A Survey on Fingerprint Biometric Recognition System

Varsha J. Rathod', Nalini C. Iyer2, Meena S M3

Dept. ojECE, BVBCET, Hubli, Karnataka, india
1 [email protected]
2 [email protected]
3 [email protected]

Abstract- Individual recognition is based on biometric identification is popular because of its various characteristics
characteristics. A biometric system is an automated method of such as: ease in collecting the data, easily accessible numerous
recognizing an individual. It is an evolving technology which is
sources (ten fingers available for collection). The use of
used in various fields like forensics, secured area and security
fingerprints has a past record from the nineteenth century. It is
system. Fingerprint method of identification is the oldest and
during this time Sir Francis Galton defined the characteristics
widely used method of authentication used in biometrics.
Fingerprint Recognition system designed uses various techniques
points of the fingerprints with which it can be identified and
in order to reduce the False Acceptance Rate (FAR) and False these were referred to as Galton points. These Galton Points
Rejection Rate (FRR) and to improve the performance of the are the base on which fingerprint recognition system is built.
system. In this paper a brief study on the advancement in the The subset of Galton points are used to represent the
techniques used by various authors in the fingerprint recognition fingerprint image and these are referred to as minutiae.
system along with their performance improvement is discussed
below. Gaps are identified based on the observations and an Organization of the rest of the paper is as follows: Section
optimum approach is proposed. II discusses about the detailed review of the related work done
in the field of fingerprint recognition. Section III discusses
Keywords- Biometric system; fingerprint recognition system; about the optimum approach for a fingerprint recognition
minutiae; reconstruction; obfuscation; 3D fingerprint technology system. Section IV discusses the conclusion and future work
I. INTRODUCTION
to be carried on to improve the performance of the system.

Identity establishment is crucial in this immensely

interconnected society. Identity of an individual who requests II. REVIEW OF THE RELATED WORK
a service can be determined by using an identity recognition
system with reliable personal recognition schemes. The The advancement in technology and the vulnerability of
intention of such schemes is to ensure that only the authorized traditional authentication system gives rise to various
user accesses the services and no one else. Biometric innovations in the field of Biometrics. Many authors and
recognition refers to the recognition of an individual based on researchers have used various techniques for fingerprint
their physiological and/or behavioral traits. Biometrics is used recognition involving various stages of enrollment,
to affirm the identity of an individual based on "who the verification and identification. In this section we summarize
person is," rather than by "what the person possesses" (e.g., an the techniques used by various contributory authors, along
Identity card) or "what the person knows" (e.g., a unique with the merits and demerits to improve the overall
password). Biometrics is used to prevent illegitimate access to performance of the fingerprint recognition system.
cellular phones, personal computers, computer networks, Virginia, et at. [2] has proposed a fingerprint recognition
ATMs, safe lockers, secure workstations, used during system by acquiring two fingerprint patterns, the first one was
telephone transactions, internet banking and in automobiles captured using a capturing device of inkless type and the next
[1]. one is the typical method of ink over paper and then the
Amongst all biometrics, fingerprint identification is one fingerprint is scanned. The following preprocessing steps were
of the most widely used and popular biometrics. Because of reported: Image Enhancement, noise reduction, binarization of
the characteristics such as unique, distinct and permanency the image, thinning and removal of false minutiae elements
over time, fingerprints have been used for identification since (mathematical morphology) and a better performing minutiae
1920's; and now due to advancement in technology and extractor algorithm for fingerprint recognition system using
availability of high speed computing systems fingerprint the two databases has been reported.
identification systems are automated. Fingerprint

