Online biometric authentication using hand vein patterns

Abstract

Online authentication is one of the basic requirements for any biometric based authentication system used in civil and commercial applications. This paper presents a new approach for online biometric authentication using hand vein patterns. In contrast to the existing approaches, our online authentication system utilizes infrared thermal images of hand vein patterns for authentication purposes. A robust peg free camera set up is employed for infrared thermal imaging. A region of interest (ROI) is extracted from the vein patterns to convolve with Gabor filter for improving the visibility of vein pattern. The outcome of this convolution is the real and imaginary parts of which only the real part is regarded as a texture. Gabor Wavelets at different orientations are convolved with the real part after partitioning it into non-overlapping windows to extract texture. The mean of the convolution on each window is taken as a feature. The experimental results on 100 users conform to the false acceptance error rate (FAR) of 0.1% for the genuine acceptance rate (GAR) of 98.5%.