Models and methods for biometric motion identification Bartosz

Abstract

Human motion is a complex signal with many different properties depending on various factors: age, gender, physical condition, emotions etc. Nevertheless there is a hypothesis which claims that human motion can be a source for biometric analysis and person identification. In the paper some methods to analyze and compare different motions are presented. Methods are examined for usefulness in motion identification. We distinguish time-series and frequency analysis for rotational signals describing mainly the motion of legs. The results of experiments are presented taking into consideration different motion representations.