A Framework for Multisensor Multiple Extended Target Tracking with Vessel Size Categorisation

Abstract

Traffic monitoring and surveillance form part of the basic aspects of several maritime applications. The observation regions around harbours have multiple modern radar sensors for adequate coverage, and the sensors produce measurements from potential targets that appear as point clouds. With measurements from multiple sensors, extra information about the target’s shape can be gained in order to improve monitoring within the particular observation region. In this paper, we present a framework for processing radar measurements from multiple sensors which outputs the extended state estimates of the vessels over an observation region jointly with their approximate size range categories. We demonstrate the performance of the framework based on real-world radar video streams with respect to its real-time capability, and finally present our implementation on both, single and multiple targets, showing promising results accompanied by some discussion.