A Biometric Authentication System That Automatically Generates Feature Points

Similar documents
A study of the Graphical User Interfaces for Biometric Authentication System

A Combined Method for On-Line Signature Verification

Behavior-based Authentication Systems. Multimedia Security

A Study on the Consistency of Features for On-line Signature Verification

Visualization of the Packet Flows using Self Organizing Maps

Online Signature Verification Technique

Feature Selection by User Specific Feature Mask on a Biometric Hash Algorithm for Dynamic Handwriting

: BIOMETRIC AUTHENTICATION TOOL FOR USER IDENTIFICATION

Exploring Similarity Measures for Biometric Databases

STUDY OF POSSIBILITY OF ON-PEN MATCHING FOR BIOMETRIC HANDWRITING VERIFICATION

BIOMET: A Multimodal Biometric Authentication System for Person Identification and Verification using Fingerprint and Face Recognition

Global Mobile Biometric Authentication Market: Size, Trends & Forecasts ( ) October 2017

An Efficient on-line Signature Verification System Using Histogram Features

Multimodal Biometric System by Feature Level Fusion of Palmprint and Fingerprint

Use of Extreme Value Statistics in Modeling Biometric Systems

On-line Signature Verification on a Mobile Platform

Recall Based Authentication System- An Overview

An Improved Iris Segmentation Technique Using Circular Hough Transform

Graphical User Authentication System An Overview P. Baby Maruthi 1, Dr. K. Sandhya Rani 2

Secure and Private Identification through Biometric Systems

INTERPRETING FINGERPRINT AUTHENTICATION PERFORMANCE TECHNICAL WHITE PAPER

Biometric identity verification for large-scale high-security apps. Face Verification SDK

Generation of Artistic Calligraphic Fonts Considering Character Structure

DEFORMABLE MATCHING OF HAND SHAPES FOR USER VERIFICATION. Ani1 K. Jain and Nicolae Duta

CSE 565 Computer Security Fall 2018

Obtaining Biometric ROC Curves from a Non-Parametric Classifier in a Long-Text-Input Keystroke Authentication Study

Hybrid Biometric Person Authentication Using Face and Voice Features

A BIOMETRIC FUSION OF HAND AND FINGER VEIN APPROACH FOR AN EFFICIENT PERSONAL AUTHENTICATION IN HEALTH CARE

Approach to Increase Accuracy of Multimodal Biometric System for Feature Level Fusion

Shape Feature Extraction for On-line Signature Evaluation

Continuous User Authentication Using Temporal Information

Exploring Games for Improved Touchscreen Authentication on Mobile Devices

CSCE 548 Building Secure Software Biometrics (Something You Are) Professor Lisa Luo Spring 2018

IJESRT. Scientific Journal Impact Factor: (ISRA), Impact Factor: 1.852

Trial decision. Appeal No Kyoto, Japan. Tokyo, Japan

Off-line Signature Verification Using Neural Network

Technical White Paper. Behaviometrics. Measuring FAR/FRR/EER in Continuous Authentication

Body Sensor Network Security

arxiv: v1 [cs.cr] 7 Jun 2015

US Secret Service National Threat Assessment Center (NTAC), Insider threat study (2004)

CIS 4360 Secure Computer Systems Biometrics (Something You Are)

Signature Verification Why xyzmo offers the leading solution

A semi-incremental recognition method for on-line handwritten Japanese text

FINGERPRINT RECOGNITION FOR HIGH SECURITY SYSTEMS AUTHENTICATION

Discovering Computers Chapter 5 Input. CSA 111 College of Applied Studies UOB

Biometrics Our Past, Present, and Future Identity

Intruders, Human Identification and Authentication, Web Authentication

Face Detection Using Radial Basis Function Neural Networks With Fixed Spread Value

Customized Data Filtering For Mobile Signature Verification

Palm Vein Technology

Electronic Signature Systems

Deprecating the Password: A Progress Report. Dr. Michael B. Jones Identity Standards Architect, Microsoft May 17, 2018

