基于联合GLMB滤波器的可分辨群目标跟踪

针对联合广义标签多伯努利(joint generalized labeled multi-Bernoulli, J-GLMB)滤波算法中群目标之间距离较近、容易关联错误的问题,结合超图匹配(hypergraph matching, HGM)提出一种基于HGM-J-GLMB滤波器的可分辨群目标跟踪算法。首先,采用J-GLMB滤波器估计群内各目标的状态、数目及轨迹信息,并利用HGM结果提升量测与预测状态之间的关联性能。其次,通过图理论计算邻接矩阵,获取群结构信息和子群数目。随后,利用群结构信息估计协作噪声,进而校正目标的预测状态。最后,通过平滑算法改善滤波效果,并设置轨迹长度阈值,使其在平滑状态达到消除短轨迹的目的。仿真实验表明,所提算法在线性系统下能有效提升群目标跟踪性能。

基于GLMB滤波的复杂场景下红外弱小目标自适应跟踪算法

针对红外弱小目标在复杂场景下受到漏检和杂波影响,导致跟踪不连续甚至失效的问题,本文提出一种红外弱小目标自适应跟踪算法。在预处理阶段,为了减少不必要的计算,首先定义一种衡量图像复杂度的算法。然后该算法通过计算红外图像多个特征得到场景复杂度来确认场景类型,再根据场景类型选取对应的检测算法提取目标候选位置、灰度以及局部直方图等特征建立对应的量测模型与似然函数。在目标跟踪阶段,为了自适应地匹配广义标签多伯努利(Generalized Labeled Multi-Bernoulli,GLMB)滤波器的滤波参数,在GLMB的基础上提出一种适应视频图像的新生算法进行航迹起始;针对红外图像序列目标检测概率未知的情况,将未知检测概率的基数化概率假设密度(Cardinality Probability Hypothesis Density,CPHD)滤波器集成到GLMB中实时估计目标检测概率以提升跟踪精度。仿真结果表明,所提出算法能有效地排除量测漏检和虚警的干扰,跟踪不同红外复杂场景下的弱小目标。

非线性量测下的机动多目标跟踪

为了解决非线性量测下机动多目标跟踪实时性差、跟踪误差大以及对杂波变化鲁棒性较差的问题,基于随机有限集理论,提出了一种采用量测转换和模糊算法改进的多模型δ-广义标签多伯努利滤波器。首先,推导了交互多模型的δ-GLMB滤波器,通过去相关无偏量测转换实现位置量测从极坐标系到笛卡尔坐标系的无偏转换,并通过预测值去除量测误差和其协方差的相关性造成的滤波估计偏差,实现了非线性场景下的机动多目标跟踪;然后,通过航迹和量测的关联新息以及目标的机动约束构建联合波门,降低了杂波量测的数量;最后引入改进的模糊算法,以目标的模型后验概率为输入,根据模型的分离程度自适应调节运动模型的过程噪声,增加滤波精度。研究结果表明:在杂波环境下,通过与CKF-JMS-δ-GLMB、CKF-IMM-δ-GLMB等非线性多模型滤波器对比,所提算法计算时间较小,且跟踪精度更高,鲁棒性强。所提算法避免了传统的非线性处理方式计算量较大的问题,并且具有较好的杂波抑制特性,提升了非线性量测下机动多目标跟踪的性能。

Effective implementation and improvement of fast labeled multi-Bernoulli filter

Effective implementation of the fast labeled multi-Bernoulli(FLMB)filter is addressed for target tracking with interval measurements.Firstly,a sequential Monte Carlo(SMC)implementation of the FLMB filter,SMC-FLMB filter,is derived based on generalized likelihood function weighting.Then,a box particle(BP)implementation of the FLMB filter,BP-FLMB filter,is developed,with a computational complexity reduction of the SMC-FLMB filter.Finally,an improved version of the BP-FLMB filter,improved BP-FLMB(IBP-FLMB)filter,is proposed,improving its estimation accuracy and real-time performance under the conditions of low detection probability and high clutter.Simulation results show that the BP-FLMB filter has a great improvement of the real-time performance than the SMC-FLMB filter,with similar tracking performance.Compared with the BP-FLMB filter,the IBP-FLMB filter has better estimation performance and real-time performance under the conditions of low detection probability and high clutter.

