Nonlinear filtering algorithms in object tracking applications

Based on the Bayesian theory framework, the extended Kalman filter and particle filter are analyzed in object tracking in real-time and dynamic systems. When EKF is applied to the object tracking, the error of estimation must be considered because of the defects of EKF in nonlinear system. Aim at these defects, this paper selects particle filter and regularized particle filter algorithms to overcome these disadvantages and improve capability such as the practicality and accuracy. The experimental results show the relationship between the tracking accuracy and the different algorithms. After the experiment, a conclusion is made between the number of particles and time-consuming.