Dynamic footprint-based user identification system : a novel biometric technique for ubiquitous computing environment

Abstract

The paradigm for using computer is slowly changing from one person-one computer to one person-many computers. This kind of change is named by ubiquitous computing and to realize ubiquitous computing, three key words should be considered, everywhere, invisible and unconsciously. In this paper, a user identification method for the realization of personalized service for ubiquitous home environment is introduced. Biometrics is the automated use of physiological or behavioral characteristics to determine or verify identity. And in view of ubiquitous home environment, easy-to-use biometrics such as gait or footprint is more appropriate than others. But, use of only gait or footprint in real application is not easy by its low recognition rate. This dissertation addresses a new biometrics, dynamic footprint which is defined as a time sequence of footprints during a single gait cycle. So, dynamic footprint is 2D projected information of 3D human walking action and includes both footprint shape information and gait information together. Based on dynamic footprint, three conspicuous problems are considered. One problem is to make a feature, from dynamic footprint, which can represent human footprint shape and to develop a user identification method via this shape feature. Another problem is to make a feature, dynamic footprint, which can represent human gait information and to develop a user identification method via this gait feature. And, the other problem is to develop a combined method which can handle both shape information and gait information together. For the first problem, user identification via footprint shape, we analyzed the previous footprint-based user identification methods and found a fact that the footprint during stepping gives more robust results than the footprint during standing since the most important part in a foot for person identification is the toe part. In addition, the overlapped footprint during a single gait cycle is very similar to the st...