Improving extent estimation of extended targets with Track-before-Detect

Extended target tracking is well-studied but only a few have considered Track-before-Detect (TbD). In this paper, we address short-comings of the current TbD model solution, to improve extent estimation. Experimenting with a particle filter, we demonstrate that the current solution is inaccurate. We propose a solution that improves extent estimation. Using intermediary solutions, we demonstrate the impact of model mismatches. Our solution matches a reasonable measurement model exactly. Hence, we verify the filter and are confident in our results. In typical situations, model mismatches lead to incorrect results, as we show for the current and intermediary solutions that give poorer results as targets move away from the sensor.