해상용 레이더를 이용한 다중 선박 추적 알고리듬 구현

Abstract

This work presents the implementation of a multi-vessel tracking algorithm using a marine radar system. The multi-vessel tracking algorithm requires additional procedures such as gating and association in addition to the conventional procedures of single target tracking algorithm which usually includes only the prediction and filtering. In this work, the ellipsoidal gate based on the normalized statistical distance was used for cutting out unlikely measurement-to-track pairings and the global nearest neighbor (GNN) method was employed for measurement-to-track association. For optimal assignment of measurement to track, the auction algorithm was embodied in the association procedure. To predict and filter the dynamic states of tracks, the Kalman filter defined in the two-dimensional Cartesian coordinate system was used and thus the radar measurements in the Polar coordinate system were converted to the Cartesian coordinate to be fed to the track filter. To validate the implemented algorithm, simulation tests were fulfilled and the results showed that the implemented algorithm operated properly for the suggested simulation cases.