International Journal of Advanced Research in ISSN : 2347 - 8446 (Online)

Computer Science & Technology (IJARCST 2014) Vol. 2 Issue 1 Jan-March 2014 ISSN : 2347 - 9817 (Print)

Biometric Fingerprint Indentification Using

Artificial Neural Network
I
Neeraj Singla, IISugandha Sharma
I,II
Department of CSE, CGC, Gharuan, Punjab, India

Abstract
Fingerprint recognition is one such area that can be used as a means of biometric verification where the ANN can play a critical rule.
Fingerprint identification is mainly used in artificial intelligence. It is popular because of their easy access, low price of fingerprint
sensors, non-intrusive scanning, and relatively good performance. In recent years, significant performance improvements have been
achieved in commercial automatic fingerprint recognition systems. The fingerprint of an individual is unique and remains unchanged
over a lifetime. No two persons have the same set of fingerprints. This property makes fingerprints an excellent biometric identifier.
So it is one of the popular and effective means for identification of an individual and used as forensic evidence.

Keywords
Artificial Neural Network; Fingerprint; Biometrics, Feature Extraction, Image Acquisition

I. Introduction like orientation. A good quality fingerprint image typically has

Fingerprints have been used for over a century and are the about 40 to 100 minutiae, but a dozen of minutiae are considered
most widely used form of biometric identification. Fingerprint sufficient to identify a fingerprint pattern [2].
identification is mainly used in artificial intelligence. It is popular
because of their easy access, low price of fingerprint sensors, non- B.Fingerprint Recognition
intrusive scanning, and relatively good performance. In recent It is one of the popular biometric techniques. It refers to the
years, significant performance improvements have been achieved automated method of verifying a match between two fingerprint
in commercial automatic fingerprint recognition systems. The images. It is formed by the ridge patterns of the finger [2].
fingerprint of an individual is unique and remains unchanged Figure 1. shows an image of the fingerprint structure.
over a lifetime. No two persons have the same set of fingerprints.
This property makes fingerprints an excellent biometric identifier.
So it is one of the popular and effective means for identification
of an individual and used as forensic evidence. A fingerprint is
formed from an impression of the pattern of ridges on a finger.
A ridge is defined as a single curved segment and a valley is
the region between two adjacent ridges. Typically, there are two
prominent types of minutiae (ridge ending and ridge bifurcation)
that constitute a fingerprint pattern. The minutiae which are
the local discontinuities in the ridge flow pattern provide the
features that are used for identification. Details such as the type,
orientation and location of minutiae are taken into account when
performing minutiae extraction [1]. Here, we propose a Fingerprint
Recognition System (FRS) based on Artificial Neural Network
(ANN). The rest of the paper is organized as follows: Section II
provides the brief description of a generic fingerprint recognition
system. Section III provides the background principles related Fig. 1: A Fingerprint Sample Image
to the working of the proposed model. All experimental results
and related discussion is provided in section IV-V. This paper is B. Artificial Neural Network
concluded by summing up the work in section VI. Some of the It is a soft computing tool that can learn patterns and predict. A
relevant literatures are cited between [6]-[7]. neural network is a biologically inspired method of computation.
Neural networks used for computation should appropriately be
II. Basic Theoritical Aspects Related to the Proposed called Artificial Neural Network(ANN)s. These only resemble the
System parallel computation generated by the biological neural network
Here we briefly cover the basic theoretical aspects related to the which is the basics of human learning.
work.
III. System Model
A. Fingerprint A generic ANN based Fingerprint Recognition System in block
These are graphical flow-like ridges and valleys present on the diagram form is shown in Figure 2.
surface of human fingers [3]. A ridge ending is defined as the ridge
point where it ends abruptly. A ridge bifurcation is defined as the
ridge point where a ridge diverges into branch ridges. A fingerprint
can be represented by the minutiae locations, types and attributes

© 2014, IJARCST All Rights Reserved www.ijarcst.com

28
ISSN : 2347 - 8446 (Online) International Journal of Advanced Research in
ISSN : 2347 - 9817 (Print) Vol. 2 Issue 1 Jan-March 2014 Computer Science & Technology (IJARCST 2014)

