Joint Detection, Tracking, and Classification of Multiple Extended Objects Based on the JDTC-PMBM-GGIW Filter
Open Access
- 5 February 2023
- journal article
- research article
- Published by MDPI AG in Remote Sensing
- Vol. 15 (4) , 887
Abstract
This paper focuses on the problem of joint detection, tracking, and classification (JDTC) for multiple extended objects (EOs) within a Poisson multi-Bernoulli (MB) mixture (PMBM) filter, where an EO is described as an ellipse, and the ellipse is modeled by a random matrix. The EOs are classified according to the size information of the ellipse. Usually, detection, tracking, and classification are processed step-by-step. However, step-by-step processing ignores the coupling relationship between detection, tracking, and classification, resulting in information loss. In fact, detection, tracking, and classification affect each other, and JDTC is expected to be beneficial for achieving better overall performance. In the multi-target tracking problem based on RFS, the overall performance of the PMBM filter satisfying the conjugate priors has been verified to be superior to other filters. Specifically, the PMBM filter propagates multiple MB simultaneously during iterative updates and model the distribution of hitherto undetected EOs. At present, the PMBM filter is only applied to multiple extended objects tracking problem. Therefore, we consider using the PMBM filter to solve the JDTC problem of multiple EOs and further improve JDTC performance. Furthermore, the closed-form implementation based on the product of a gamma Gaussian inverse Wishart (GGIW) and class probability mass function (PMF) is proposed. The details of parameters calculation in the implementation process and the derivation of class PMF are presented in this paper. Simulation experiments verify that the proposed algorithm, named the JDTC-PMBM-GGIW filter, performs well in comparison to the existing JDTC strategies for multiple extended objects.Keywords
Funding Information
- National Natural Science Foundation of China (61971103, 62101112, 6142105190307)
- Science and Technology on Electronic Information Control Laboratory (61971103, 62101112, 6142105190307)
This publication has 42 references indexed in Scilit:
- Target Threat Assessment Based on Improved RBF Neural Network2015
- Joint detection, tracking and classification of a manoeuvring target in the finite set statistics frameworkIET Signal Processing, 2015
- An Efficient, Variational Approximation of the Best Fitting Multi-Bernoulli FilterIEEE Transactions on Signal Processing, 2014
- Joint tracking and classification based on aerodynamic model and radar cross sectionPattern Recognition, 2014
- Extended Object Tracking with Random Hypersurface ModelsIEEE Transactions on Aerospace and Electronic Systems, 2014
- Random Set Theory for Multisource-Multitarget Information Fusion2008
- Survey of maneuvering target tracking. part v: multiple-model methodsIEEE Transactions on Aerospace and Electronic Systems, 2005
- Detection, classification, and tracking of targetsIEEE Signal Processing Magazine, 2002
- Tracking low elevation targets in the presence of multipath propagationIEEE Transactions on Aerospace and Electronic Systems, 1994
- Letter to the Editor—An Algorithm for Ranking all the Assignments in Order of Increasing CostOperations Research, 1968