多模型GGIW-GLMB算法跟踪机动群目标

针对多个机动群目标跟踪问题,提出了一种多模型伽马高斯逆威夏特-广义标签多贝努利(MM-GGIW-GLMB)算法。采用多模型算法对群目标进行运动建模,利用最适高斯(BFG)近似在预测阶段对多模型进行融合,减小了多模型算法的运算量,为进一步提高算法在目标机动阶段的跟踪性能,引入强跟踪滤波器(STF)对BFG算法得到的预测状态协方差进行修正。利用最优次模式分配(OSPA)距离及其一倍标准差和航迹标签正确率衡量算法对机动群目标的跟踪性能。仿真结果表明,本文算法能够提升对机动群目标的跟踪精度和稳定性。

基于GLMB滤波和Gibbs采样的多扩展目标有限混合建模与跟踪算法

本文针对杂波条件下多扩展目标的状态估计,目标个数估计,扩展目标形状估计问题,提出了一种基于标签随机有限集(Labelled random finite sets,L-RFS)框架下多扩展目标跟踪学习算法,该学习算法主要包括两方面:多扩展目标动态建模和多扩展目标的跟踪估计.首先,结合广义标签多伯努利滤波器(Generalized labelled multi-Bernoulli,GLMB)建立了扩展目标的量测有限混合模型(Finite mixture models,FMM),利用Gibbs采样和贝叶斯信息准则(Bayesian information criterion,BIC)准则推导出有限混合模型的参数来对多扩展目标形状进行学习,然后采用等效量测方法来替代扩展目标产生的量测,对扩展目标形状采用椭圆逼近建模,实现扩展目标形状与状态的估计.仿真实验表明本文所给的方法能够有效跟踪多扩展目标,并且在目标个数估计方面优于CBMeMBer算法.此外,与标签多伯努利滤波(LMB)计算比较表明:GLMB和LMB算法滤波估计精度接近,二者精度高于CBMeMBer算法.

Multi-Bernoulli Filter for Tracking Multiple Targets Using Sensor Array

This paper presents a multi-Bernoulli filter for tracking the direction of arrival(DOAs)of time-varying number of targets using sensor array.Our method operates directly on the measurements of sensor array and does not require any detection.Firstly,more information is reserved and compared with the after-detection measurements using a finite set of detected points.It can significantly improve the tracking performance,especially in low signal-to-noise ratio.Secondly,it inherits the advantages of the multi-Bernoulli approximation which models each of the targets individually.This allows more accurate multi-target state estimation,especially when targets cross.The proposed filter does not need clustering step and simulation results showcase the improved performance of the proposed filter.

标签随机有限集下多机动目标跟踪快速实现

针对传统δ-GLMB(δ-Generalized Labeled Multi-Bernoulli)滤波和δ-JGLMB(δ-Joint Generalized Labeled Multi-Bernoulli)滤波计算复杂度较高和实际场景中目标机动的问题,文中提出了IMM-δ-FGLMB(Interactive Multi-modelδ-GLMB)滤波和IMM-δ-FJGLMB(Interactive Multi-modelδ-JGLMB)滤波两种跟踪算法。在滤波新生阶段提出量测驱动Bernoulli成分筛选过程(Measurement Driven for Filtrate the Bernoulli Components,MDFBC),通过利用量测信息来对新生的Bernoulli成分进行粗筛,去除无贡献的新生Bernoulli成分,进而为算法减少计算量。在滤波的预测阶段,通过利用多个模型对机动目标的状态进行交互式预测,从而避免使用单个模型进行状态预测时出现的模型失配问题。实验验证了所提算法在多机动目标跟踪和计算效率方面的优良性能。

Multiple-model GLMB filter based on track-before-detect for tracking multiple maneuvering targets

A generalized labeled multi-Bernoulli(GLMB)filter with motion mode label based on the track-before-detect(TBD)strategy for maneuvering targets in sea clutter with heavy tail,in which the transitions of the mode of target motions are modeled by using jump Markovian system(JMS),is presented in this paper.The close-form solution is derived for sequential Monte Carlo implementation of the GLMB filter based on the TBD model.In update,we derive a tractable GLMB density,which preserves the cardinality distribution and first-order moment of the labeled multi-target distribution of interest as well as minimizes the Kullback-Leibler divergence(KLD),to enable the next recursive cycle.The relevant simulation results prove that the proposed multiple-model GLMB-TBD(MM-GLMB-TBD)algorithm based on K-distributed clutter model can improve the detecting and tracking performance in both estimation error and robustness compared with state-of-the-art algorithms for sea clutter background.Additionally,the simulations show that the proposed MM-GLMB-TBD algorithm can accurately output the multitarget trajectories with considerably less computational complexity compared with the adapted dynamic programming based TBD(DP-TBD)algorithm.Meanwhile,the simulation results also indicate that the proposed MM-GLMB-TBD filter slightly outperforms the JMS particle filter based TBD(JMSMeMBer-TBD)filter in estimation error with the basically same computational cost.Finally,the impact of the mismatches on the clutter model and clutter parameter is investigated for the performance of the MM-GLMB-TBD filter.

