Fingerprint Image Enhancement, Thinning and Matching

Document Sample
Fingerprint Image Enhancement, Thinning and Matching Powered By Docstoc
					    International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)
       Web Site: Email:,
Volume 1, Issue 2, July – August 2012                                          ISSN 2278-6856

     Fingerprint Image Enhancement, Thinning and
                                 Dinesh Kumar Misra1, Dr.S.P. Tripathi2 and Ashutosh Singh3
                              PhD Research Scholar, Teerthanker Mahaveer University, College of Engineering,
                                                     Moradabad, Uttar Pradesh, India

                       Professor and HoD , Computer Science and Engineering, IET, Gautam Budha Technical University,
                                             Sitapur Road, Lucknow, Uttar Pradesh, India

                             Ashutosh Singh, Integral University, Department of Computer Science & Engineering,
                                                       Lucknow, Uttar Pradesh, India

                                                                      fingerprint with the permanent uniqueness. Even the
Abstract: Fingerprint image enhancement method is based               fingerprints in twins are not the same. In practice two
on the intrinsic characteristics of fingerprint patterns in the       humans with the same fingerprint have never been found.
Fourier domain rather than spatial domain. Since signal               A fingerprint contains narrow ridges separated by narrow
components of fingerprints ridges are localized in the Fourier
                                                                      valleys and these ridges flows almost parallel to each
domain. So to enhance the localized signal components and
attenuate the other noise component, fingerprint filters is
                                                                      other. The ridges are the dark area of the fingerprint and
designed in Fourier domain one to enhance ridge direction             the valleys are the white area that exists between the
and other for ridge direction enhancement. Fingerprint                ridges [9,12,15]. However in the research on finger print
matching is the process used to determine whether two sets of         verification, we can distinguish it with the help of
fingerprint ridge detail come from the same finger. If the two        minutiae, which are the some abnormal points on ridges.
finger print image is given, the system first extract                 There are two types of termination of minutiae,
orientation field and then minutiae features and establishes
                                                                      immediate ending of ridges or a point from where edges
alignment of two image using algorithm. There exist many
algorithms that do fingerprint matching in many different             ends abruptly called termination and the point on ridge
ways. Some methods involve matching minutiae points                   from where ridge split into two or more branches is
between the two images, while others look for similarities in         known as bifurcation as shown in Fig.1. The most
the bigger structure of the fingerprint. In this Paper, we            commonly used minutiae in current fingerprint matching
propose a method for fingerprint enhancement in Fourier               techniques are ridge endings and bifurcations because
domain and matching based on minutiae matching. However,              they can be easily detected by only looking at points that
unlike conventional minutiae matching algorithms, this
algorithm also takes into account region and line structures
                                                                      surround them.
that exist between minutiae pairs. This allows for more               Henry Faulds , Francis Galton and Edward Henry and
structural information of the fingerprint to be accounted for,        among others had established scientific basis for using
thus resulting in stronger certainty of matching minutiae.            fingerprints as a method of personal identification in late
Also, since most of the region analysis is pre-processed, it          19th century. Since then the finger matching was mostly
does not make the algorithm slower.                                   used by law enforcement agencies worldwide to establish
Keywords: Fingerprint, Minutiae, Minutiae matching,                   the identity of suspect/victims based on partial prints or
Thinning, Scanning, Edge Enhancement, Region
                                                                      latent and to identify repeat offenders based on prints of
                                                                      all their fingers. But these techniques are finding many
                                                                      other applications like identity management, access
1. INTRODUCTION                                                       control, attendance and time management etc.
The subject of fingerprint image enhancement and                      Due to rising concern on about security and frauds many
processing for matching has been examined and many                    Government and commercial organizations have
methods are proposed. Most of the methods have several                substantially increased their own deployment of finger
steps, out of that one step is to detect fingerprint ridge            based matching and recognition system in several no
direction and other step is to use the direction to enhance           forensic applications including physical and logical
the image [2, 3, 5, 6]. Two of the fundamentally                      access control , ATM transaction ,border control and
important conclusions that have risen from research are:              consumer device services. Fingerprint is dominant
(i) a person's fingerprint is natural structure and is not of         biometric trait in this application compared to other
changing nature (ii) everyone in this world has his own               common traits such as face, iris and voice and new

