Layered multiple-model algorithm with application to tracking maneuvering and bending extended target in clutter

In many practical situations has a layered structure. This paper presents a novel and highly efficient MM algorithm, called Layered MM (LMM) algorithm, which takes full advantage of the layered structure. Compared with the standard MM algorithm, it is expected that the LMM algorithm will have a substantial saving in computation and similar performance. The need for such a layered structure and layered algorithm is motivated by the problem of tracking maneuvering and bending extended targets in a cluttered environment. The basic idea of the LMM algorithm is to incorporate the coupling between the layers using time-varying transition probabilities of the models. In conjunction with a fast point-pattern matching algorithm and a modified conditional matching pairs support finding algorithm proposed recently, the LMM algorithm is applied to solve the problem of tracking maneuvering and bending extended targets in a cluttered environment.