Mobile Biometric Authentication: Pros and Cons of Server and Device-Based

CHAPTER 5 FEASIBILITY STUDY ON 3D BIOMETRIC AUTHENTICATION MECHANISM

Reducing FMR of Fingerprint Verification by Using the Partial Band of Similarity

International Journal of Advanced Research in Computer Science and Software Engineering

Secure Fingerprint Matching with External Registration

Extremely-Large-Scale Biometric Authentication System - Its Practical Implementation

NOVATEUR PUBLICATIONS INTERNATIONAL JOURNAL OF INNOVATIONS IN ENGINEERING RESEARCH AND TECHNOLOGY [IJIERT] ISSN: VOLUME 2, ISSUE 1 JAN-2015

Dynamic Stroke Information Analysis for Video-Based Handwritten Chinese Character Recognition

Discovering Computers Chapter 5 Input

Biometric Cryptosystem Using Online Signatures

Multimodal Biometrics for Voice and Handwriting

EBOOK 4 TIPS FOR STRENGTHENING THE SECURITY OF YOUR VPN ACCESS

Gaithashing: a two-factor authentication scheme based on gait features

The Future of Authentication

Signature Verification: Why xyzmo offers the leading solution

Using Biometric Authentication to Elevate Enterprise Security

Obtaining Receiver Operating Characteristic Curves from Commercial Biometric Systems

Off-line signature verification: a comparison between human and machine performance

Spatial Topology of Equitemporal Points on Signatures for Retrieval

DOUBLE CHECKING: MULTIMODAL, INTEGRATIVE & CONTINUOUS VERIFICATION TECHNOLOGY OF SIGNATURE & FINGER PRINT IDENTIFICATION

An Efficient Secure Multimodal Biometric Fusion Using Palmprint and Face Image

BIOMETRIC MECHANISM FOR ONLINE TRANSACTION ON ANDROID SYSTEM ENHANCED SECURITY OF. Anshita Agrawal

RULE BASED SIGNATURE VERIFICATION AND FORGERY DETECTION

Signature Recognition by Pixel Variance Analysis Using Multiple Morphological Dilations

Face Detection Using Radial Basis Function Neural Networks with Fixed Spread Value

The Analysis of Traffic of IP Packets using CGH. Self Organizing Map

Robust color segmentation algorithms in illumination variation conditions

Identification, authentication, authorisation. Identification and authentication. Authentication. Authentication. Three closely related concepts:

EVALUATING DIFFERENT TOUCH-BASED INTERACTION TECHNIQUES IN A PUBLIC INFORMATION KIOSK

Off-line Signature Verification Using Contour Features

Repositorio Institucional de la Universidad Autónoma de Madrid.

Touchless Fingerprint recognition using MATLAB

IDENTIFICATION THINK EXCELLENCE, CHOOSE MULTIMODALITY

Information Security Identification and authentication. Advanced User Authentication II

WiFi concierge at home network focusing on streaming traffic

How accurate is AGNITIO KIVOX Voice ID?

CS 528 Mobile and Ubiquitous Computing Lecture 11b: Mobile Security and Mobile Software Vulnerabilities Emmanuel Agu

International Journal of Scientific & Engineering Research, Volume 7, Issue 11, November ISSN

A Visualization Tool to Improve the Performance of a Classifier Based on Hidden Markov Models


International Journal of Computer Science Trends and Technology (IJCST) Volume 5 Issue 5, Sep Oct 2017

CAN WE ESCAPE PASSWORDS?

Mahmood Fathy Computer Engineering Department Iran University of science and technology Tehran, Iran

User Signature Identification and Image Pixel Pattern Verification

Keywords Wavelet decomposition, SIFT, Unibiometrics, Multibiometrics, Histogram Equalization.

