Multifactor biometric sketch authentication

Abstract

In this paper we propose a multifactor biometric sketch authentication method based on biometric sketch recognition and a user's personal knowledge about the sketch`s content, which is negotiated between the biometric authentication system and the user during enrollment. The used sketch recognition algorithm is based on the active shape structural model (ASSM) for analyzing the structural variability of sketches built up from a set of deformable shapes. For increasing the reliability of the biometric sketch authentication method the user's knowledge as authentication factor has been added by fulfilling specific sketching tasks of varying complexity given by the authentication system. An evaluation and testing framework for biometric algorithms was used to prove the accuracy of the method. For this purpose the biometric sketch algorithm has been adapted to the framework, a compiled sample database for comparability testing between users has been generated, and attack classes ranging from none, over partial to complete knowledge about the user's sketch has been developed and used. The evaluation of the test results shows that particularly the user's knowledge as an added authentication factor leads the used sketch recognition algorithm to high accuracy.