Multitarget tracking using the Joint Multitrack Probability Density

This paper addresses the problem of detecting and tracking multiple targets in a Bayesian framework. First, we introduce the definition of joint multitrack probability density (JMKPD) which is the probability of having a certain number of tracks, each one clearly identified with an ID number, and a kinematic state. We develop the a priori model needed to solve the Bayesian problem and a particle filter implementation with two layers, one that deals with the false alarms and track initiation, and another that deals with track maintenance and track end.