Energy Minimization Model Based Target Tracking

Abstract

This paper proposes a novel method to deal with the target tracking. Specifically, the information of observation model, appearance model, exclusion model, dynamic model, trajectory persistence model and trajectory regulation model are first used to construct objective tracking functions; then, the gradient descent method is adopted to achieve an approximate minimum of the constructed objective functions, and to obtain the status of tracking targets; finally, continuous energy minimization based intelligent extrapolation method is utilized to obtain the final continuous and smooth trajectories. Experimental results on PETS 2009/2010 benchmark video database demonstrate the effectiveness and efficiency of the proposed scheme.