The Application Of Improved Method Research On Particle Filter In Target Tracking Technology

With the development of modern science and engineering technology, the two theory systems in target tracking filter domain are formed, which are classic Wiener filter and modern Kalman filter.In this paper, main techniques of multiple targets tracking in integrated bridge system are discussed. Those techniques are studied in detail, including data association and filter algorithm.Firstly we introduced basic principles of data fusion and targets tracking.This paper introduced the liner filtering algorithm--kalman filter, introduced the basic principles and processing of the KF. Kalman filter is own pretty good filtering performance when the system is liner and non-Gaussian, but when the system is nonlinear and Gaussian, the filtering performance will descend. In order to solve this problem, this paper introduced a new nonlinear filter algorithm particle filtering, introduced the basic principles and processing of the PF.Secondly this paper analyses the merit and shortcoming of PF, and how to improve the particle filter, including the adopting good importance density functions and resampling. So we can reduce the degeneracy of particle filter. In view of the shortcoming of the PF, this paper introduced an improved particle filter--Regularized particle filter. Regularized particle filter algorithm is based on regularized resampling algorithm, according to disperse posterior distribution, we rebuild its sequence distribution, then sampling the particle from the approximate posterior distribution, it can reduce the particle degeneracy phenomenon.Thirdly this paper introduced Joint Probability Data Association (JPDA) and Interacting Multiple Models (IMM). Also this paper introduced basic principles of JPDA and IMM. In the IMM algorithm using regularized particle filter taking place of kahnan filter. The experiments are done using kahnan filter, particle filter and regularized particle filter. The simulating results show that the tracking precision and the tracking effect are enhanced. And then complete the software designing with VC++ and MATLAB. Lastly, the paper summarizes the research ideas and points out the remaining problems and the future of particle filter.