Distributed tracking of extended targets using random matrices

This paper studies the problem of extended target tracking using multiple sensors in a distributed manner. The approach of using a random matrix for the target extension is used. The weighted Kullback-Leibler divergence is applied to fuse sensor-based estimates via exchanging information among neighboring sensors. It is shown that closed form solutions can be derived to fuse sensor-based estimates of the target state and the extension matrix, respectively. In particular, the resulting solutions coincide with the same recursive form for the posterior density functions of the target state and the extension matrix. A numerical example is provided to verify the effectiveness of the proposed filter.