Multiple Extended Targets Tracking Based On 3-Dimensional Projection TBD And Trajectory Maintenance

With the rapid development of modern radar technology,certain new systems such as high precision radar and double-sided array radar have brought many performance advantages.But they also bring some problems.Typical multi-target tracking systems assume that in every scan there is at most one measurement for each target.With the emergence of high precision radar,the echo signal of the same target may be distributed in multiple detection units of the radar,which makes the hypothesis of point target no longer hold true.In other words,multiple measurements may be made on the same target at a scanning time.This special target is often referred to as an extended target.The traditional multi-target tracking technology based on point model is often unable to adapt to the extended model state,so the performance of traditional algorithm is not good.On the other hand,the efficiency of target tracking is greatly improved by two-sided array radar that have the unique structure of two kinds of polarization modes.At the same time,the radar can avoid to blind for weak targets to a certain degree.Because the two antennas of the radar are placed back to back,the measurement rate of the radar is twice that of ordinary radar,but the number of clutter point is also twice that of ordinary radar.An urgent problem for two-sided array radar is how to perform multi-target tracking in dense clutter.In this paper,we propose a new algorithm to effectively track multiple extended targets in complex environments for both high resolution radar and two-sided array radar.The algorithm consists of three steps.The first step,the spatiotemporal clutter map is set up to suppress clutter.In this paper,the spatiotemporal clutter map algorithm is a data processing method.Firstly,a clutter density map is established by using measurements from the past time.The clutter area in the monitoring area can be found by the clutter density map.Then,combined with the clutter density map,the current measurements in different regions is given to different weights.The algorithm is performed to clutter suppression for both single radar and multiple radar.The algorithm can effectively reduce false track establishment.The second step,a 3-dimensional projection based TBD(3DP-TBD)is to address the problem of target extension,weak targets and dense clutter.The tracklets in straight lines can be obtained by the 3DP-TBD.Firstly,the proposed detection algorithm bases its principle on track before detection(TBD)that it has been proven to be effective in tracking weak targets in dense clutter.Secondly,in order to realize the detection and tracking of the target,the idea of 3d projection can make use of the characteristics of the linear motion state of the target to quickly accumulate the point of the target in the time dimension.The method is suitable for both detection and tracking of point target and extended target.But the disadvantage of this method is that it can only detect straight track.In the third step,the algorithm of track association and maintenance is performed to integrate into a maneuvering trajectory which is formed by tracklets in straight lines.According to the difference of track association strategy,two kinds of algorithms are proposed in this paper.One is the algorithm of track association based on the exhaustive method,and the other is the algorithm of track association based on the lion reproduction model.The latter method is an improvement on the former.By using the idea of meta-heuristic algorithm and the lion propagation model,the efficiency of track association is improved and the computation is reduced.Finally,the proposed algorithm is compared with the two PHD algorithms.A large number of synthetic data and the real data are tested to verify the superiority of the proposed approach.