Target Clustering Based Multi-Bernoulli Filter for Superpositional Sensors

TL;DR

This paper introduces a novel multi-Bernoulli filtering approach for superpositional sensors, utilizing target clustering and Gaussian implementation to improve multi-target tracking in complex scenarios.

Contribution

It proposes a target clustering based multi-Bernoulli filter tailored for superpositional sensors, with a conjugate density formulation and sigma point Gaussian implementation.

Findings

Effective in handling multiple target interactions

Performs well with overlapping targets

Demonstrates robustness in complex scenarios

Abstract

The sensor whose output is a function of the sum of contributions from targets present in the surveillance area is called superpositional sensor. In this letter, target clustering based multi-Bernoulli filter for superpositional sensors is proposed.Targets are clustered according to the set of resolution cells illuminated by them. Single target posterior density is strictly derived, and densities of all the targets are combined to a approximate multi-target posterior, which makes the multiBernoulli density is conjugate with respect to the likelihood of superpositional sensors. The Gaussian implementation of the proposed algorithm is also presented, where the multidimensionality and the nonlinearity of update equation are handled by sigma point transformation. The simulation results illustrate that the proposed algorithm is effective confronted with the interaction of multiple targets and long term overlapping of two targets.