Tracking Algorithm Research Based On Particle Filter

Many filtering algorithms can get perfect tracking results within the linear and Gaussian model;But when the objects are in the background of multi-model, non-Gaussian,strong noise,the estimation accuracy of classical algorithms such as kalman filter and extended kalman filter decreases significantly and even appears divergentce phenomenon.Particle filter is an recursive filtering method based on Monte Carlo simulation technology and Bayesian estimation.As a nonlinear filter algorithm based on Bayesian estimation,particle filter has original advantage at treating the parameter estimation and state filtering aspects of nonlinear non-Gaussian time-varying systems.Thus a great development is obtained.Aiming at the defects of existing methods in target tracking, combining with the practical issues,the improved algorithms are presents in this paper to achieve better results.The main work is summarized as follows:1.Considering the importance of density function to improving the degradation phenomena and the accuracy of particle filter,the thesis discusses a TSEPF algorithm. The algorithm makes the generated samples closer to the real sampling through integrating of the latest observation information.Simulation results show that the performance of the algorithm is superior to EPF,UPF and several other filtering algorithms.2.In order to resolve the depletion phenomenon of particles after resampling,the thesis presents a improved algorithm through introducing a MCMC(Markov chain Monte Carlo) step to increase the diversity of particles.The algorithm is applied to target tracking,the simulation results also show that the accuracy of PF-MCMC algorithm is higher than the traditional PF algorithm and it can track the target more exactly.3.When the targets show strong motor capacity,the thesis presents a IMM-PF algorithm that combines interacting multiple model algorithm(IMM) and particle filter algorithm for the problems of single-model adaptive filter.The simulation results show that IMM-PF algorithm is practicable and effective under the non-linear maneuvering target tracking conditions,the performance of IMM-PF algorithm is better than the IMM algorithm which uses the extended Kalman filter in tracking.