Volume 1, Issue 2 July-August 2012                                                                                     Page 17
   International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)
       Web Site: Email:,
Volume 1, Issue 2, July – August 2012                                          ISSN 2278-6856

emerging traits including gait, ear and palm vein.             points in another image then the points are said matched.
                                                               Still Finger print matching is difficult pattern recognition
                                                               problem due to large intra-class variations like variation
                                                               in finger print image of same finger and inter class
                                                               variations like similarity between finger print images
                                                               from different fingers.      It is the idea of this paper to
                                                               filter and rectify the above constraint.

                                                               3. IMPLEMENTATION PROCESS
                                                               Finger print image extraction is first step in this
                                                               implementation process. This is mainly done to improve
                                                               the image quality and make it clearer for further
                                                               operations. Often Finger print image from various sources
                (a)                  (b)                       lack sufficient contrast and clarity. Hence enhancement
               Figure 1 Minutiae Features                      by way image extraction is performed to improve
                                                               accuracy of matching .It increases contrast between ridges
In this paper Section 2 describes the Minutiae matching        and furrows and connect the some of the false broken
method. Section 3 explains implementation process.             points of ridges due to insufficient amount of ink or poor
Section 4 deals testing procedure. Section 5 describes         quality of sensing devices. It can be done by utilizing
output results as net outcome of proposed method and           Histogram Equalization.
section 6 at the end future work.                              Histogram equalization [12] is a technique of improving
                                                               the global contrast of an image by adjusting the intensity
                                                               distribution on a histogram. This allows areas of lower
2. MINUTIAE MATCHING METHOD                                    local contrast to gain a higher contrast without affecting
Fingerprint minutiae matching scheme [4,9,13,14] can be        the global contrast. Histogram equalization accomplishes
classified in three groups:                                    this by effectively spreading out most frequent intensity
   2.1 Correlation Based Matching                              values. The original histogram of fingerprint image has
In this process two finger print images are superimposed       bimodal type (in Fig.2.a), the histogram after equalization
and correlation between corresponding pixels is computed       occupies all the ranges from 0 to 255 and the
for different alignment e.g. displacement and rotation.        visualization effect is enhanced as shown in Fig.2.b
   2.2 Minutiae Based Matching                                 The result of histogram equalization is shown for one of
In this process of matching minutiae are extracted from        the case of fingerprint image in fig.3.
two fingerprints and stored as set of points in two
dimensional planes. It consists of finding the alignment
between template and the input minutiae sets that result
in the maximum number of minutiae pairings.
   2.3 Pattern or Image based matching
Pattern based matching use algorithms to compare the
basic fingerprint patterns like arch, whorl or loop between
a previously stored template and candidate fingerprint.
For this purpose image is required to be aligned in same
                                                                 Figure 2 (a) Original Histogram (b) Histogram after
orientation. In matching process algorithms finds a
central point on the fingerprint image and centers on the
image. In pattern based algorithm, the template contains
the type, size and orientation of pattern within the aligned
fingerprint image. The candidate fingerprint image is
graphically compared with the template to determine the
degree to which the match.
Fingerprint matching technique uses one of the above
techniques but most of the cases minutiae matching is
utilized. Matching of minutiae is described above. Other
technique of minutiae matching is minutiae distance
relative to other minutiae around it. If multiple points in     Figure 3 (a) Original Image    (b) Enhanced image
                                                                            after Histogram Equalization
one image have similar distances between them, multiple

Volume 1, Issue 2 July-August 2012                                                                               Page 18
   International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)
       Web Site: Email:,
Volume 1, Issue 2, July – August 2012                                          ISSN 2278-6856