Semi-Fragile Watermarking in Biometric Systems: Template Self-Embedding

Ujma A. Mulla 1 1 PG Student of Electronics Department of, B.I.G.C.E., Solapur, Maharashtra, India. IJRASET: All Rights are Reserved

CHAPTER 6 EFFICIENT TECHNIQUE TOWARDS THE AVOIDANCE OF REPLAY ATTACK USING LOW DISTORTION TRANSFORM

Transcription:

A Biometric Authentication System That Automatically Generates Feature Points Hiroshi Dozono 1, Youki Inaba 1, Masanori Nakakuni 2 1 Faculty of Science and Engineering, Saga University, 1-Honjyo Saga, 840-8502 JAPAN 2 Information Technology Center, Fukuoka University, 8-19-1, Nanakuma, Jonan-ku, Fukuoka 814-0180 JAPAN Abstract Recently, personal information in the mobile devices have been threatened as the popularization of mobile devices because of the lack attention to the information. The purpose of our research is to develop the reliable and convenient authentication system for mobile devices. In this paper, the development of the biometric authentication system, which generates feature points from freehand pattern and uses the points as the anchors for drawing pattern and points for detecting pen speed, is introduced using the experimental results. Keywords: Biometrics, Authentication, Touch panel, Mobile device, Tracing Authentication 1. Introduction Recently, computerization of the information has proceeded very rapidly. Along with the computerization, the mobile devices, such as smartphones and tablets, are spread among people very rapidly. For example, about 70% of Japanese will possess smartphones as ubiquitous devices in 2016. However, the owners of mobile devices do not take care of the security of the personal information which stored in the mobile devices. The development of the mobile devices targets the performance or usability, and security is trampled upon. For this problem, the authentication method that is flexible and convenient for the users is proposed. If the authentication method is not annoying and does not require special knowledge about security, the user will naturally use the authentication system, and will have consciousness to the security. 2. Authentication Methods using Touch Panel As the popularization of the mobile devices such as smartphones and tablet devices, the devices which equipped with touch panel become widespread. And, these devices grow in usage and popularity with the people who are not familiar with the conventional computers. In these situations, a simple authentication method which uses a touch panel is desired. As the authentication method using touch panels, the signature written on the touch panel is often used[1]. However, it is difficult to write identical signature on slippery touch panel especially for capacitive type panel that is recently equipped to almost all mobile devices. As another method, the authentication method which uses the knowledge factors in selecting the symbols or positions in the image is often used[2]. However, this method has a weakness for snooping. Android devices use the lock screen which uses the knowledge factor with connecting the points displayed on the touch panel (Figure 1). However, the patterns which can be drawn are not flexible, and users tend to select simple patterns because the points are fixed. Thus, this method also has a weakness for snooping. For avoiding snooping, Fig. 1: Android lock screen biometric authentication[3] is effective. Biometric authentication is classified into two types; biometric authentication using biological features and biometric authentication using behavioral features. As the biological features, the fingerprints, vein patterns and iris patterns are often used. They can achieve high accuracy. However, special sensor devices, such as fingerprint readers, are required for implementation. As the behavioral features, keystroke timings[4][5] and calligraphy of handwritten patterns or signature[1] are often

used. They can be obtained from conventional input devices. However, the accuracy of authentication is lower than that of biological features. At the same time, high accuracy is not always necessary for the devices that are used personally because authentication is used as the lock method just in case the device is stolen or possessed by malicious users. In this paper, the biometric authentication method which uses the biological features obtained from touch panel is proposed. We have proposed an authentication system using the behavior biometrics during drawing the symbol displayed in the touch panel[6]. This system uses pen speed and pen pressure at all sampling times as behavior biometrics, and marks 0.1 as Equal Error Rate (EER). However, capacitive type panel, which is mostly used recently, cannot detect pen pressure, and it requires much computational cost for matching all pen speeds and pen pressures at all sampling time. For this problem, we propose an authentication system which generates feature points automatically. as a behavior factor for biometric authentication. The integration of multi-modal biometrics is reported in [7][8], and the combination of knowledge factor and behavior factor is reported in [9]. The feature points are generated on the points which have large curvature factors of curving line. To detect both of shelving curve and sharp curve simultaneously, the curvature factors are calculated in long intervals and short intervals. The detection of feature points in short interval tends to generate dense points due to jiggling of the finger or touch pen. To avoid this problem, the dense points are integrated to one point. Figure 3,4 shows the examples of generating feature points. In case 1, the feature points are successfully gen- 3. The System Which Generates Feature Points At first, we developed a system which generates feature points. As the environments of development, Xcode 3.2.6 and ios SDK 4.3 is used, and ipad is used as the target machine. As shown in Figure 2, this system generates feature points automatically from the freehand curving line. These Fig. 3: Generation of feature points(case 1) Fig. 4: Generation of feature points(case 2) erated in both shelving curve and sharp curve. In case 2, the feature points generated closely during drawing circle twice are integrated Fig. 2: The system which generate feature points 4. Authentication using Knowledge Factor points are displayed on the touch panel during authentication to increase reproducibility of the registered curving line. By displaying the points without curving line, the pattern of connecting points is used as a knowledge factor for authentication. The pen speeds between the points are used In this section, the results of authentication experiments using the connection pattern of feature points as knowledge factor are reported. The pattern is represented in the order of the connections of feature points. It faces the possibility of two problems. At first, the pattern may be guessed from the feature points displayed on the touch panel. Secondly, the

