A sequential Monte Carlo method for particle filters

An object oriented particle filter framework is proposed based on sequential Monte Carlo methods. Particle filter is an extensively used algorithm for vision based tracking systems. However, little work has been done in the past literature to investigate the implementation strategies of the particle filter algorithm. In this paper, we propose a framework based on open source particle filter libraries and evaluate respective advantages and disadvantages. The results support the proposed object oriented particle filter being a most useful tool for computer vision based stochastic prediction.