| Multi-target tracking(MTT)is one of the hotspots in the field of multi-sensor data fusion,and it has been widely used in military and civil fields.The traditional MTT method is based on classical probability theory,which is mainly based on solving multi-target data association problem.In complicated environment,the tracking problems are usually affected by unknown target number,dense clutter,low detection rate,etc.,and results degradations in tracking accuracy.In recent years,the method of probability hypothesis density(PHD)based on random finite set(RFS)arises more attentions.By using RFS theory,the target state and measurements are described in a probability hypothesis density space,thus effectively avoids the data association problem.However,most current RFS based multi-target tracking methods are proposed for single-sensor application.Obviously,it should be wise to achieve the estimation results through fusing multi-sensor information but how to construct an effective and efficient multi-sensor estimator is not a trivial problem.To this end,this thesis focuses on the multi-sensor multi-target tracking problem with dense clutter and low detection rate.The main work and research results are as follows:1)In order to reduce the tracking effect of single sensor PHD filter in high-clutter environment,a distributed multi-sensor data fusion structure model is constructed,and an adaptive multi-sensor data fusion algorithm based on GMPHD filter is proposed.The simulation results show that the proposed algorithm improves the tracking accuracy compared with single sensor.2)The limitation of conventional track fusion algorithm in different clutter environment and detection rate limits the improvement of tracking effect.To solve this problem,a distributed multi-sensor data fusion model with feedback is proposed,and two different multi-sensor PHD fusion algorithms are proposed: max fusion algorithm and product fusion algorithm.Simulation results on different scenes show that the proposed algorithm outperforms the traditional methods.3)A multi-sensor IMM-GMPHD filter algorithm for multi-maneuvering target tracking is proposed,which can be applied to multi-maneuvering target tracking problem.It effectively deals with multi-maneuvering target tracking in clutter environment.Simulation results show that the proposed algorithm can achieve highertarget state estimation accuracy when the target is maneuvering. |