  3.1 Thinning                                                    3.2 Scanning a Fingerprint
Thinning [1, 15] was done using the Zhang-Suen                 Some of the algorithms require a linear scan of the
algorithm as described below, A Fast Parallel Algorithm        fingerprint image. Scanning is achieved by moving a
for Thinning Digital Patterns. A 3x3 window is moved           fixed size window across the picture in a grid-like pattern
down throughout the image and calculations are carried         as shown in fig.5. This can be seen in the image to the
out on each pixel to decide whether it needs to be in the      right. However, it is possible that areas of interest do not
image or not. To the right is a description of the window      lie squarely in the one of the windows. To account for
and the classification given to the pixels that surround the   this the window is then shifted just vertically, just
centre pixel. The algorithm runs two sub-iterations            horizontally, and then vertically and horizontally by half
continuously until the image reaches a stable state.           the window size and the grid scan is completed again.
Table1. 3x3 windows to verify need of pixel in image.          Therefore, it takes four scans of the image to do the linear
                                                               scan. This is not a problem because it is used for the pre-
                                                               processing of a fingerprint image (which only occurs
                                                               once) and is done in a linear manner.

  Pseudo Code for Zhang –Suen Thinning:
    Let A(P) be the number of 01 patters in the order set
    P2… P9
    Let B(P) be the number of non-zero neighbors of P                   Figure 5 Scanning of finger print Image
    Do until image is stable (i.e. no changes made)
    Sub-iteration 1:                                             3.3 Edge Enhancement
    Delete P from image if:                                    Regions are defined by the fingerprint edges that bound
    a) 2 = B(P) = 6                                            them. However, because of the nature of the fingerprint
    b) A(P) = 1                                                and current scanning techniques, ridge detail can be
    c) P2* P4* P6= 1                                           missing from the scanned fingerprint. Furthermore, the
    d) P4*P6*P8=1                                              thinning algorithm can also eliminate some of the edge
    Sub –iteration 2:                                          detail. Most missing edges take the form of gaps in an
    a) and b) above                                            edge and are usually easily identified by a human. Still
    c’) P2*P4*P8 =1                                            problem occurred in computer to recognize them. In the
    d’) P2*P4*P6 =1                                            course of the project one of the edge enhancement
                                                               techniques utilized, which caught some of the simple
This processes effectively thins the image, however, it        gaps, however, did not fill in the bigger or more complex
sometimes creates undesirably artifacts. In the example        ones.
of a Zhang-Suen thinned fingerprint there are gaps               Pseudo Code for Edge Enhancement:
between edges as well as regions of small area that need           Scan Image as explained in previous section:
to be removed for proper regional analysis as shown in             For each window:
Fig.4.                                                                Find all the endpoints in the window
Example of Zhang-Suen Thinned fingerprint:                         For each endpoint pair
                                                                      Look at the line that goes through the points
                                                                   If the line is "strong"
                                                                      Draw the line between the points

                                                               A line is said to be strong if many points on the line are
                                                               already in or near points in the existing edge detail. If the
                                                               line is strong then it is most likely that the ridge was
                                                               supposed to span the gap and therefore can be drawn in.
                                                               Note that because the image has been thinned, it is easy to
      Figure 4 Input image           Thinned Image             find endpoints in the image. Endpoints are merely pixels
                                                               that only have one neighboring pixel.

Volume 1, Issue 2 July-August 2012                                                                                Page 19
   International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)
       Web Site: Email:,
Volume 1, Issue 2, July – August 2012                                          ISSN 2278-6856

                                                               were used. Two separate minutiae pairs were used and
   3.4 Region Coloring                                         the corresponding minutiae pairs were found on each of
Once the edge enhancement process completed and edges          the 4 scans as shown in fig.7 and Fig.8.
have been properly added to the image, the region
coloring [11] process started. The region coloring is          5. RESULT
similar to taking the image into Microsoft Paint and using     Four input Image 1,2,3 and 4 are taken as input for
the color fill tool on all the valleys of the fingerprint,     region coloring, then spurious region removal, if present
however, the code does it in a little more efficient way.      any by using algorithm written .The results are shown
The image is scanned and colors are associated with            below as output Image 5,6,7 and 8 respectively.
white space in the image. While the image is being
colored in each pixel looks at the pixels around it to         Input Fingerprints:
determine what color it should be and color equivalences
are set up. Once again image is being scanned, replacing
all the colors in the equivalence set with just one of the
colors from them. Therefore, each pixel in a region has
the same color number associated with it.

