CAJ | 학술논문

针对多目标跟踪中概率假设密度(PHD)滤波器在杂波模型与先验知识不匹配情况下滤波性能急剧下降的缺点，将增广状态空间引入PHD滤波器，提出了一种新的未知杂波环境下的 PHD滤波器。该滤波器利用增广状态空间区分目标状态空间与杂波状态空间，通过量测对杂波模型进行估计，不需要杂波先验知识，避免了因杂波强度的先验知识选择不当而造成PHD滤波器跟踪性能下降的问题。仿真结果表明，该算法在未知杂波环境下，具有稳定的跟踪效果；在保证实时性的前提下，其跟踪精度与传统 PHD滤波器在杂波模型匹配情况下相当。
침대다목표근종중개솔가설밀도(PHD)려파기재잡파모형여선험지식불필배정황하려파성능급극하강적결점，장증엄상태공간인입PHD려파기，제출료일충신적미지잡파배경하적 PHD려파기。해려파기이용증엄상태공간구분목표상태공간여잡파상태공간，통과량측대잡파모형진행고계，불수요잡파선험지식，피면료인잡파강도적선험지식선택불당이조성PHD려파기근종성능하강적문제。방진결과표명，해산법재미지잡파배경하，구유은정적근종효과；재보증실시성적전제하，기근종정도여전통 PHD려파기재잡파모형필배정황하상당。
Aiming at improving the poor performance of the Probability Hypothesis Density(PHD) filter when the clutter model and the prior knowledge are mismatched,a novel PHD filter into which we introduce the augmented state space and which is used under the unknown clutter circumstance is proposed in this paper.The proposed filter can distinguish the target state space and the clutter state space by the augmented state space.Using the estimate of the unknown clutter model from the measurement,the filter can avoid the tracking performance reduction caused by the improper model selection of the unknown clutter.Simulation results show that the proposed algorithm can achieve a stable tracking performance under the unknown clutter circumstance and a tracking accuracy equal to that of the conventional PHD filter used in the unknown clutter circumstance in the real-time context.