摘要
粒子滤波方法是一种针对非刚性目标运动跟踪的有效工具。运用基于贝叶斯估计的粒子滤波算法,对复杂的运动背景下目标移动进行跟踪。论述了贝叶斯估计理论,推导粒子滤波过程,并将状态粒子决定的区域所对应的色彩直方图用作测量,与目标参考直方图相比较,得出最佳的后验估计。运用窗口粒子平均方法确定目标的坐标,实现跟踪。算法采用单目标以及多目标序列图象进行跟踪实验,并与均值移动(mean-shift)跟踪算法结果进行比较,证明该跟踪算法更为有效。
Particle Fiher（PF） is an efficient tool for non-rigidity objects tracking.The paper presents a Bayesian-based PF method for objects tracking in dynamic scenes and discusses the Bayesian estimation algorithm and the PF process.Used the color histograms correspond to the particles＇ rectangles in the sequence images as the measurement and they are compared to the reference histogram so that to obtain the optimized posteriori probabilities.The robust mean technique is chosen to ascertain the objects＇positions.Single object and multi objects tracking are performed in the experiments.The results are compared with that of mean-shift algorithm.It＇s proved that this method is more efficient.
出处
《计算机工程与应用》
CSCD
北大核心
2007年第8期62-64,102,共4页
Computer Engineering and Applications
关键词
目标跟踪
贝叶斯估计
粒子滤波
色彩直方图
object tracking
bayesian estimation
Particle Fihe （PF）
color histogram