978-1-4673-7910-6/15/$31.00 ©2015 IEEE 323

 Further Virginia, et at.[3] improved the techniques by (FAR)lFalse Rejection Rate(FRR) region than the other
acquiring two fingerprint databases, replaced the first: using algorithms. Comparison of the results for these systems using
an optical fingerprint reader and the second technique all the fingerprints shows that, the Equal Error Rate (EER) of
remained the same. Thinning algorithm developed employs the proposed algorithm is 1.90%, while that of the POC-based
Morphology techniques and results show that morphological algorithm and minutiae-based algorithm are 3.03% and 4.81%
machine, understood as the particular implementation of the respectively. System performance using only low-quality
morphological functions described, improve the extraction fingerprints shows that, the EER of the proposed algorithm,
process of the Ridge Map developed [2]. POC-based and minutiae-based algorithm are 0.00%, 0.54%
and 10.31% respectively. Therefore results prove that the
Fingerprint images captured depends on various
proposed algorithm is useful in verifying low-quality
conditions and thus are of poor quality. This degradation
fingerprints.
results in creation of a significant number of spurious
minutiae and genuine minutiae being ignored. To reduce the Minutiae extraction in a fingerprint based on both
spurious minutiae created Raymond Thai et at.[4] has reported location and orientation attributes has been reported by
that a sequence of image enhancement techniques are used for Karthik et al.[6] using Fuzzy vault implementation.
fingerprint image enhancement. It consists of four important Implemented technique has been evaluated on FVC2002-DB2
stages to perform normalization, orientation estimation, ridge (Fingerprint Verification Competition 2002- Database2)[35]
frequency estimation and Gabor filtering. Along with the and MSU-DBI(Michigan State University Database) [36]
additional three stages: segmentation, binarization, and fingerprint databases. Further to secure a 128 bit AES
thinning. Experiments were conducted using both synthetic encryption key a fully automatic and fuzzy vault system based
images and real fingerprint images and were used to assess the on fingerprint minutiae has been implemented. To align the
performance of the implemented techniques. The fingerprint template the data from the orientation field is used.
implemented enhancement techniques should be applied A public-domain fingerprint database is used and the proposed
before minutiae extraction. Extraction of minutiae was methodology is implemented showing higher Genuine Accept
performed using Crossing Number method. Experiments Rate (GAR) and lower False Accept Rate (FAR).
conducted have shown that the used method is capable of
Aliaa et al. [7] in her work "Fingerprint Recognition
precisely detecting the ridge endings and bifurcations. In some
System Using Hybrid Matching Technique" has reported that
exceptional cases, if the extracted minutiae are different from
in the feature extraction algorithm four main steps are used,
the original minutiae an additional stage is necessary to verify
they are: for an image determining the centre point, forming
the minutiae. This stage corresponds to an image post­
small squares or blocks around the centre point, using Gabor
processing stage, Minutiae Validation algorithm. The
filtering region of interest is and computing the average
experimental results show that the different types of spurious
absolute deviation from the mean.
minutiae can be removed using an additional post-processing
stage. Further the first prototype of a fingerprint identification
system using Minutiae Tree was developed by Praveer,et al.
Further, the performance of fingerprint
[8]. The methodology was scalable to large datasets. The
recognition/validation is highly effected by the fingertip
amount of time required for enrollment of a template was very
surface condition, which varies several conditions such as
less, and it remained constant. To validate these assumptions
environmental conditions, age, illness and personal causes.
the system was tested on real world and synthetic datasets.
This problem, has been addressed by Koichi Ito, et al.[5], in
The original indexed was not changed whenever a new
the paper "A fingerprint recognition algorithm using phase­
fingerprint was enrolled, thus the indexing system was
based image matching". The phase components in 2D DFT is
dynamic.
used for low-quality fingerprint images. This helps in
achieving a fingerprint recognition system with high The Fingerprint reconstruction techniques used so far
performance. A pressure sensitive sensor is used to capture the generated a partial fingerprint because of the spurious
fingerprint images from 30 subjects. The image captured was minutiae were included which were not present in the original
of 256 x 384 pixels. Among these captured images, 10 were minutiae template. This problem with the reconstruction of
of low-quality. The performance of the proposed algorithm fingerprint images has been addressed by Jianjiang, et al.[9]
based on phase components in 2D has been compared with a where in, the reconstruction algorithm reported reconstructs
classic minutiae-based and a simple Phase Only Correlation­ the phase image, which is later converted into the grayscale
based algorithm. Reports show that the proposed algorithm image. The proposed reconstruction algorithm provides the
exhibits significantly higher performance. This is because the image of a whole fingerprint and it consisted of very few
Receiver Operating Characteristic (ROC) curve of the spurious minutiae. Thus novel fingerprint reconstruction
proposed system is located at lower False Acceptance Rate scheme proposed focuses on converting the minutiae

