摘要
粒子滤波算法是一种基于贝叶斯估计的蒙特卡罗方法,适用于非线性非高斯系统的分析,被广泛应用于跟踪、定位等问题的研究中。为了解决粒子滤波算法在重采样后,丧失粒子多样性的问题,本文在粒子滤波算法的重采样步骤后,加入了马尔可夫链蒙特卡罗(Markov Chain Monte Carlo,简称MCMC)移动步骤,增加粒子的多样性。利用粒子滤波算法和MCMC粒子滤波算法对目标跟踪问题进行了仿真,并且通过分析仿真实验结果,比较了两种算法的性能,结果说明加入MCMC粒子滤波算法的性能优于粒子滤波算法。
The particle filter algorithm is a Monte Carlo method based on Bayesian, it was appropriate to the estimation for nonlinear and non-Gaussian systematic analysis, and has been widely used in tracking, targeting issues. To address the problem that the particle diversity is losing after the resampling step. The Markov chain Monte Carlo （Markov Chain Monte Carlo, referred to as MCMC） move step was joined after the particle filter algorithm resampling steps, to increase the diversity of particles. The target tracking problem was simulated with the particle filter algorithm and MCMC particle filter algorithm, and through analyzing the simulation results, a comparison between particle filter algorithm and improved particle filtering algorithm performance was made. The results show that the MCMC particle filter algorithm is superior to the particle filter algorithm,
出处
《电子测试》
2009年第12期19-22,86,共5页
Electronic Test