Figure 2: Process logic of the Complete System

It involves the following modules: Figure 3: System Model Design

1. Image Acquisition In this proposed model, a multi stage approach is used. These
An image can be captured by a digital scanner, digital camera, stages are image acquisition, image preprocessing, thinning,
sensor etc. It involves recording of a scene by a camera, sampling feature extraction, classification and decision.
and quantization and storing it in a computer for manipulation. Fingerprint image preprocessing includes image segmentation,
The best of digital image capture record represents loss of data normalization, ridge orientation estimation, ridge frequency
as the sampling process involved in digitization selects samples estimation, thinning etc. Image segmentation is the process
at specific intervals only. of separating the foreground regions in the image from the
background regions. The foreground regions correspond to the
2. Image Preprocessing Stage clear fingerprint area containing the ridges and valleys, which is the
It includes various stages which should be taken for making an area of interest. The background corresponds to the regions outside
image suitable for manipulation and interpretation by subsequent the borders of the fingerprint area, which do not contain any valid
stages. The steps include removal of noise and variation of intensity fingerprint information. When minutiae extraction algorithms
recorded, sharpening, improving the contrast and strengthening the are applied to the background regions of an image, it results in
texture of the image. Another important aspect is image restoration the extraction of noisy and false minutiae. Thus, segmentation is
which extracts image information from a degraded form to make employed to discard these background regions, which facilitates
it suitable for subsequent processing and interpretation [2]. the reliable extraction of minutiae. The segmented images are then
Feature extraction: It is a process through which certain vital normalized. Image normalization is required to standardize the
information and details of an image section is captured for intensity values in an image by adjusting the range of gray-level
subsequent interpretation. values so that it lies within a desired range of values.
After normalization, we have done the thinning process. Thinning
3. Classification is a morphological operation that successively erodes away the
This is the key component of fingerprint recognition system and foreground pixels until they are one pixel wide. A standard
determines the system‟s performance to a large extent. In this thinning algorithm [7] is employed, which performs the thinning
proposed model an Artificial Neural Network is used as classifier operation using two sub iterations. This algorithm is accessible in
for recognition. It can generate multiple classes and decision MATLAB via the „thin‟ operation under the „bwmorph‟ function.
boundaries and it produces the correct result by classifying the The application of the thinning algorithm to a fingerprint image
feature extracted templates and matching these features with preserves the connectivity of the ridge structures while forming
known patterns in the feature database. a skeletonised version of the normalized image.
The thinned images are next considered for the minutiae feature
IV. Design and Implementation of the Proposed extraction. The minutiae feature extraction algorithm which
System extracts the main minutiae features required for matching two
In this work the focus is to implement the system that provides fingerprints. Here, Crossing Number (CN) method is used for
reliability, accuracy and reduced overall match speed. minutiae extraction of fingerprints [8]. The ridge pixel can be
The steps of the algorithm of the system model is shown in Figure divided into bifurcation, ridge ending and non-minutiae point
3. based on it. The CN algorithm is working on pixel representation
„1‟ or „0‟, but the decision of minutiae point can be selected
for each pixel value. CN method extracts the ridge endings and
bifurcations from the skeleton image by examining the local
neighborhood of each ridge pixel using a 3x3 window. The CN
for a ridge pixel P is given by:
CN= , (1)

www.ijarcst.com 29 © All Rights Reserved, IJARCST 2014

International Journal of Advanced Research in ISSN : 2347 - 8446 (Online)
Computer Science & Technology (IJARCST 2014) Vol. 2 Issue 1 Jan-March 2014 ISSN : 2347 - 9817 (Print)

