A new type of random finite set for multi-target tracking

Random finite set (RFS) based approaches in multi-target filtering are well-established to estimate the states and its number. However, RFS-based approaches have not given the labels of the states over the time which help to produce the trajectories of targets in multi-target tracking. This paper discusses the limitation of the label RFS (LRFS) approach introduced in [1] where an LRFS is an RFS whose elements are added labels. Then a new type of RFS called Indexed RFS (IRFS) approach is proposed where an IRFS is also an RFS which adds indices to its elements. Unlike the LRFS approach, the density of an IRFS is defined to be proportional to the density of the deindexed version. Therefore, the important properties are preserved under this definition. It is shown in this paper that the IRFS approach offers simpler interpretations compared to the LRFS approach.