Optimisation of Geometric and Holistic Feature Extraction Approaches for a Footstep Biometric Verification System

This paper reports some experiments which assess the potential of footsteps as a biometric. We present a semiautomatic capture system and report results on a database with independent development and evaluation datasets comprised of more than 3000 footsteps collected from 41 persons. We present an optimisation of geometric and holistic features extraction approaches. An equal error rate of 10% is obtained with holistic features classified with a support vector machine. The database is freely available to the research community.