Gait recognition based on time-frequency analysis

Abstract

In this paper, we extract model-based gait features and investigate the time-frequency representations of the feature signals. A novel gait recognition approach is proposed, which is based on time-frequency analysis of gait feature signals using the Wigner distribution. Time-frequency analysis using theWigner distribution is aimed at capturing gait information that is not extractable using other time-domain or frequency-domain techniques. Experiments, conducted based on the above approach, yielded encouraging results.