Analysis of Algorithms used in Biometric using Fingerprint Authentication for 3D Authentication System Ayushi Mathur 1, Mr. Rupal Gupta 2 1 Research Scholar, CCSIT, Teerthankar Mahaveer University, Moradabad 2 Assitant Professor, CCSIT, Teerthankar Mahaveer University, Moradabad 1 ayushi.mathur92@gmail.com 2 rupal.gupta07@gmail.com Abstract- In today s era, security has great significance on everyone s forum. Biometric appears as a promising tool for 3D Authentication System to further increase the degree of security. Several biometric features like face recognition, fingerprint recognition, iris recognition, heartbeat recognition etc. are widely used. Biometric allude to metrics related to human characteristics. In computer science, Biometrics authentication is used as a form of identification and access control. Biometric identifiers are reliable and unique for individuals based on their physiological (shape of the body) and behavioural (the pattern of behaviour of a person) characteristics. In this paper, we will analyse the algorithms used in biometrics using finger print authentication for 3D authentication system. Keywords- Biometric, Authentication System, Fingerprint recognition, Algorithms, Minutiae Based Algorithm, Image Based Algorithm I. INTRODUCTION To make personal identification, Biometric relies on something that you are so it can inherently differentiate among who is an authorized person and who is a fraudulent person [1, 2, 3, 4]. Biometric authentication is a system that is used for unique physiological and behavioural characteristics of individuals for identification of a person to secure access. Biometric is not used for personal identification to establish absolute yes or no but used for identification to achieve positive [2, 3]. Biometric is a measurable characteristics like facial structure, fingerprints etc. For measuring person s characteristics, we use Biometrics Devices which gives computer vision, pattern recognition etc. Biometric Authentication System uses following performances: Equal error rate False match rate Failure to enrol rate Template capacity Receiver operating characteristic False non-match rate Failure to capture rate Biometric authentication is the most foolproof or at least the hardest to forge or spoof. A. Biometric Overview The following requirements are satisfies for any physiological and behavioural human characteristics than are used for making personal identification [1, 2]: Universality: each person must have characteristic Uniqueness: two persons should not have same characteristics Acceptability: indicates people willing to accept biometric systems up to what extent Collectability: indicates quantitatively characteristic can be measured Circumvention: how easy is to fool system by fraudulent techniques? Performance: achievable accuracy of identification Permanence: characteristic is invariant with time The use of Biometrics is rapidly increasing day by day and these are widespread in 398
forensics for identification of criminals. In civilian applications, biometrics is also widely used such as [9]: Banking security Information system security National ID systems Physical access control Voter & driver registration Nowadays biggest growth area of biometric is Smartphone. They are used as a portable biometric sensor i.e. camera is used for iris or face recognition, microphone is used for voice recognition and keyboard is used for typing rhythm. Smartphone-based approach advantage is that there is no need to purchase any special biometric hardware as users have their phone with them any time needed to log on to the system. B. Types of Biometrics Biometric methods introduced are: Fingerprint recognition Face recognition Signature dynamics Palm or hand geometry Typing rhythm Voice recognition Eye Scans Universality of face recognition and eye scans is high and low for signature dynamics but medium for fingerprint recognition and palm or hand geometry. medium for palm or hand geometry but high for fingerprint recognition and eye scans. Acceptability of face recognition, voice recognition and signature dynamics is high, medium for fingerprint recognition and palm or hand geometry but low for eye scans. II. BIOMETRIC AS FINGERPRINT AUTHENTICATION As uniqueness and permanence of fingerprint recognition is higher than other biometric recognitions, so we mostly use fingerprint authentication system as a means of identification of a person [2, 5]. Fingerprint of a person can never be changed throughout person s death except deep physical injuries and severe burns [11, 12, 13]. Fingerprints present on human fingers are graphical flow-like ridges [10]. On the surface of fingerprint, ridge patterns are formed in womb. Fingerprints are categorized into six categories i.e.: Arch Tented Arch Right Loop Left Loop Whorl Twin Loop Uniqueness for face recognition, voice recognition and signature dynamics is low, medium for palm or hand geometry but high for fingerprint recognition. Performance of face recognition, voice recognition and signature dynamics is low, Arch Tented Arch 399