Example of Region coloring:

                                                                      Image1              Image2

 Figure 6 (a) Input Image               (b) Output Image
                                                                      Image3             Image4
  3.5. Spurious Region Removal                                        Figure 7 Four input Image (1, 2, 3, and 4)
As an artifact of the thinning process small regions are
created where there should be no region. Because the           The pre-processing algorithms were run on each of the 4
matching process relies on the region pre-processing to be     different scans of the same finger to obtain the output
accurate these regions must be removed.                        fingerprints:
  Pse udo Code for Re gi on Re moval :
    For each window in scan:
        For each region encased in the window that is
    sufficiently small
         Remove the border between the erroneous
    region and the region it borders with least.

This will make it so that regions that are small are spilled
into a surrounding region that it most likely was supposed
to belong to.                                                         Image 5            Image 6

Once pre-processing is completed, the image is used for
the matching process. In matching process, first we had
to show that the region data was a good indication of
matching minutiae. That is we need to show that the
amount of regions between two minutiae is consistent
across many different scans of the same fingerprint. To
test our hypothesis 4 different scans of the same finger             Image 7             Image 8

Volume 1, Issue 2 July-August 2012                                                                             Page 20
   International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)
       Web Site: Email:,
Volume 1, Issue 2, July – August 2012                                          ISSN 2278-6856