user cannot remember the pattern from the feature points. The first problem is examined in this section. The settings of the experiment are as follows. 3 patterns of feature points shown in Figure 5 are registered by users. Ten users triy to guess the pattern from the feature points displayed on the screen. Table 1 shows the results. The number in the threshold values for authentication are set, and authentication succeeds if all differences of pen speeds are less than the thresholds. As the behavior factor varies in each drawing, and the dispersion is different for each user, the threshold is set as ασ according to the variance σ 2 of the pattern that is traced three times in registration phase. The procedure of authentication experiment is as follows. Step 1: Generation of Feature Points The user, who is registered on the system, draws a curving line on touch screen. Then, the feature points are generated as mentioned in Section III, and the connecting pattern is registered for the user as shown in Figure 6. The name of the user is set in the text box on the top of screen for later analysis using the recorded data. Fig. 5: Registered patterns of feature points Table 1: Result of authentication experiment using knowledge factor A B C Complete 0 0 0 Starting point 0 0 2 table denotes the number of successful users. No user can completely guess the patterns. And, only two users can guess the starting point of pattern C. From this experiment, the authentication using knowledge factor is secure enough. However, if the number of feature points decreases or the registered pattern becomes too simple, it may not secure. An algorithm which evaluates the security of registered pattern is required in practical use. 5. Biometric Authentication using Behavior Factor of Pen Speed We have reported the biometric authentication using behavior factors of pen calligraphy of pen speed and pen pressure[6]. However, recent capacitive touch panel, which is equipped with almost all recent devices, cannot detect pen pressure. Considering the compatibility, the biometric authentication using pen speed is effective. However, it may be not secure enough as biometrics, thus the authentication using the knowledge factor of connecting patterns is simultaneously used. In [6], the distribution of the pen speed along the curving line is used as feature value. However, it requires large computation costs because the number of points tends to be large, and it may be problematic for mobile devices. In this paper, much simpler method is used. The pen speeds between the feature points are used as feature value. For each connection between feature points, Fig. 6: Registration screen Step 2: Registration of Behavior Factor of Pen Speed The user traces the pattern on the touch screen which displays the feature points without curving lines in three times. Then, the averages and variances of pen speeds between the feature points are registered as a behavior factor of pen speed. Step 3: Authentication The user who is authenticated draws the curving line on the touch screen which displays the feature points without curving line. Then, the connecting pattern and pen speeds between the feature points are examined, and if both of knowledge factor and behavior factor meet the condition of authentication, the user is authenticated as true user. Next, the experimental results using a behavior factor of pen speed are reported. As the indexes of evaluation, FRR and FAR are used. FRR is the False Rejection Rate, which is the rate of falsely rejecting the true user as false user. FAR is the False Acceptance Rate, which is the rate of falsely accepting the false user as true user. At first, the factor α