324 2015 International Conference on Green Computing and Internet of Things (ICGCloT)
representation to the phase representation. The experiment compatible with 2D fingerprint recogllltlOn techniques, the
was performed for plain fingerprints on FVC2002 DB1 A common way is to develop an Automatic Fingerprint
(Fingerprint Verification Competition 2002 Database 1). The Identification System (AFIS) using 3D fingerprints and to
experiment was performed for rolled fingerprints on NIST unwrap 3D fingerprints into 2D equivalent images, and then
SD4 (National Institute of Standards and Technology Special the existing algorithms for 2D fingerprint processing can be
Database 4). Reports show that the reconstruction of utilized. The 3D fingerprints used in this paper were taken
fingerprints with the minutiae-only template in FVC2002 DBI from 5 human subjects and all ten fingers of each subject were
needs 1.40 seconds and NIST SD4 takes around 4.16 seconds scanned twice using a TBS scanner and the output were
per image. unraveled 2D equivalent fingerprint images. False reject rate
was adopted to measure the recognition performance and a
To mask the identity an individual may alter his
commercial software VeriFinger was used for recognition.
fingerprint pattern purposefully. This type of identity disguise
According to the results reported, the false reject rate was very
is referred to as Fingerprint obfuscation. Fingerprint image
high and the number of falsely extracted minutiae was large
quality assessment software (e.g., NIST Fingerprint Image
for 2D to 3D fingerprints recognition. Due to the various
Quality (NFIQ)) fails to identify altered fingerprints.
advantages non-contact 3D fingerprint technology has the
Considering Fingerprint Obfuscation as a major threat,
ability to replace the traditional fingerprint acquisition and
Soweon Woon,et al.[10], proposed an algorithm based on two
recognition system in many applications.
levels: finger level and subject level and a quantitative
analysis was performed. NFIQ had limited ability to detect The benefits of 3D fingerprint technology are higher user
altered fingerprints, thus an algorithm was developed to convenience, hygiene, and improved accuracy. To replace a
automatically detect altered. The proposed algorithm and the 2D fingerprint recognition system by a 3D fingerprint
NFIQ were tested on a large public domain fingerprint recognition system the key limitations are bulkiness and high
database (NIST SDl4). Correct detection of the altered cost systems, which is due to the use of multiple imaging
fingerprints for the proposed algorithm and NFIQ algorithm cameras. Ajay Kumar, et al.[14], has investigated and
were 66.4 percent and 26.5 percent respectively at a false developed a low-cost solution to the above stated problem 3D
positive rate of 0.3 percent. fingerprint identification system by using a single camera. The
proposed system and results presented show that the use of
In concern with the security of fingerprint template,
Lambertian reflectance based shape from shading technique
research on secure fingerprint recognition system based on
helps in developing a more accurate 3D fingerprint
Frobenius norm was proposed by Weizhou Zhao, et al.[11].
identification. Results show that there was a reduction in the
In the ROI of an image the main information can be shown by
Equal Error Rate (EER). The EER for 2D minutiae matching,
Frobenius norm. The system becomes more secure when the
when combined with 3D curvature match scores and when
template is generated by combining Frobenius and random
combined with 3D minutiae match scores were 2.12%, 1.73%
matrix, thus making the fingerprint recognition system much
and 1.17% respectively.
more secure. For the proposed fingerprint recognition system,
the experiments on Fingerprint Verification Competition 2002 III. OBSERVATIONS
Database (FVC2002) show that the system is practicable and
Observations made based on the study of the above
efficient. But, since the random matrix was stored in a
contributory papers are: Fingerprint Recognition system
smartcard or a token and if it was lost then it would cause
employs pre-processing techniques such as image
inconvenience to the authorized user.
enhancement, noise reduction, segmentation, binarization,
Image enhancement plays an important role in any thinning algorithm were implemented ([2],[3],[4]). Minutiae
fingerprint recognition system. This has been reported by extraction was performed based on location and orientation
Azad noor, et al. [12], considering four main stages in pre­ ([2],[3],[4],[6],[7],[8],[9],[11]). To validate the minutiae
processing, they are normalisation, orientation estimation, extracted a post-processing stage, minutiae validation
ridge frequency estimation and Gabor filtering. The enhanced algorithm was implemented ([4]).
image is further processed by segmentation, binarisation, and
Performance of fingerprint recognition system was
thinning. The performance of the algorithm shows that EER
improved by using phase-based image matching ([5]). Fuzzy
is 3.5% or less on optical and capacitive images.
vault implementation with low FAR was implemented ([6]).
Further improvement to the fingerprint recognition Performance of the system was improved by introducing
system was incorporated by using touch less live scan devices image enhancement techniques, minutiae tree, image
that generated 3D representation of fingerprints. Wei Zhou,et reconstruction, forbenius norm ([4],[7],[8],[1O]).Further
al,[13], has evaluated the performance of 2D to 3D improvement to fingerprint recognition system was due to the
Fingerprint Recognition. To make the developed system

