GLMB Tracker with Partial Smoothing

Nguyen, Tran Thien Dat; Kim, Du Yong

doi:10.3390/s19204419

Cited by 9 publications

(8 citation statements)

References 55 publications

(102 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the average sense, the response of the δ-GLMB smoother to target birth is faster than that of the CPHD smoother, whereas its response to target death is slightly quicker than that of the CPHD smoother. The prompt response of the δ-GLMB smoother attributes mainly to the fast response of the δ-GLMB filter [34] to target birth or death because the smoothing density relies on both the filtering density and the measurements indicating the true target number. The PHD and MB smoothers suffer from the premature target death [23], [25] at 40s, 45s, which are also seen by the large OSPA cardinality errors at those moments in Fig.…”

Section: Numerical Resultsmentioning

confidence: 99%

“…where the subtraction sign in B + − B + denotes the set difference operation, and r ( ) B denotes the birth probability of label ∈ B + . Replacing ω (33) by (34) leads to ω (I ,ξ,B + ,S…”

Section: B Ranked Assignment Formulationmentioning

confidence: 99%

“…A computationally efficient LMB smoother was proposed in [33], and they show that it improves the tracking performance over the PHD and MB smoothers, and the LMB filter. A GLMB tracker with partial smoothing was detailed in [34], in which they update the trajectory tuples using the output of the GLMB filter at each time step, and then perform single-object forward-backward smoothing. An approximate δ-GLMB (δ-GLMB-A) smoother was proposed in [35], in which the derivation assumes that no targets births and deaths occur in the smoothing period.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Multitarget Tracking Using One Time Step Lagged Delta-Generalized Labeled Multi-Bernoulli Smoothing

Liang

et al. 2020

IEEE Access

View full text Add to dashboard Cite

Aiming at improving the tracking performance of the delta-generalized labeled multi-Bernoulli (δ-GLMB) filter, we present a one time step lagged δ-GLMB smoother in this work, which also inherently outputs targets trajectories and differs from the Probability hypothesis density (PHD), Multi-Bernoulli (MB), and Cardinalized probability hypothesis density (CPHD) smoothers that are incapable of generating target trajectories directly. Under the standard multitarget measurement likelihood and state transition kernel, we show that a δ-GLMB distributed multitarget filtering density would result in a same distributed one time step lagged multitarget smoothing density. An efficient implementation of the proposed smoothing algorithm using the standard ranked assignment technique is also given. Numerical results show that the proposed smoother is capable of tracking a time-varying number of targets, in the presence of measurement origin uncertainty, target detection uncertainty, and clutter, and show that the proposed smoother outperforms the δ-GLMB filter, and the PHD, MB, and CPHD smoothers of the same time lag on both the estimates of target number and state and it also outperforms the LMB and the approximated δ-GLMB smoothers of the same time lag on target number estimate. INDEX TERMS Delta-generalized labeled multi-Bernoulli, multitarget tracking, random finite set, smoothing.

show abstract

Section: Numerical Resultsmentioning

confidence: 99%

Section: B Ranked Assignment Formulationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Multitarget Tracking Using One Time Step Lagged Delta-Generalized Labeled Multi-Bernoulli Smoothing

Liang

et al. 2020

IEEE Access

View full text Add to dashboard Cite

show abstract

“…the mode or mean. Alternatively, the entire trajectory of object`2 I ⇤ can be estimated using the forward-backward algorithm, starting from its current filtering density p (⇠ ⇤ ) (•,`) and propagating backward to its time of birth [20], [54].…”

Section: Multi-sensor Glmb Filtermentioning

confidence: 99%

A Bayesian Filter for Multi-View 3D Multi-Object Tracking With Occlusion Handling

Ong

et al. 2022

IEEE Trans. Pattern Anal. Mach. Intell.

View full text Add to dashboard Cite

This paper proposes an online multi-camera multi-object tracker that only requires monocular detector training, independent of the multi-camera configurations, allowing seamless extension/deletion of cameras without retraining effort. The proposed algorithm has a linear complexity in the total number of detections across the cameras, and hence scales gracefully with the number of cameras. It operates in the 3D world frame, and provides 3D trajectory estimates of the objects. The key innovation is a high fidelity yet tractable 3D occlusion model, amenable to optimal Bayesian multi-view multi-object filtering, which seamlessly integrates, into a single Bayesian recursion, the sub-tasks of track management, state estimation, clutter rejection, and occlusion/misdetection handling. The proposed algorithm is evaluated on the latest WILDTRACKS dataset, and demonstrated to work in very crowded scenes on a new dataset.

show abstract

“…This filter can estimate not only the number of the targets but also their trajectories, simultaneously [12]. It has been applied to several problems as tracking with merged measurements [13], track-before-detect [14,15], extended targets [16], cell biology [17,18], sensor scheduling [19], spawning targets [20], distributed data fusion [21], field robotics [22,23] and computer vision [24]. The GLMB filter for multitarget tracking with two sensors has been developed in [25,26].…”

Section: Introductionmentioning

confidence: 99%

Tracking Multiple Marine Ships via Multiple Sensors with Unknown Backgrounds

Nguyen

Liu

2019

Sensors

Self Cite

View full text Add to dashboard Cite

In multitarget tracking, knowledge of the backgrounds plays a crucial role in the accuracy of the tracker. Clutter and detection probability are the two essential background parameters which are usually assumed to be known constants although they are, in fact, unknown and time varying. Incorrect values of these parameters lead to a degraded or biased performance of the tracking algorithms. This paper proposes a method for online tracking multiple targets using multiple sensors which jointly adapts to the unknown clutter rate and the probability of detection. An effective filter is developed from parallel estimation of these parameters and then feeding them into the state-of-the-art generalized labeled multi-Bernoulli filter. Provided that the fluctuation of these unknown backgrounds is slowly-varying in comparison to the rate of measurement-update data, the validity of the proposed method is demonstrated via numerical study using multistatic Doppler data.

show abstract

GLMB Tracker with Partial Smoothing

Cited by 9 publications

References 55 publications

Multitarget Tracking Using One Time Step Lagged Delta-Generalized Labeled Multi-Bernoulli Smoothing

Multitarget Tracking Using One Time Step Lagged Delta-Generalized Labeled Multi-Bernoulli Smoothing

A Bayesian Filter for Multi-View 3D Multi-Object Tracking With Occlusion Handling

Tracking Multiple Marine Ships via Multiple Sensors with Unknown Backgrounds

Contact Info

Product

Resources

About