Where is the pixel value in the neighborhood of P. For a pixel Table II: ANN Specifications
P, its eight neighboring pixels are scanned in an anti-clockwise Fingerprint with CN features of
direction as follows: Input Data Size
length 160
P4 P3 P2 SNR 0 to 3dB
MLP with two hidden layers.
P5 P P1 First hidden layer 1.25 times the
ANN type length of feature vector and second
hidden layer 0.75 times the length
P6 P7 P8 of feature vector.
ANN training Back propagation with Levenberg-
After the CN for a ridge pixel has been computed, the pixel can method Marquardt optimization
then be classified according to the property of its CN value. With Average training
MLP-200 to 3000
this formula, if CN=1 it corresponds to the end point and if CN=3, epochs
it corresponds to Bifurcation point of minutiae. Other properties of Mean Squared
CN are described in Table I. In applying this algorithm, border area 10-4
Error(MSE) goal
may be ignored, since there is no need to extract minutiae point
on border area of the image that will gives more false minutiae The MSE plot shown by the ANN during training while configuring
points. the FRS is shown in Figure 4. Original, thinned and minutiae
extracted fingerprints are shown in Figures 5 to 7. The tested
Table I: Properties of CN results of recognition for various training epochs for the FRS is
shown in Table III. The epochs are between a few hundreds to a
few thousands and the success rate is around 84 to 94%.
Computation time is counting based on tic/toc command from
Matlab, a command to start stopwatch timer and measure the
time required for the operation between tic and toc statements.
The training time required is between 35 to 38 seconds for a set
of ten samples each. The results are derived by performing fifteen
trials for each of the sample sets and the average results quoted.
The strength of the proposed system is its speed, computational
efficiency, robustness, decision and high precision which shall
make it suitable for certain application.

After a successful extraction of minutiae, they are stored in a Table III: Average Success Rates Achieved Between a Few
template, which may contain the position, direction, type and quality Numbers of Training Epochs
of the minutiae. In this proposed model ANN is used as classifier Epochs % Success rate of FRS
for recognition. Here, a feed forward back propagation neural
200 84
network is configured for the classification of the fingerprints. For
the fingerprint approach, the ANN input layer has 160 neurons and 500 92
is trained for 200 to 3000 epochs. The results obtained are average 1000 93
values of atleast fifteen trails for the epochs considered. 2000 93.6
3000 94
V. Experimental Details and Results
The performance of FRS is analyzed in terms of computational
speed and reliability. The overall computational time taken by
the system is reduced to a greater level. A total of 40 identical
fingerprint images have been provided to the system for training,
validation and testing of the system. After extensive training,
the system is subjected to certain variations with signal to noise
ratio (SNR) range between 0 to 3dB to achieve robustness and
proper recognition. The ANN considered is configured using the
specifications shown in Table II.
The default performance function for feed forward neural network
is mean squared error (mse) which is the average squared error
between the network outputs and the target outputs.

Figure 4: MSE plot

© 2014, IJARCST All Rights Reserved www.ijarcst.com

30
ISSN : 2347 - 8446 (Online) International Journal of Advanced Research in
ISSN : 2347 - 9817 (Print) Vol. 2 Issue 1 Jan-March 2014 Computer Science & Technology (IJARCST 2014)

Biography:-

ER. NEERAJ SINGLA, pursuing M.Tech

- CSE from CGC Gharuan and don
B.Tech. degree from punjab technical
university.

Figure 5: Original Fingerprint image

ER. SUGANDHA SHARMA is working as

Assistant professor in CGC Gharuan in
CSE department and she has almost 4 year
Figure 6: Thinned Fingerprint image teaching experiece and she has done her
M.E. From Punjab university.

Fig. 7: Minutiae Extracted Fingerprint

VI. Conclusion
The specialty of the work is associated with the fact that if the ANN
is configured properly it can tackle the variations in the fingerprint
images and that way provides the insights for developing a
system which requires these samples as inputs for verification
and authorization. A system designed to provide authentication
decision using these inputs can be a reliable means of verification
as has been observed from experimental results.

References
[1] Jain, A., Hong, L., Pankanti, S., and Bolle, R. “An identity
authentication system using fingerprints. In Proceedings of
the IEEE (September 1997), vol. 85, pp. 1365–1388.
[2] R. Thai, “Fingerprint Image Enhancement and Minutiae
Extraction”, 2003.
[3] A. K.Jain, S. Pankanti and R. Bolle, Biometrics: Personal
Identification in Networked Society[M]. Boston: Kluwer
Academic Publishers, 1999.
[4] A. K.Jain, L. Hong, S. Pankanti and R. Bolle, “On-line
Fingerprint verification”.
[5] N.K.Ratha, S.Chen and A.K. Jain “ Adaptive Flow Orientation
Based Feature Extraction in Fingerprint Images”.

www.ijarcst.com 31 © All Rights Reserved, IJARCST 2014