Adaptive Rao-Blackwellized Particle Filter and Its Evaluation for Tracking in Surveillance
The Kalman filtering as a recursive linear estimator is inadequate when applying to non-Gaussian densities and multimodal. Particle Filters sampling in high-dimensional spaces can be inefficient. The use of Rao-Blackwellized particle filters can drastically reduce the size of the space over which we need to sample. In this paper, we proposed an efficient method for using subspace representations by applying adaptive Rao-Blackwellized particle filter to integrate out the subspace coefficients in the state vector. When an object moves in three-dimensional space, our system tracks the object using Adaptive Rao-Blackwellized particle filter and fuzzy control system. Notwithstanding the use of stochastic methods and agile motion, the algorithm runs in near real-time. The fuzzy controller was designed to track moving objects with zero tracking errors.