Generalized probabilistic data association for vehicle tracking under clutter

Vehicle tracking under clutter is an important prerequisite for numerous vehicular applications. In this paper, we propose a generalization of the existing integrated probabilistic data association method in order to model situations where several true and additional clutter observations originated from one object. We will show that the proposed method outperforms the existing one. Furthermore, we will demonstrate a system utilizing a camera sensor and the proposed algorithm for detecting and tracking vehicles under clutter.