Right Loop Left Loop Fig. 1 Biometric Authentication Architecture Whorl Fig. 2 Six categories of Fingerprint There are two critical points in a fingerprint i.e. core and delta. Fingerprint is recognized by an automatic pattern recognition system. For Fingerprint recognition, there are three types of fundamental stages [6]: Data acquisition: In this stage, through user interface the fingerprint data is acquired. The image obtained is stored in database. Feature extraction: In this stage, fingerprint features are extracted and then stored in database along with the details. Matching: In this stage, the decision is made for a person to authenticate identity who intends to access the system. The Fingerprint Authentication system architecture is depicted in Fig.1 Fingerprint images are acquired by pressing finger on flat surface of an electronic fingerprint sensor. Based on the sensing technology, electronic fingerprint sensor has three families [8]: Silicon sensor (Solid-state) Optical sensor Ultrasound III. FINGERPRINT AUTHENTICATION: Advantage ADVANTAGE AND DISADVANTAGE Require less time for enrolment with fingerprint scanning system Fingerprints identifiers are unique and specific to individuals Easy to use Require small storage space It is standardized Provides highest level of security Impossible to punch for someone else Disadvantage It is very intrusive Deep physical injuries and severe burns interfere with scanning process Not appropriate with children due to quickly changing size of fingerprint 400
IV. ALGORITHMS USED TO MATCH FINGERPRINT We can match fingerprint through many different techniques such as Euclidean distance, etc. Euclidean distance is calculated between two minutiae points. There are different algorithms for matching fingerprint patterns like minute matching algorithm, image based (or pattern matching) algorithm. In this paper we have analyse two algorithms i.e. minutiae based and image based algorithm and have shown the comparison between them. A. Minutiae Based Algorithm So far work carried out in the area of fingerprint recognition is based on Minutiae Based Algorithm [6, 7]. The major features of fingerprint ridges in minutiae based algorithm are ridge ending, bifurcation and short ridge or dot. Ridge ending is the point at which ridge terminates. Bifurcation is the point at which single ridge split into two ridges. Short ridge or dot is the ridge shorter than average ridge length. A good quality of fingerprint contains around 40 to 100 minutiae [2, 7]. The two individuals can not have common more than seven minutiae. Minutiae base algorithm depends upon local discontinuities in ridge flow pattern and for verification, only small part of finger image is required. In this we use two algorithms: minutiae-extraction algorithm (fingerprint detection) and minutiae-matching (matching fingerprint i.e. input fingerprint and database fingerprint) algorithm. Minutiae-Extraction consist three components: Oriental field estimation, Ridge detection and Minutiae detection. For minutiae-matching we use point pattern matching algorithm. a. Minutiae Extraction Fingerprint authentication is based on minutiae patterns matching. Minutiae extraction consisting three components: Orientation field estimation: The orientation field of fingerprint image is represented as the inborn nature of fingerprint image. Ridge extraction: On ridge the grey-level values attained their local maxima along a direction from normal to a local ridge orientation is the ridges important property in fingerprint image. In this the ridges are thinned. Minutiae extraction and post processing: In this the thinned ridges are mapped. If undesired spikes and breaks are presented in thinned ridges then it may detect many minutiae. To remove spikes and join broken ridges we apply a procedure before minutiae detection. b. Minutiae Matching We can match fingerprint by different strategies, such as point pattern matching, image based matching, ridge pattern matching, graph based scheme, etc. The point pattern matching is the minutiae matching. It is robust and simple matching algorithm. In alignment based algorithm, a minutia matching is decomposed into two stages: Alignment stage Matching stage B. Image Based Algorithm This algorithm uses both feature of fingerprint: macro and micro. For verification, large size of finger image and image area is required as compared to minutiae based algorithm. Requirement of memory is more. The basic fingerprint 401
