe burns or deep
                                                                          2.     Excessively dry fingers lead to fragmented ridges.
     physical injuries.
                                                                                 Sweaty fingerprints lead to bridging between successive
4.   Easy collecting ability: The process of collecting                          ridges.
     fingerprints has become very easy with the advent of
                                                                              The quality of fingerprint encountered during
     online sensors. These sensors are capable of capturing
     high resolution images of the finger surface within a                verification varies over a wide range. It is estimated that
     matter of seconds.                                                   roughly 10% of the fingerprint encountered during
                                                                          verification can be classified as ’poor’. Poor quality
5.   High performance: Fingerprints remain one of the most                fingerprints lead to generation of spurious minutiae. In
     accurate biometric modalities available to date with                 smudgy regions, genuine minutiae may also be lost, the net
     jointly optimal FAR (false accept rate) and FRR (false               effect of both leading to loss in accuracy of the matcher.
     reject rate). Forensic systems are currently capable of
     achieving FAR of less than 10-4.                                          Most of the time, fingerprints show non-stationary
                                                                          nature. Due to the non-stationary nature of the fingerprint
6.   Wide acceptability: While a minority of the user                     image, using a single filter that operates on the entire image
     population is reluctant to give their fingerprints due to            is not effective. Instead, the filter parameters have to be
     the association with criminal and forensic fingerprint               adapted to enhance the local ridge structure. A majority of
     databases, it is by far the most widely used modality for            the existing techniques are based on the use of contextual
     biometric authentication.                                            filters whose parameters depend on the local ridge frequency
    The fingerprint surface is made up of a system of ridges              and orientation. Human experts routinely use context when
and valleys that serve as friction surface when we are                    identifying. Fingerprint image enhancement is a crucial step
gripping the objects. The surface exhibits very rich structural           in an automatic fingerprint recognition system. Quite a lot of
information when examined as an image. The fingerprint                    approaches for filtering the fingerprint image have been
images can be represented by both global as well as local                 suggested. Most of them perform oriented band pass
features as depicted in Figure 3.                                         filtering. However, such filters, for example, Gabor filters
                                                                          produced good results. Summarily, the performance of a
                                                                          fingerprint feature extraction and matching algorithms




                                                                    173                                   http://sites.google.com/site/ijcsis/
                                                                                                          ISSN 1947-5500
                                                                    (IJCSIS) International Journal of Computer Science and Information Security,
                                                                    Vol. 9, No. 8, August 2011



