Footstep modeling and detection using locally stationary autoregressive model

It is known that footsteps are useful to estimate human states. Recent years, the footstep analysis based on the Autoregressive (AR) model is adopted to achieve footstep based surveillance system. Accurate footstep analysis would be powerful tool in various applications, home security service, surveillance and understanding of human action since the gait expresses personality, age and gender. The conventional footstep analysis models the footstep using AR model whose the order of AR coefficient and length of analysis window is fixed. This means that the suitable AR model for footstep analysis is not discussed. In this paper, we apply the Locally Stationary Autoregressive (LSAR) model and AIC to perform an appropriate footstep analysis.