Target tracking based on audio and video information fusion

Abstract

Particle filter is an effective technique to deal with the state estimation of nonlinear and non Gauss dynamic systems. Aiming at the problem of moving target tracking under the condition of illumination and occlusion, a particle filter algorithm based on audio and video information fusion is studied. This algorithm overcomes the deficiency of the algorithms based on a single signal source by using the time—space relativity and the complementarity of the audio and video. Pedestrian tracking experiment based on audio and video information fusion shows that the fusion algorithm is more stable and accurate than the single video tracking algorithm.