基于粒子滤波的运动物体检测跟踪

Moving object detection and tracking based on particle filtering

提出了一种状态空间模型粒子滤波算法,并应用于运动目标的跟踪。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结果,在预设搜索区域用粒子群找到和目标模板最相似的中心位置,并以该位置作为观测值,进行跟踪。仿真实验结果和两种实际条件下效果比较表明该算法在跟踪低常速运动中精准性高,是一种有效的目标跟踪方法。

A particle filtering algorithm of state space model is put forward in this paper to track moving targets. The method based on Bayesian estimation uses the particle set to represent the probability. The recursive Bayesian filtering is a- dopted to approximate the optimization results. The particle swarm is employed in the presetting search area to find the central location most similar to target template. The location is taken as the observation value for the next frame tracking. The simu- lation result and comparison of the two actual effects show that the algorithm is an effective method for tracking the targets which moves in low normal speed.