Tracking extended targets in high clutter using a GGIW-LMB filter

摘要

Due to increasing sensor resolutions, the commonly used point-target assumption in multi-object tracking algorithms is violated. Recently, several algorithm based on Gaussian inverse Wishart (GIW) or Gamma GIW (GGIW) distributions have been proposed which facilitate the tracking of so-called extended targets which generate multiple measurements per scan. Using GGIW distributions, the target extent and the measurement rate are estimated in addition to the targets' kinematics. In this contribution, the GGIW Labeled Multi-Bernoulli (GGIW-LMB) filter is applied to a scenario with a huge amount of clutter measurements. Additionally, three different target birth models are compared for different clutter rates: static birth distributions, adaptive birth distributions, and adaptive two-step distributions.