基于粒子滤波与Gibbs抽样的多目标被动跟踪研究  

Multi-Target Tracking Based on Particle Filter and Gibbs Sampling

作　　者：孟范栋[1] 黄文斌[1] 

机构地区：[1]海军潜艇学院,山东青岛266071

出　　处：《电光与控制》2010年第12期1-4,共4页Electronics Optics & Control

基　　金：国防预研基金(401040101)

摘　　要：多目标跟踪问题是目前目标跟踪领域的一个重要研究方向,其中被动跟踪更加复杂也更具有实战意义。为了提高多目标跟踪算法的精度及稳定性,减少算法的计算开销,采用了粒子滤波与Gibbs抽样相结合的方法。粒子滤波能很好地解决目标跟踪中状态估计的非线性问题,将其应用扩展到多目标的跟踪维持;运用Gibbs抽样解决多目标跟踪中的数据关联问题,提高了关联的准确度,减小了计算开销。仿真试验证明:上述算法能较好地解决多目标跟踪问题,具有较好的估计精度。Multi-target tracking is a focus of study in target tracking area at present, in which the passive tracking is complicated and very useful in actual combat. In order to improve the accuracy and stability of multi-target tracking algorithm and to reduce computation cost, the particle filtering was used together with Gibbs sampling method. The particle filter can resolve the problem of nonlinearity of state evaluation in target tracking, and extend it to the domain of multiple targets tracking. The Gibbs sampler was used for data association to reduce the computation cost and increase the accuracy of association. Simulation results demonstrated the availability of the studied tracking algorithm.

关 键 词：多目标跟踪 被动跟踪 粒子滤波 吉布斯抽样 

分 类 号：V271.4[航空宇航科学与技术—飞行器设计]