depend heavily upon the quality of the input fingerprint
image.
    Minutiae or small details mark the regions of local
discontinuity within a fingerprint image. These are locations
where the ridge comes to an end (type: ridge ending) or
branches into two (type: bifurcation). Other forms of the
minutiae includes a very short ridge (type: ridge dot), or a
closed loop (type: enclosure). The different types of minutiae
are illustrated in figure given below which is repeated here
for clarity. There are more than 18 different types of
minutiae among which ridge bifurcations and endings are the                             Figure 5. Position of the neighboring pixels for minutiae detection.
most widely used. Other minutiae type may simply be
expressed as multiple ridge endings of bifurcations. For                                  After the CN for a ridge pixel has been computed, the
instance, a ridge dot may be represented by two opposing                           pixel can then be classified according to the property of its
ridge endings placed at either extremity. Even this                                CN value. As shown in figure, a ridge pixel with a CN of 0
simplification is redundant since many matching algorithms                         corresponds to a ridge dot, a CN of 1 corresponds to a ridge
do not even distinguish between ridge ending and                                   ending and a CN of 3 corresponds to a ridge bifurcation.
bifurcations since their types can get flipped during                              Despite decades of research on fingerprint verification,
acquisition.                                                                       reliably matching fingerprints is still a challenging problem
                                                                                   due to the following reasons:
                                                                                   1.    Poor quality images: Sweat, excessively dry fingers,
                                                                                         creases, bruises obscure the ridge structure leading to
                                                                                         generation of spurious minutiae and leading to absence
                                                                                         of genuine ones. Furthermore, intrinsic images cannot be
                                                                                         reliably extracted from poor quality images resulting in
                                                                                         further errors during feature extraction and matching
                                                                                         phases. It was shown in FVC2000 that 80% of the errors
                                                                                         were caused by 20% of the fingerprints. These can be
                                                                                         corrected to an extent by proper enhancement algorithm.
                                                                                   2.    Representational limitation: Human experts utilize
                                                                                         visual and contextual information present in the image
                                                                                         itself in addition to minutiae information while
                                                                                         performing the matching. None of the algorithms or
                                                                                         representations captures this information and utilizes it
                                                                                         in the process of matching. Minutiae based
  Figure 4. Different types of minutiae details present in a fingerprint                 representation completely avoid using the textural and
                                image.
                                                                                         visual information present in the image. While the
                                                                                         texture based matching approaches do not utilize the
                V.     MINUTIAE EXTRACTION                                               positional properties of the minutiae. The correlation
                                                                                         based approaches utilize the global ridge structure of the
    Ridge ending and bifurcations are the representative                                 image, but do not explicitly represent or utilize textural
features of a fingerprint image. In automatic identification,                            or minutiae information. Thus all algorithms mentioned
the two basic features are referred to as minutiae. To                                   in literature rely on incomplete information. It is not
determine the position of minutiae we have to use an                                     known as to how much discriminatory information
efficient minutiae detection algorithm. A 3 x3 window for                                actually exists in each of the fingerprint representations
minutiae determination is placed on a binary image. A pixel                              or in the image itself.
P with its 8 neighboring points (P1….. P2) are defined as
well. The order of neighbors is assigned in an anticlockwise                       3.    Sensor size: Most of the commercial sensors are
direction beginning from the right hand middle point. After                              governed by economy of production than by accuracy
performing the minutiae detection, the value of the CN                                   requirements. Many commercial solid state sensors have
(Crossing Number) determines the information about the                                   area less than 1 square inch. Thus the area of the sensor
minutiae point. The value of the CN can be estimated as                                  is less than the surface area of the fingerprint. It has
                                                                                         shown that the accuracy of the matcher relates directly
   CN =                                     Where P9 = P1                                to the sensor area. The partial prints so obtained do not
                                                                                         have sufficient discriminatory information leading to
    Where, Pi is the pixel value in the neighborhood of P. For
                                                                                         increased error rates during recognition. In effect, the
a pixel P, its eight neighboring pixels are scanned in an
                                                                                         small solid state sensors are undermining public opinion
anticlockwise direction as follows:
                                                                                         about reliability of fingerprints that the forensic




                                                                             174                                   http://sites.google.com/site/ijcsis/
                                                                                                                   ISSN 1947-5500
                                                                    (IJCSIS) International Journal of Computer Science and Information Security,
                                                                    Vol. 9, No. 8, August 2011



