An Embedded Background Modeling Method for Detecting Object Left Behind and Very Still Person

摘要

This paper proposes an embedded background modeling method for detecting object left behind and very still person in public places. In doing so, we employ a new background subtraction algorithm based on multivariate background models in which bivariate background modeling technique is reinforced by embedding a stochastic background. Moreover it utilizes logical and statistical analysis to differentiate between objects left behind and very still persons. We evaluate the robustness and efficiency of the method by using our own video sequences taken in an international airport and our University Campus as well as PETS 2006 datasets.