patterns are compared between previously stored template and input fingerprint. It requires that the previously stored template images and the input fingerprint images should be align in same orientation. The image based algorithm in the fingerprint images finds the central point. In this algorithm, the previously stored template contains size, type and orientation of patterns. The input fingerprint image is compared with the previously stored template to determine their matching degree which they match so that match score generate [14]. V. COMPARISION TABLE 1 S. No. Minutiae Based Algorithm 1. Depends upon local discontinuities in ridge flow pattern 2. Require extensive preprocessing operation 3. For verification, require small part of finger image 4. Uses only minutiae location Image Based Algorithm Uses both features of fingerprint: macro and micro Require very less preprocessing operation For verification, require large size of image Uses grey level information Image based algorithm is better than the minutiae based algorithm. As both minutiae base algorithm and image based algorithm has some strengths and weaknesses. For matching low quality image, minutiae based algorithm has worst performance. So some feature of image based techniques is extracted and integrated into minutiae based technique. By combining both algorithms, the low quality fingerprint images will improve the accuracy of fingerprint matching. VI. CONCLUSION In this paper we have analysed two algorithms used for matching fingerprint i.e. minutiae based and image based algorithm. The minutiae based algorithm doesn t provide exact or correct match for low quality image. If there is a spike or breaks in finger then this algorithm consider those breaks as minutiae and when compared to previously store template through input fingerprint, the error occurs. In image based algorithm we require a large size of fingerprint image to store in database which requires large memory and checks the orientation between previously store template and input fingerprint. In our paper we are proposing the combination of these two algorithms. If we abstract some features of minutiae based algorithm and integrate into image based algorithm, then it will provide better result for low quality image and will also give correct matches of all fingerprints to exact individuals. Its performance can also be superior to both algorithms individually. ACKNOWLEDGEMENT I would like to acknowledge Mr. Rupal Gupta, my research guide for his immense support, genuine concern and invaluable advice that helped and guided me whenever the need arose. REFERENCES [1] R. Clarke, Human identification in information system: Management challenges and public policy issues, Info. Technol. People, Vol. 7, No. 4, pp. 6-37, 1994 [2] Anil k. Jain, Lin Hong, Sharath Pankanti and Ruud Bolle, An Identity-Authentication System Using Fingerprints, 402
Proceedings of the IEEE, Vol. 85, No. 9, pp. 1365-1388, Sep. 1997 [3] E. Newham, The Biometric Report. New York: SJB Services, 1995 [4] S. G. Davies, Touching Big Brother: How biometric technology will fuse flesh and machine, Info. Technol. People, Vol. 7, No. 4, pp. 60-69, 1994 [5] H. C. Lee and R. E. Gaensslen, Eds., Advances in Fingerprint Technology, New York: Elsevier, 1991 [6] M. James Stephen and P. V. G. D Prasad Reddy, Implementation of Easy Fingerprint Image Authentication with Traditional Euclidean and Singular Value Decomposition Algorithms, Int. J. Advance. Soft Comput. Appl., Vol. 3, No. 2, July 2011 [7] Virginia Espinosa-Durg, Minutiae Detection Algorithm for Fingerprint Recognition, IEEE AESS Systems Magazine, pp. 7-10, March 2002 [8] Fernando Alonso-Fernandez, Josef Bigun, Julian Fierrez, Hartwig Fronthaler, Klaus Kollreider and Javier Ortega- Garcia, Fingerprint Recognition, Guide to Biometric Reference Systems and Performance Evaluation, Chp. 4, pp. 51-90, 2009 [9] Shrikhande, Santosh P., and H. S. Fadewar, Finger vein recognition using Discrete Wavelet Packet Transform based features, 2015 International Conference on Advances in Computing Communications and Informatics (ICACCI), 2015 [10] Anil Jain, Introduction to Biometrics, Biometrics, 1999 [11] W. Balder, Dermatoglyphics: Science transition, Vol. 9, pp. 95, 1991 [12] M. Kuchen, C. Newell, A Model for fingerprint formation, Europhys letters, Vol. 68, No. 1, pp. 141-147, 2004 [13] Gualberto Aguilar, Gabriel Sanchez, Karina To scano, Moises Salinas, Mariko Nakano, and Hector Perez, Fingerprint Recognition, Second International Conference on Internet Monitoring and Protection (ICIMP 2007) [14] Subhra Mazumdar and Venkata Dhulipala, Biometric Security using Fingerprint Recognition 403