Biometrics and Bioinformatics

Aditya Abhyankar ¹, Amit Vijayat ², Sunil Kumar ³, Stephanie Schuckers ⁴

Affiliations
1 Vishvakarma Institute of Information Technology, Pune, IN
2 Hyderabad, IN
3 Electrical and Computer Engineering Department, San Diego State University, San Diego, CA
4 Clarkson University, NY, US

Abstract

Securing biometric information has become essential with growing biometric applications in different sectors of society. Vulnerability assessment plays a key role in improving the security of any security system by identifying the potential vulnerabilities and proposing countermeasures to mitigate the threats posed by them. In this work self-generated and dynamic helper data based system is proposed to encrypt the biometric templates. Biometric information is statistically learned and probabilistic matching is performed to discriminate genuine from imposters. We call this system as One Time Biometric Transformation (OTBT) system. The system was tested using CASIA iris database and by probabilistic matching an EER of 1.96% is achieved. Strength analysis of the system for three different challenging databases is also presented.

Full Text

     

Aditya Abhyankar ¹, Stephanie Schuckers ²

Affiliations
1 Vishvakarma Institute of Information Technology, Pune, IN
2 Clarkson University, NY, US

Abstract

By virtue of using simple, inexpensive techniques it is possible to delude fingerprint scanners. An antidote against such fraudulent attacks is suggested in this paper. This work emphasizes on quantifying liveness for fingerprint scanners using wavelet energy analysis of fingerprint images. From the images captured at two time stamps the perspiration changes along the finger ridges are sensed and used as liveness measure. Multi-resolution analysis and wavelet packet analysis is performed for low and high frequency data. Orthogonal Daubechies filters are designed to analyze the non-stationary information efficiently.Only those wavelet coefficient which experience an energy change of more than 40% are used for liveness after normalization. The proposed algorithm was applied to a data set of approximately 58 live, 50 spoof and 28 cadaver fingerprint images captured at 0 sec and 2 or 5 sec. The algorithm was tested for different underlying fingerprint sensing technologies including optical (secugen), optoelectrical (ethentica) and capacitive DC (precise biometrics). Classification results are presented for all the scanners and for both the time windows.

Keywords

Fingerprints, Spoofing, Liveness, Wavelet Analysis, Multiresolution Analysis, Wavelet Packet Analysis, Energy Distribution Analysis.

Full Text