A Marine Radar Dataset for Multiple Extended Target Tracking
Jaya Shradha Fowdur, Marcus Baum, Frank J. Heymann · elib (German Aerospace Center) · 2019
The marine radar remains one of the most extensively used sensor for maritime surveillance. Owing to improved technologies, it can nowadays be exploited to gain information about the extents of targets, since multiple measurements can be obtained from a single target. This paper introduces an open multitarget marine radarbased dataset subjected to a linear-time joint probabilistic data association (JPDA) filter for tracking extended targets using ellipsoidal approximations. A tailored version of the Multiplicative Error Model-Extended Kalman Filter* (MEM-EKF*) algorithm is used for estimating the orientations and kinematic properties of multiple targets recursively. Using the automatic identification system (AIS) information as ground truth, the positional errors are evaluated using the optimal sub pattern assignment (OSPA) metric and the performance of the algorithm for orientation estimation is rationally discussed with respect to the ground truth and dataset scenario. The dataset proposed is intended for comparing different algorithms for the purpose of multiple extended target tracking (METT).