A modified robust multi-target tracking method with high clutter and underwater nonstationary measurement noisea)

Area of Science:

• Underwater acoustics
• Signal and information processing
• Multi-target tracking

Background:

• Robust tracking of multiple underwater targets is crucial but challenging due to clutter and nonstationary noise.
• Existing random finite set (RFS) algorithms struggle with observation likelihood contamination and poor accuracy in such environments.

Purpose of the Study:

• To propose a robust two-stage modified variational Bayesian delta-generalized labeled multi-Bernoulli multi-target tracking algorithm.
• To enhance tracking accuracy and robustness in cluttered underwater environments with nonstationary measurement noise.

Main Methods:

• Introduced Sage-Husa (SH) estimation for initial measurement noise covariance correction.
• Employed a variational Bayesian framework for further optimization of measurement noise covariance estimates.
• Utilized a delta-generalized labeled multi-Bernoulli filtering update stage.

Main Results:

• Effectively alleviated clutter contamination of the likelihood function in adaptive RFS.
• Improved discrimination between targets and clutter under complex noise conditions.
• Significantly enhanced the robustness of underwater multi-target active tracking through real-time noise correction.

Conclusions:

• The proposed algorithm significantly outperforms traditional and existing adaptive generalized labeled multi-Bernoulli methods.
• Demonstrated superior performance in simulations and experiments with strong clutter and nonstationary noise.
• Offers a robust solution for underwater multi-target active tracking.