δ-广义标签多伯努利滤波算法的非线性扩展

针对高斯混合(Gaussian mixture,GM)实现的变分贝叶斯-δ-广义标签多伯努利(variational Bayesian-δ-generalized labeled multi-Bernoulli,VB-δ-GLMB)滤波算法在非线性场景下跟踪性能较低这一问题,结合基于临近点算法(proximal point algorithm,PPA)和变分贝叶斯(variational Bayesian,VB)的迭代优化与容积卡尔曼滤波(cubature Kalman filtering,CKF),提出一种适用于非线性模型的机动多目标跟踪算法。该算法在GM-VB-δ-GLMB的基础上采用逆伽马(inverse-Gamma,IG)和高斯乘积混合分布近似量测噪声协方差和状态联合后验分布;利用PPA-CKF-VB(PCKF-VB)方法对传递过程中的高斯项参数进行预测更新;最后为提高滤波精度进行变分贝叶斯容积RTS(VB cubature Rauch-Tung-Striebel,VB-CRTS)平滑。仿真结果表明,对于量测噪声未知的非线性系统,所提的算法与现有的VB-δ-GLMB算法相比目标跟踪精度有显著提高。

基于幅度信息的标签多伯努利滤波算法

杂波环境下,现有多目标跟踪滤波器会出现性能衰减。对此,提出了基于幅度信息(amplitude information,AI)的广义标签多伯努利(generalized labeled multi-Bernoulli,GLMB)滤波算法(AI-GLMB)。通常杂波幅度低于目标回波幅度,通过引入幅度信息对目标状态进行扩展,建立幅度似然函数,推导新的更新方程,并给出了算法的序贯蒙特卡罗实现方法。仿真结果表明,AI-GLMB算法能有效适应高杂波环境,同幅度信息辅助的概率假设密度滤波算法、幅度信息势平衡多伯努利滤波算法及传统GLMB滤波算法相比,其跟踪精度更高。

基于广义标签多伯努利滤波的可分辨群目标跟踪算法

针对杂波条件下可分辨群目标的状态估计、目标个数与子群个数估计问题,提出了一种基于标签随机有限集(Label random finite set,L-RFS)框架下的可分辨群目标跟踪算法,该算法主要包括两个方面:可分辨多群目标动态建模和多群目标的跟踪估计.本文工作主要包括:1)结合图论中的邻接矩阵对可分辨群目标运动进行动态建模.2)利用基于L-RFS的广义标签多伯努利滤波(Generalizes label multi-Bernoulli,GLMB)算法对目标的状态和个数进行估计,并且通过估计邻接矩阵得到群的结构和个数估计.3)通过个数不同、结构不同的三个子群目标在二维平面分别做线性和非线性运动进行算法验证.仿真分析表明本文算法能够准确估计出群目标中各目标的状态、个数以及子群的个数,并且能获得目标的航迹估计.

多模型广义标签多伯努利滤波器

标准广义标签多伯努利算法没有对目标状态转移密度进行深入分析,在带入确定运动模型的情况下无法对机动目标进行跟踪。针对这个问题,参考基于马尔可夫跳变分支合并策略的多模型算法,提出了交互多模型广义标签多伯努利算法、一阶广义伪贝叶斯广义标签多伯努利算法,以及二阶广义伪贝叶斯广义标签多伯努利算法,并将这三种多模型算法与同样针对机动多目标的马尔可夫跳变系统广义标签多伯努利算法进行比较。仿真结果表明,与马尔可夫跳变系统广义标签多伯努利算法相比,所提三种算法具有更低的计算时间消耗和更高的跟踪精度。其中,一阶广义伪贝叶斯广义标签多伯努利算法计算时间消耗最低,二阶广义伪贝叶斯广义标签多伯努利算法跟踪精度最高,交互多模型广义标签多伯努利算法综合性能最好。

基于交互多模型的分组δ-广义标签多伯努利算法

为了解决马尔科夫跳变系统广义标签多伯努利滤波器在多机动目标跟踪场景需要计算大量模式假设分支,并且需要频繁对假设分支进行剪枝,导致算法存在计算量过高并且影响跟踪精度的问题,提出一种基于交互多模型的分组δ-广义标签多伯努利滤波器。滤波器采用航迹分组策略,不同组的航迹独立进行关联映射与分支权重计算,降低了关联的计算复杂度,可以实现不同航迹组之间并行滤波。另外,为了处理机动目标场景引入交互多模型,给出基于交互多模型的分组滤波递推方程。仿真结果表明,所提出的滤波器跟踪精度更高,计算速度更快,可以用于跟踪多个机动目标的场景。

