关键词： 视频跟踪, 粒子滤波, 时序重要性采样(SIS), 算法, 重新采样

Abstract: The problem of visual tracking under non-Gaussian and non-linear assumption is addressed.Recent advances in sequential Monte Carlo techniques,especifically in particle filter algorithm,allow us to model and track the posterior distribution defined by Bayesian model using a collection of targets states.That collection can be viewed as samples from the posterior of interest.The proposed algorithm takes the degeneracy problem and resampling into consideration.Experimental results show that our approach can yield robust and stable tracking performance.

Key words: visual tracking, particle filter, sequential importance sampling (SIS) algorithm, resampling