in the threshold value is set during the experiment of the authentication of the author as to achieve less than 10 % for FRR as shown in Table 2. To examine the effectiveness of Table 2: Result of authentication experiment using pen speed of author A B C FRR 0.069 0.100 0.095 behavior factor, the experiments are conducted with giving the knowledge factor of connecting patterns to all users. Thus, following results of experiments denote the security of behavior factor without knowledge factor. At first, experiments are conducted with showing the connecting pattern with straight lines displayed between the feature points as shown in Figure 7. 6 users A-F register The number in brackets denotes the number of feature points. For all users, FRR becomes less than 0.1, and FAR is also small except user D. The EER is not calculated in these experiments. However, it will be more or less than 0.05 on average. Next, the experiments with showing the authentication process of true user to false users are conducted. In the practical case, the false user may snoop the pattern just before spoofing the true user, and can remember the curving line in authentication process. In these experiments, the lines connecting the feature points are not displayed on authentication screen as in Figure 9. Figure Fig. 9: Authentication screen without lines 10 shows the results. User A-F are not identical to those in Fig. 7: Authentication screen with lines the pattern, and for each calculation of FAR and FRR, more than 10 trials are conducted. Figure 8 shows the results. Fig. 10: Result of authentication experiment using behavior factor(2) - under snooping Fig. 8: Result of authentication experiment using behavior factor (1) the previous experiments. In this case, FAR becomes worse because the pattern is snooped. However, the average is less than 0.1, and EER will be more or less than 0.08 on average. In both experiments, FAR becomes larger for smaller number of feature points. Thus, an algorithm which evaluates the security of registered pattern is required also for behavior factor in practical use.

These experimental results are the extreme case in which the connecting pattern is known by false users. In the practical use, the combination of knowledge factor and behavior factor will meet secure authentication. 6. Conclusion In this paper, we propose the authentication which generates the feature points from the handwritten curving line automatically. The pattern of connecting feature points is applied to the authentication using a knowledge factor, and pen speeds between the feature points are applied to the authentication using a behavior factor. The effectiveness of both authentication is examined by experiments, and the combination of the authentication will accomplish secure system. As the future work, the authentication accuracy should be more strengthened. In this paper, simple algorithm using threshold is applied. More sophisticated algorithm like neural network should be applied to authentication system. Recently, the cloud system using HTML 5 WEB applications is spread to mobile devices. The application of this method to WEB applications will achieve the security on cloud system. References [1] J. J. Brault and R. Plamondon: A Complexity Measure of Handwritten Curves: Modelling of Dynamic Signature Forgery, IEEE Trans. Systems, Man and Cybernetics, 23:pp.400-413(1993) [2] Hiroshi Dozono, Takayuki Inoue and Masanori Nakakuni,et.al, Study of Biometric Authentication Method using Behavior Characteristics on Game Consoles, Proc. of SAM 2009, (2009) [3] R. Bolle, J. Connell, S. Pankanti, N. Ratha, and A. Senior, Guide to Biometrics, Springer, 2004 [4] F. Monrose and A.D. Rubin: Keystroke Dynamics as a Biometric for Authentication, Future Generation Computer Systems, March(2000). [5] H. Dozono and M. Nakakuni et.al, The Analysis of Key Stroke Timings using Self Organizing Maps and its Application to Authentication, Proceedings of the International Conference on Security and Management 2006, pp.100-105(2006) [6] Hiroshi Dozono and Masanori Nakakuni, et.al: The Analysis of Pen Pressures of Handwritten Symbols on PDA Touch Panel using Self Organizing Maps, Proceedings of the International Conference on Security and Management 2005, pp.440-445(2005) [7] S.Dokic,A.Kulesh,et.al, An Overview of Multi-modal Biometrics for Authentication, Proceedings of The 2007 International Conference on Security and Management, pp.39-44(2007) [8] Hiroshi Dozono and Masanori Nakakuni et.al: An Integration Method of Multi-Modal Biometrics Using Supervised Pareto Learning Self Organizing Maps., Proc. of the Internal Joint Conference of Neural Network 2008,(2008) [9] Hiroshi Dozono, Takayuki Inoue and Masanori Nakakuni, A study of the Graphical User Interfaces for Biometric Authentication System, Proc. of SAM 2012, (2012)