动态背景下基于粒子滤波的运动目标跟踪方法

摘要

On an intelligent video monitor system, it is a hard problem to accurately track a moving target in a dynamic background. A particle filter based method is taken in the paper to track a moving target in a dynamic background. The method is based on Bayesian estimation, represents probability by particle sets, and approximates the optimal estimation through recursive Bayesian filter. Experimental results prove that, as an effective target tracking method, the proposed method can accurately track a moving target in a dynamic background.%在智能视频监控系统中,实现对动态背景下的运动目标准确跟踪是一个难点问题.使用一种基于粒子滤波的方法来对动态背景下的运动目标进行跟踪.该方法基于贝叶斯估计,利用粒子集来表示概率,通过递推的贝叶斯滤波来近似逼近最优化的估计结果.实验结果证明,该方法可准确跟踪动态背景下的运动目标,是一种有效的目标跟踪方法.