基于差集表遍历搜索的互素阵列DOA估计器

为降低波达方向(direction-of-arrival,DOA)估计中阵元硬件成本和阵元间耦合,提出基于差集表遍历搜索的互素稀疏阵列DOA估计器。该估计器兼顾低复杂度和高精度的特点。其低复杂度在于:仅需依据互素稀疏阵列的阵元坐标即可构造出差集表,以该差集表为指导,可实现观测阵元的协方差矩阵到Nyquist虚拟阵列协方差矩阵的快速转换,进而借助多信号分类分解实现多目标DOA估计;其高精度在于:差集表遍历搜索措施可提升信号子空间的维度,进而提高了空间谱分辨率。鉴于高频段、低波长的阵列信号处理应用日益展开,该DOA估计器具有较为广阔的应用前景。

A robust Poisson multi-Bernoulli filter for multi-target tracking based on arithmetic average fusion

The coalescence and missed detection are two key challenges in Multi-Target Tracking(MTT).To balance the tracking accuracy and real-time performance,the existing Random Finite Set(RFS)based filters are generally difficult to handle the above problems simultaneously,such as the Track-Oriented marginal Multi-Bernoulli/Poisson(TOMB/P)and Measurement-Oriented marginal Multi-Bernoulli/Poisson(MOMB/P)filters.Based on the Arithmetic Average(AA)fusion rule,this paper proposes a novel fusion framework for the Poisson Multi-Bernoulli(PMB)filter,which integrates both the advantages of the TOMB/P filter in dealing with missed detection and the advantages of the MOMB/P filter in dealing with coalescence.In order to fuse the different PMB distributions,the Bernoulli components in different Multi-Bernoulli(MB)distributions are associated with each other by Kullback-Leibler Divergence(KLD)minimization.Moreover,an adaptive AA fusion rule is designed on the basis of the exponential fusion weights,which utilizes the TOMB/P and MOMB/P updates to solve these difficulties in MTT.Finally,by comparing with the TOMB/P and MOMB/P filters,the performance of the proposed filter in terms of accuracy and efficiency is demonstrated in three challenging scenarios.

Vo 滤波器算法在多目标跟踪中应用研究

标签随机有限集（LRFS）对多目标航迹问题提供了一个新的研究方向，然而国内关于其研究还很少，关于其相关的算法的总结文献基本没有。本文基于标签随机有限集，对其在多目标系统中应用的各种算法原理和其之间关系进行了综述。主要论述了LRFS框架下的一般的标签伯努利滤波器(GLMB)、脉冲扩展标签多伯努利(-GLMB)算法等相关算法，着重梳理了它们的算法之间的关系，并进行了比较和总结。

Variational Bayesian labeled multi-Bernoulli filter with unknown sensor noise statistics

It is difficult to build accurate model for measurement noise covariance in complex backgrounds. For the scenarios of unknown sensor noise variances, an adaptive multi-target tracking algorithm based on labeled random finite set and variational Bayesian （VB） approximation is proposed. The variational approximation technique is introduced to the labeled multi-Bernoulli （LMB） filter to jointly estimate the states of targets and sensor noise variances. Simulation results show that the proposed method can give unbiased estimation of cardinality and has better performance than the VB probability hypothesis density （VB-PHD） filter and the VB cardinality balanced multi-target multi-Bernoulli （VB-CBMeMBer） filter in harsh situations. The simulations also confirm the robustness of the proposed method against the time-varying noise variances. The computational complexity of proposed method is higher than the VB-PHD and VB-CBMeMBer in extreme cases, while the mean execution times of the three methods are close when targets are well separated.

An efficient measurement-driven sequential Monte Carlo multi-Bernoulli filter for multi-target filtering

We propose an efficient measurement-driven sequential Monte Carlo multi-Bernoulli(SMC-MB) filter for multi-target filtering in the presence of clutter and missing detection. The survival and birth measurements are distinguished from the original measurements using the gating technique. Then the survival measurements are used to update both survival and birth targets, and the birth measurements are used to update only the birth targets.Since most clutter measurements do not participate in the update step, the computing time is reduced significantly.Simulation results demonstrate that the proposed approach improves the real-time performance without degradation of filtering performance.

Generalized labeled multi-Bernoulli filter with signal features of unknown emitters

A novel algorithm that combines the generalized labeled multi-Bernoulli(GLMB) filter with signal features of the unknown emitter is proposed in this paper. In complex electromagnetic environments, emitter features(EFs) are often unknown and time-varying. Aiming at the unknown feature problem, we propose a method for identifying EFs based on dynamic clustering of data fields. Because EFs are time-varying and the probability distribution is unknown, an improved fuzzy C-means algorithm is proposed to calculate the correlation coefficients between the target and measurements, to approximate the EF likelihood function. On this basis, the EF likelihood function is integrated into the recursive GLMB filter process to obtain the new prediction and update equations.Simulation results show that the proposed method can improve the tracking performance of multiple targets,especially in heavy clutter environments.