CBMeMBer Filter for Extended Object Tracking Using Box Particle

A CBMeMBer filter for extended object detection and tracking using box particle is proposed for the problem that measurement affects by inaccuracy and vague in extended target tracking. Firstly, states and observations of extended target are modeled as a Bernoulli random finite set and a Poisson random finite set respectively. Then the Bernoulli filter is implemented based on interval analysis, the pseudo likelihood function and state updating function of extended target appropriate to interval measurements are derived. Finally, the state of extended targets is recursively estimated using Bernoulli filter box particles in non-linear models. Simulation results show that the algorithms proposed by our study could lead to significantly lower computational complexity with keeping tracking performance.