2015 International Conference on Green Computing and Internet of Things (ICGCloT) 325
introduction of 3D fingerprint recognition techniques ([11], [II] Weizhou Zhao, Hui Zhang, Secure Fingerprint Recognition Based on
Frobenius Norm, 2012 International Conference on Computer Science
[12]). sand Electronics Engineering

Based on the above listed observations, an optimal [12] A New Algorithm for Minutiae Extraction and Matching in Fingerprint
by Azad Noor
approach for fingerprint recognition system can be designed
[[3] Wei Zhou, Jiankun Hu, Ian Petersen, Song Wang, Mohammed
and implemented using the techniques of image enhancement, Bennamoun, Performance Evaluation of 2D to 3D Fingerprint
minutiae extraction and better fingerprint matching Recognition, 2013 6th International Congress on Image and Signal
techniques. Processing (CISP 2013)

[14] Ajay Kumar, Cyril Kwong, Towards Contactless, Low-Cost and

Accurate 3D Fingerprint Identification, 20[3 IEEE Conference on
Computer Vision and Pattern Recognition
IV. CONCLUSION
Fingerprint recogmtIOn is a method used for
authentication of an individual by measuring, analyzing and
verifying the unique characteristics of fingerprints. This paper
presents a literature survey on the various techniques used in
fingerprint recognition. The approaches are distinguished
based on the several factors like: the data type which they
need to handle, depending on the techniques used for image
enhancement, minutiae extraction and post-processing stages.
In order to achieve desired accuracy and system performance,
two methods have been widely used, first is minutiae and
second one is Gabor filter based. Future research work can be
carried out to improve the quality of the image by improving
the image enhancement technique and develop a better
matching technique.

REFERENCES
[I] Vishwanath G. Garagad, Nalini C. Iyer, A Novel Technique of Iris
Identification for Biometric Systems, IC ACCI 2014.

[2] Virginia Espinosa-Duro, Minutiae Detection Algorithm For Fingerprint

Recognition, EUPMt-UPC (Polytechnic University of Catalonia), 2000

[3] Virginia Espinosa-Duro, Fingerprints Thinning Algorithm, EUPMt-UPC

(Polytechnic University of Catalonia), 2002

[4] Raymond Thai, Fingerprint [mage Enhancement and Minutiae

Extraction, The University of Western Australia, 2003

[5] Koichi Ito, Ayumi Morita, Takafumi Aoki, Tatsuo Higuchi, Hiroshi
Nakajima, and Koji Kobayashi, A Fingerprint Recognition Algorithm
Using Phase-Based [mage Matching for Low-Quality Fingerprints,
IEEE 2005

[6] Karthik Nandakumar, Student Member, IEEE, Anil K. Jain, Fellow,

[EEE, and Sharath Pankanti, Senior Member, [EEE, Fingerprint-Based
Fuzzy Vault: [mplementation and Performance, [EEE Transactions On
Information Forensics And Security, Vol.2, No.4, December 2007

[7] Fingerprint Recognition System Using Hybrid Matching Techniques

Aliaa A.A. Youssif, Morshed U. Chowdhury, Sid Ray and Howida
Youssry Nafaa, Helwan University, Cairo, Egypt

[8] Praveer Mansukhani, Sergey Tulyakov, and Venu Govindaraju, Fellow,

[EEE, A Framework for Efficient Fingerprint Identification Using a
Minutiae Tree, IEEE Systems Journal, Vol.4, No.2, June 2010

[9] Jianjiang Feng, Member, IEEE, and Anil K. Jain, Fellow, [EEE,
Fingerprint Reconstruction: From Minutiae to Phase, IEEE Transactions
On Pattern Analysis And Machine Intelligence, Vol. 33, No.2, February
2011

[[0] Soweon Yoon, Student Member, [EEE, Jianjiang Feng, Member, IEEE,
and Anil K. Jain, Fellow, IEEE, Altered Fingerprints: Analysis and
Detection, [EEE Transactions On Pattern Analysis And Machine
Intelligence, Vol.34, No.3, March 20[2

326 2015 International Conference on Green Computing and Internet of Things (ICGCloT)