无线层析网络中基于粒子滤波的时变多目标跟踪

发布时间：2018-02-25 23:03

本文关键词： 粒子滤波时变多目标跟踪无线层析成像最佳子模式分配　出处：《北京理工大学学报》2017年05期 　论文类型：期刊论文

【摘要】：研究将粒子滤波(PF)理论应用于无线层析网络中的时变多目标跟踪(MTT).传统的基于无线层析成像(RTI)的时变多目标跟踪方法存在延迟问题,即与真实的时变目标数目相比,估计所得时变目标数目存在滞后,并且跟踪精度较低.基于PF的时变多目标跟踪方法利用维度可变的粒子来估计目标数目,实现目标跟踪.该方法不存在延迟问题,并且能提高目标的跟踪性能.研究通过在一个9.5m×9.5m的监测区域内进行实验来验证该算法的有效性.实验结果表明基于RTI的时变多目标跟踪方法的最佳子模式分配(OSPA)误差为0.485m,而基于PF的时变多目标跟踪方法的OSPA误差为0.362m,其性能比基于RTI的方法提高了25%.
[Abstract]:This paper studies the application of particle filter theory to time-varying multi-target tracking in wireless tomography networks. The traditional time-varying multi-target tracking method based on wireless tomography (RTI) has the problem of delay, that is, compared with the real number of time-varying targets. The estimated number of time-varying targets has lag, and the tracking accuracy is low. The PF based time-varying multi-target tracking method uses variable particles to estimate the number of targets to achieve target tracking. There is no delay problem in this method. The effectiveness of the algorithm is verified by experiments in a monitoring area of 9.5 m 脳 9.5 m. The experimental results show that the optimal sub-pattern allocation of time-varying multi-target tracking method based on RTI is OSPA). The error is 0.485m, while the OSPA error of PF based time-varying multi-target tracking method is 0.362m. its performance is 25% higher than that based on RTI.
【作者单位】：北京理工大学信息与电子学院;
【基金】：国家自然科学基金资助项目(61101129,61227001,61471045) 北京青年英才计划(YETP1182)
【分类号】：TN713;TN92

【相似文献】