Regional Variance for Multi-Object Filtering

Delande, Emmanuel; Üney, Murat; Houssineau, Jérémie; Clark, Dan E.

doi:10.1109/tsp.2014.2328326

Cited by 52 publications

(69 citation statements)

References 27 publications

(95 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recent developments within the estimation framework for stochastic populations [19], such as target classification [37] or information-theoretic policies for closed-loop sensor management [38], are fully relevant to the context of SSA and directly applicable to the DISP filter. The regional statistics for multiobject filters [39], estimating the number of objects with associated uncertainty in any desired region of the state space, could be exploited in the context of a SSA scenario to provide, for example, a dynamical assessment of the density of objects on particular orbits. Originally developed within the FISST framework, the regional statistics could be adapted to any filter derived from the estimation framework for stochastic populations [19] as well, including the DISP filter.…”

Section: Further Workmentioning

confidence: 99%

Novel Multi-Object Filtering Approach for Space Situational Awareness

Delande

Frueh

Franco

et al. 2018

Journal of Guidance, Control, and Dynamics

Self Cite

View full text Add to dashboard Cite

Surveillance activities with ground-based assets in the context of space situational awareness are particularly challenging. The observation process is indeed hindered by short observation arcs, limited observability, missed detections, measurement noise, and contamination by clutter. This paper exploits a recent estimation framework for stochastic populations for space situational awareness surveillance scenarios. This framework shares the flexibility of the finite set statistics framework in the modeling of a dynamic population of objects and the representation of all the sources of uncertainty in a single coherent probabilistic framework and the intuitive approach of traditional trackbased techniques to describe individual objects and maintain track continuity. We present a recent multi-object filtering solution derived from this framework, the filter for distinguishable and independent stochastic populations, and propose a bespoke implementation of the multitarget tracking algorithm for a space situational awareness surveillance activity. The distinguishable and independent stochastic populations filter is tested on a surveillance scenario involving two ground-based Doppler radars in a challenging environment with significant measurement noise, limited observability, missed detections, false alarms, and no a priori knowledge about the number and the initial states of the objects in the scene. The tracking algorithm shows good performance in initiating tracks from object-generated observations and in maintaining track custody throughout the scenario, even when the objects are outside of the sensors' fields of view, despite the challenging conditions of the surveillance scenario.Nomenclature α y = probability of existence of previously detected target with observation path y cH = probability of existence of hypothesis H H = hypothesis, i.e., subset of compatible tracks in Y (H, n) = multitarget configuration, describing a possible composition of population X H tjt−1 ; H t = set of all hypotheses, propagated by the filter for distinguishable and independent stochastic populations (DISP) lz; x = likelihood of association between target with state x and observation with state z P tjt−1 , P t = law of whole population of targets, propagated by the distinguishable and independent stochastic populations filter P a = law of population of appearing targets p d x = probability of detection of target with state x p fa z = probability that observation with state z is false alarm p y = probability distribution of previously detected target with obs. path y p ϕ = probability distribution of each yet-to-be-detected target wH; n = joint probability of existence of targets in configuration (H, n) X = population of targets X = target state space X = target state space augmented with empty state ψ x = target state (e.g., position and velocity coordinates) Y tjt−1 , Y t = set of all observation paths, propagated by distinguishable and independent stochastic populations filter y = observation path or track characterized by said o...

show abstract

Section: Further Workmentioning

confidence: 99%

Novel Multi-Object Filtering Approach for Space Situational Awareness

Delande

Frueh

Franco

et al. 2018

Journal of Guidance, Control, and Dynamics

Self Cite

View full text Add to dashboard Cite

show abstract

“…The PHD filter is one step of the maximum a posterior (MAP) estimation for PET and the update of PHD filter is shown to be the first step of the Shepp-Vardi algorithm which is very famous for image processing (Snyder and Miller 1991;Streit 2009Streit , 2010. Delande et al (2014) have investigated the high-order statistics in target number of the PHD filter. Assume that a realisation of a point process Φ is a set of points {x 1 , .…”

Section: The Probability Hypothesis Density (Phd) Filtermentioning

confidence: 99%

Micro-object motion tracking based on the probability hypothesis density particle tracker

et al. 2015

View full text Add to dashboard Cite

Tracking micro-objects in the noisy microscopy image sequences is important for the analysis of dynamic processes in biological objects. In this paper, an automated tracking framework is proposed to extract the trajectories of micro-objects. This framework uses a probability hypothesis density particle filtering (PF-PHD) tracker to implement a recursive state estimation and trajectories association. In order to increase the efficiency of this approach, an elliptical target model is presented to describe the micro-objects using shape parameters instead of point-like targets which may cause inaccurate tracking. A novel likelihood function, not only covering the spatiotemporal distance but also dealing with geometric shape function based on the Mahalanobis norm, is proposed to improve the accuracy of particle weight in the update process of the PF-PHD tracker. Using this framework, a larger number of tracks are obtained. The experiments are performed on simulated data of microtubule movements and real mouse stem cells. We compare the PF-PHD tracker with the nearest neighbor method and the multiple hypothesis tracking method. Our PF-PHD tracker can simultaneously track hundreds of micro-objects in the microscopy image sequence.

show abstract

“…The first phase focussed mainly on the exploitation of the novel high-order regional statistics for multi-object filters, estimating the size of the target population, with associated uncertainty and in any desired region of the surveillance scene. Available for any multi-object filtering solution derived from the FISST framework [20], it is a valuable tool for a comparison between popular multiple-target tracking filters [10] or, as illustrated in the first phase, for an assessment of their extensions adapted to multi-sensor scenarios. Since the availability of the PHD filter [18] and the growing popularity of the multi-object filtering solutions, the axes of development in multi-object filtering have focussed almost exclusively on solutions estimating the target population as a whole without maintaining individual information on specific targets (i.e.…”

Section: Study Overview: Motivation and Objectivesmentioning

confidence: 99%

“…Broadly speaking, the algorithm exploited in this study propagates as much information as possible under the proviso of the modelling assumptions above, and aims at filling the role of a reference filter for the future assessment of multi-object filters, in particular through the exploitation of the higher-order regional statistics [10].…”

Section: Filter Derivation: Key Modelling Assumptionsmentioning

confidence: 99%

“…They led to the construction and the practical extraction of higher-order statistics describing the output of multi-object filters [10]. Notably, the novel regional variance in target number can be exploited to quantify the uncertainty in the estimation of the target number, evaluated in any arbitrary region of the surveillance scene, provided by a given multi-object filter.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation