Ground moving target tracking with VS-IMM particle filter based on road information

Tracking target moving on ground is constrained by terrain, especially the road segments which may intersect at one junction. The ground target moves along the road generically, this paper proposes a variable structure interacting multiple model particle filter (VS-IMM-PF) algorithm based on road information. Models in filters based on state estimation process or process noise correction is determined by whether the target is near one junction, which is judged through uncertainty region. In simulation, a moving ground moving target indication (GMTI) sensor is used in air-to-ground tracking. It´s showed that the advanced algorithm can track ground moving target effectively, of which accuracy is improved, compared with the algorithm only using process noise correction. In addition, the operation time decreases obviously as the number of particles reduces.