Joint tracking and classification for extended targets in maritime surveillance

Abstract

In the framework of target tracking, joint tracking and classification (JTC) based approaches have been recently introduced to adjust the trajectory estimation to a given type of target. In a maritime surveillance framework, these approaches leverage marine radar capabilities to provide high resolution measurements. It can then be helpful to choose relevant motion models regarding the tracking algorithm. Among them, the authors recently proposed to take advantage of the target extent measurements for single or multiple-target tracking. This work was based on a point source hypothesis regarding the position measurement of each target. However, this is no longer suitable for extended target tracking. An extension of our previous work to track properly extended targets whose position measurement can be obtained from any point source of its surface is thus proposed. For this purpose, random matrix theory is leveraged. Eventually, the relevance of the proposed approach with simulated data is analyzed.