                                                                feature- feature extraction process,” IEICE
  Figure 8 Output Image 5,6,7,8 (corresponding output           transactions, Vol.J72-D-II, N0.5, pp 724-732, 1989.
      image of input image1, 2, 3, 4 respectively)          [8] A.K.Jain, S Prahbakar, S.Hong,, L Pankanti, “Filter
                                                                Bank based Finger print Matching,” IEEE
                                                                transaction on Image Processing,Vol.9,No.5,pp.846-
6. CONCLUSION AND FUTURE WORK                                   859, May,2000, DOI :10.1109/83.841531
Future work on this project would include the creating of   [9] Salil Prabhakar, “Fingerprint classification and
a matching algorithm that uses the regional data created        matching using filter bank”, Ph.D Thesis, 2001.
in this pre-processing system. The efficient matching       [10] Eu Zhu, Jianping Yin,Guomin Zhang, “Fingerprint
algorithm have to be developed in theory and in code so          matching based on global alignment of multiple
that our goal of getting faster and more accurate matched        reference minutiae, “ Pattern Recognition 38,pp1685-
image than with pre-existing software. Also, more work           1694,2005.
can be done on the edge enhancing algorithm as it does      [11] J. Feng, “Combining Minutiae Descriptors for
not properly detect missing edges that are supposed to be        Fingerprint matching,” Pattern recognition, pp.342-
curved. It can be accomplished by edge orientation, edge         352, Jan.2008.
enhancement using energy minimization principle.            [12] S. Bana and Dr. D. Kaur, “Fingerprint recognition
                                                                 using image segmentation,” IJEAST, Vol. No.5,Issue
                                                                 No.1,pp.012-023, 2008.
            ACKNOWLEDGEMENT                                 [13] Manvjeet Kaur, Mukhwinder Singh, Akshay
Authors are highly thankful to Teerthankar Mahaveer              Giridhar and Parvinder S Sandhu , “Finger print
University (TMU) , Moradabad, U.P., India for all kind of        Verification System using Minutiae Extraction
support and co-operation to pursue research work under           Technique,” World Academy of          Science ,
able Professor Dr. S.P.Tripathi of Gautam Budhha                 Engineering and Technology,46 , 2008.
Technical University (GBTU), Lucknow, U.P. India.           [14] A.K.Jain,   J    Feng    and     K  Nandakumar,
                                                                 “JainFpMatching”, IEEE          Computer Society,
REFERENCES                                                       0018/9162, pp.36-45, 2010.
[1] L.Lam, S.W.Lee, and C.Y. Suen, “Thinning                [15] C Saraswat and Amit Kumar, “An Efficient
    Methodology-A Comprehensive Survey,” IEEE                    Automatic Attendance System using Fingerprint
    Transaction on Pattern Analysis and machine                  Verification Technique,” IJCSE, Vol.02, No.2,
    intellgence,vol.14,no.9,1992.                                pp.264-269, 2010.
[2] Toshio Kamei and Masanori Mizoguchi ITRL, NEC
    Corporation, Japan: “Image Filter Design For            Dinesh Kumar Misra has received B.Tech (ET) from
    Fingerprint Enhancement”- Proc. ISCV, 95, pp109-        IETE, New Delhi, M.Tech.(CSE) from AAIDU,
    114, 1995.                                              Allahabad, U.P., India, MBA (HR) from IGNOU, New
[3] L. Hong, A. K. Jain, S. Pankanti and R. Bolle,          Delhi and Perusing Ph.D in CSE form College of
    “Finger print Enhancement,’ Proc. First IEEE            Engineering, TMU, Moradabad, U.P., India on Digital
    WACV, pp.202-207, Sarasota, 1996.                       Image processing. He has published 5 International and 3
[4] N.K Ratha, K.Karu,S.Chen, A.K.Jain, “A Real-time        National papers. Presently he is working as
    matching system for large fingerprint database, IEEE    Scientist/Engineer in Department of Space, ISRO,
    transaction on Pattern Analysis and Machine             Lucknow U.P., India since 1993.
    Intelligence Vol..18, No.8,pp.799-813,1996              Dr. S.P. Tripathi, Professor and Head in CSE
[5] Hong Y. Wan, A.K.Jain, “Fingerprint Image               Department, IET, GBTU, Lucknow, U.P., India. He is Ph
    enhancement      Algorithm     and  performance         .D in CSE from Lucknow University, Lucknow, U.P.,
    Evaluation, IEEE Trans., Pattern Analysis &             India and able Professor to supervise more than 8 Ph.D
    Matching,Intell.20,pp.777-789,1998.                     students. He has published more than 25 research paper.
                                                            He has keen interest in research in the field of Data
[6] Lin Hong, Wan Yi-fei and A. Jain, “Finger print
                                                            structure, Fuzzy Logic and Data Base Management.
    image Enhancement: Algorithm and Performance
    Evaluation,” IEEE transaction on Pattern Analysis       Ashutosh Singh, B.Tech in CSE from Integral
    and Matching Intellgence, Vol.20, no.8, pp777-789,      University, Lucknow, U.P., India is Final year student
    1998.                                                   and working on the above cited work.
[7] K. Asai, Y. Hoshino and K. Kiji, “Automated
    fingerprint Identification by minutiae-network

Volume 1, Issue 2 July-August 2012                                                                          Page 21

Description: International Journal of Emerging Trends & Technology in Computer Science (IJETTCS) is an online Journal in English published bimonthly for scientists, Engineers and Research Scholars involved in computer science, Information Technology and its applications to publish high quality and refereed papers. Papers reporting original research and innovative applications from all parts of the world are welcome. Papers for publication in the IJETTCS are selected through rigid peer review to ensure originality, timeliness, relevance and readability. The aim of IJETTCS is to publish peer reviewed research and review articles in rapidly developing field of computer science engineering and technology. This journal is an online journal having full access to the research and review paper. The journal also seeks clearly written survey and review articles from experts in the field, to promote intuitive understanding of the state-of-the-art and application trends. The journal aims to cover the latest outstanding developments in the field of Computer Science and engineering Technology.