A user identification system using signature written with mouse

Abstract

A user identification system is very important for protecting information from illegal access. There are identification systems using standard devices (keyboard or mouse) and systems using special devices. A user identification system using mouse is proposed in [6]. In their system, users write a simple figure object and the successful verification rate is 87%. However the simple object is too easy to prevent impersonation. In order to realize a more reliable user identification system using mouse, we propose a new system to identify users using a complex figure object, signature. New techniques we utilize in our system are as follows: the normalization of input data, the adoption of new signature-writing-parameters, the evaluation of verification data using geometric average means and the dynamical update of database. We have implemented our user identification system and conducted experiments of the implemented system. The successful verification rate in our system is 93%.