     community has worked so hard to establish over the last                           Black squares are used to denote the position of the
     century.                                                                      minutiae points. Then the following is the Distance Vector
                                                                                   for the shown fingerprint image:
4.   Non-linear distortion: The fingerprint image is obtained
     by capturing the three dimensional ridge patterns on the                       DV =(0,2,1,0,0,1,0,1,0,0,0,1,0,2,0,0,0,1,1,0,1,0,0,1,0,1,0,1,0,0,0,0)
     surface of the finger to a two-dimensional image.                                 At the time of enrollment, Distance Vector of each
     Therefore apart from skew and rotation assumed under                          fingerprint image will be calculated. The Distance Vector of
     most distortion models, there is also considerable                            each image will be stored with the template fingerprint
     stretching. Most matching algorithms assumed the prints                       image. At the time of fingerprint matching the Distance
     to be rigidly transformed (strictly rotation and                              Vector of the taken fingerprint image will be compared with
     displacement) between different instances and therefore                       the stored template fingerprint image. Answer of matching
     perform poorly under such situations.                                         depends upon the threshold value we decide for the Distance
                                                                                   Vector. For example, if we assume that there can be
                                                                                   discrepancy of the two Distance Vectors at most at 5 places
                                                                                   (or 5 rows) then this is our threshold value. If we want more
                                                                                   secure system we can decrease the value of threshold to 3 or
                                                                                   2.
                                                                                       In the minutiae extraction phase, we have checked about
                                                                                   three different types of minutiae points i.e. Ridge Dot, Ridge
                                                                                   Ending and Ridge Bifurcation. Many times due to the poor
                                                                                   quality of the image or the imperfections in the fingerprint
                                                                                   image certain minutiae can be missed or in some cases some
                                                                                   spurious minutiae may be generated that makes the
                                                                                   verification algorithm less reliable. We define a new feature
                                                                                   as Distance vector that verifies whether the employed
                                                                                   algorithm performed well or not. At the time of enrollment,
           Figure 6: An illustration of the non-linear distortion                  Distance Vector of each fingerprint image will be calculated.
                                                                                   The Distance Vector of each image will be stored with the
5.   Intra user variation: Due to the limited sensor size, there                   template fingerprint image. At the time of fingerprint
     is a large variation between fingerprints of the same user                    matching the Distance Vector of the taken fingerprint image
     acquired at different instances of time.                                      will be compared with the stored template fingerprint image.
                                                                                   Answer of matching depends upon the threshold value we
        VI. PROPOSED APPROACH MATCHING                                             decide for the Distance Vector.
        THROUGH DISTANCE VECTOR METHOD                                                Here we recapitulate the ideas proposed in this thesis. We
    This is the heart of our proposed methodology. Many                            have developed a novel scheme for fingerprint image
times due to the poor quality of the image or the                                  matching. The proposed method is based on minutiae based
imperfections in the fingerprint image certain minutiae can                        matching approach. This method verifies the reliability of the
be missed or in some cases some spurious minutiae may be                           minutiae extraction phase along with the participation in the
generated that makes the verification algorithm less reliable.                     matching phase itself.
We define a new feature as Distance vector that verifies
whether the employed algorithm performed well or not. A                                               VII. IMPLEMENTATON
Distance vector is a vector that counts the number of
minutiae points in each row of the fingerprint image. For an                           In this paper, we develop a novel scheme for fingerprint
instance, suppose we have a fingerprint of size 32x32 pixels.                      image matching. The proposed method is based on minutiae
After applying the Minutiae Extraction phase, we got                               based matching approach. This method verifies the reliability
minutiae points in the fingerprint as demonstrated in Figure                       of the minutiae extraction phase along with the participation
7.                                                                                 in the matching phase itself.
                                                                                       In the minutiae extraction phase, we will check about
                                                                                   three different types of minutiae points i.e. Ridge Dot, Ridge
                                                                                   Ending and Ridge Bifurcation. Many times due to the poor
                                                                                   quality of the image or the imperfections in the fingerprint
                                                                                   image certain minutiae can be missed or in some cases some
                                                                                   spurious minutiae may be generated that makes the
                                                                                   verification algorithm less reliable. We define a new feature
                                                                                   as Distance vector that verifies whether the employed
                                                                                   algorithm performed well or not.
                                                                                       At the time of enrollment, Distance Vector of each
Figure7: A 32x32 example fingerprint image having minutiae points in its
                                                                                   fingerprint image will be calculated. The Distance Vector of
different pixel positions




                                                                             175                                http://sites.google.com/site/ijcsis/
                                                                                                                ISSN 1947-5500
                                                                       (IJCSIS) International Journal of Computer Science and Information Security,
                                                                       Vol. 9, No. 8, August 2011



each image will be stored with the template fingerprint                                [16] S. Prabhakar, A. Jain and S. Pankanti, “ Learning fingerprint minutiae
image. At the time of fingerprint matching the Distance                                     location and type,” vol. 8, pp. 1847–1857, 2003.
Vector of the taken fingerprint image will be compared with                            [17] N. Ratha, S. Chen, and A. K. Jain, “Adaptive flow orientation based
                                                                                            feature extraction in fingerprint images Pattern Recognition,” vol. 28,
the stored template fingerprint image. Answer of matching                                   pp.1657–1672, 1995.
depends upon the threshold value we decide for the Distance
Vector.

