An innovative fingerprint feature representation method to facilitate authentication using neural networks

Abstract

Authentication systems enable the verification of claimed identity; on computer systems these are typically password-based. However, such systems are vulnerable to numerous attack vectors and are responsible for a large number of security breaches. Biometrics is now commonly investigated as an alternative to password-based systems. There are numerous biometric characteristics that can be used for authentication purposes, each with different levels of accuracy and positive and negative implementation factors. The objective of the current study was to investigate fingerprint recognition utilizing Artificial Neural Networks (ANNs) as a classifier. An innovative representation method for fingerprint features was developed to facilitate verification by ANNs. For each participant, the method required the alignment of their fingerprint samples (based on extracted local features), and the selection of 8 of these aligned features common to their samples. The six attributes belonging to each of the selected features were used for ANN input. Unlike the common usage, each participant had one dedicated ANN trained to recognize only their fingerprint samples. Experimental results returned a false acceptance rate (FAR) of 0.0 and a false rejection rate (FRR) of 0.0022, which were comparable to (and in some cases, slightly better than) other research efforts in the field. © Springer-Verlag 2013.