Particle Filter Algorithm Of Target Tracking Research

Target tracking technology plays an important role in both the military and civilian, which is widely applied in aerospace, medical, national economy and other areas. The key of target tracking is to extract the target state information; therefore, the filtering algorithm for recursive target state estimation is an important part of the target tracking technology. The state space models of target tracking systems often represent nonlinear non-Gaussian in real situations. Particle filter (PF) plays more excellent performance in nonlinear and non-Gaussian situations than the traditional Kalman filter and the improved algorithms of Kalman filter, so it has been widely focused by researchers in target tracking.The basic particle filter algorithm has some shortcomings, such as particle degeneracy, and particle diversity after resampling process. In this paper, several shortcomings in PF are studied in the framework of PF; some improved algorithms are proposed and applied in several single target tracking problems. The main achievements are as follows:1. To deal with the problem of particle diversity after basic resampling process, a new resampling algorithm based on random Sigma points-RSPR algorithm is proposed. The proposal algorithm is used of splitting random Sigma points to improve particles diversity. By the simulation of two general models and the passive radar with a single target tracking model, it show that the proposal algorithm can alleviate the loss of particle diversity effectively.2. To improve the accuracy of the state estimation for the basic particle filter, a novel resampling algorithm which is introduced the current observational information is proposed. The novel algorithm moves the particle points toward the high likelihood region to improve the final state estimation accuracy. By the simulation of two general models and the passive radar with a single target tracking model, it show that the proposal algorithm not only improve the accuracy of the state estimation, but also increase particle diversity effectively.3. To deal with the problem of the particle degeneracy, the latest improved method of particle filter-particle flow for nonlinear filter is studied. The new filter smoothly migrate the set of particles used to represent the prior density into a new set of particles that represent the posterior density, the new method can solve the problem of particle degeneracy and realize the parallel computing process. Based on the simulation results, it shows the new algorithm can deal with particle degradation, but the new method has some shortcomings.