Biometric authentication using finger nail plates

Abstract

This paper attempts to bring a new inventive and non-mainstream biometric development to the fore. A completely automated and unified approach to authenticate individuals using finger nail plate surface images has been proposed. There has not been any attempt in utilizing the texture and the contour information of the nail-plate for human authentication in literature. This has motivated us to explore the nail plate based identification for security applications and applying approaches that ascertain the best possible performance. The complex technique of Interferometry is perhaps the most widely used approach in the literature to carry out analysis on nail-bed which is the inner part of the nail unit. In this paper, we propose a very convenient and efficient method by acquiring low resolution images of nail plate surface which is the outermost part of the nail unit. The contour and texture characteristics of nail plates from three fingers are represented by the appearance and shape based feature descriptors. The paper presents two ways of integrating the nail-plate features from three fingers: (1) score level rules for fusion of matching scores and (2) the classifier based fusion of matching scores by employing decision tree and support vector machines. The experimental results from 180 users and a total of 2700 nail plate images validate the contributions from this paper.