                         VIII. CONCLUSION                                                                        AUTHORS PROFILE
    This is going to be an explorative investigation where
efforts will be put on lighting up the method of fingerprint                           Sumeet Dhawan has done B.Tech Computer Science and Engineering
recognition using distance vector method. We have described                            from Adesh Institue of Engineering and Technolgy, AIET Faridkot,India in
and discussed the fingerprint matching methods. Besides                                2009. Currently he is doing MTech in Department of Information
                                                                                       Technology from Adesh Institue of Engineering and Technology, AIET
this, we have trained and tested the different experiments                             Faridkot, India.
found. The tests were made on fingerprint matching using
distance vector methods. This paper contains information                               Amit Makkar has done B.Tech Computer Science and engineering
about fingerprint matching and distance vector so that a                               Currently he is working as Assistant Professor in Department of Computer
general understandings about the back end procedure can be                             Science and Engineering in Adesh Institue of Engineering and Technolgy,
formulated.                                                                            AIET Faridkot, India. He has 15 years of work experience.


                             REFERENCES
[1]    S. Prabhakar, S. Pankanti and A. K. Jain, “ Biometric recognition:
       Security and privacy concerns,” IEEE Security and Privacy
       Magazine, vol. 1, pp. 33–42, 2003.
[2]    B. Schneier, “Inside risks: The uses and abuses of
       biometrics,”Communications of the ACM, 42:136, 1999.
[3]    S. Pankanti, S. Prabhakar and A. K. Jain, “On the individuality of
       fingerprints,”Transactions on PAMI, vol. 24, pp.1010–1025.
[4]    A. M. Baze, G. T. B Verwaaijen, S. H. Garez, L. P. J. Veelunturf and
       B. J. van der Zwaag, “A correlation-based fingerprint verification
       system,” In ProRISC2000 Workshops on Circuits, Systems and
       Signal Processing, November 2000.
[5]    A. M. Choudhary and A. A. S. Awwal, “Optical pattern recognition
       of fingerprints using distortion-invariant phase-only filter”, In SPIE,
       Photonic devices and algorithms for computer, vol. 3, pp. 162–170,
       October 1999.
[6]    Ardeshir Goshtasby, “Piecewise cubic mapping functions for image
       registration,” Pattern Recognition, vol. 20, pp.25–533, 1987.
[7]    S. Greenberg, M. Aladjem, D. Kogan and I. Dimitrov, “Fingerprint
       image enhancement using filtering techniques,” In International
       Conference on Pattern Recognition, vol. 3, pp.326–329,2000.
[8]    E. P. Simoncelli and H. Farid, “Steerable wedge filters for local
       orientation analysis,” Transactions on Image Processing, vol. 5,1996.
[9]    B.G.Sherlock, D.M.Monro and K.Millard, “Fingerprint enhancement
       by directional fourier filtering,” In Visual Image Signal Processing,
       vol. 141, pp. 87–94, 1994.
[10]   R. Cappelli, A. Lumini, D. Maio, and D. Maltoni, “Fingerprint
       classification by directional image partitioning” IEEE Transactions on
       Pattern Analysis and Machine Intelligence, vol.21,1999.
[11]   G. T. Candela, P. J. Grother, C. I. Watson, R. A. Wilkinson and C. L.
       Wilson, “ Pcasys – a pattern-level classification automation system
       for fingerprints,” April 1995.
[12]   M. Kaas and A. Witkin. Analyzing oriented patterns, “Computer
       Vision Graphics Image Processing,” vol. 37, pp. 362–385, 1987.
[13]   D. Maio and D. Maltoni, “Neural network based minutiae filtering in
       fingerprint images,” In 14th International Conference on Pattern
       Recognition, pp. 1654–1658, 1998.
[14]   A. Jain, L. Hong and R. Bolle, “on-line fingerprint verification,”
       IEEE Trans. Pattern Analysis and Matching Intelligence, vol. 19, no.
       4, pp. 302-314, April 1997.
[15]   R. W. Zhou, C. Quek and G. S. Ng, “Novel Single Pass Algorithm,”
       Pattern Recognition Lett., IEEE Transactions on Image Processing,
       vol. 16, no. 12, pp. 1267-1275, 1995




                                                                                 176                                  http://sites.google.com/site/ijcsis/
                                                                                                                      ISSN 1947-5500

								
To top