基于概率假设密度的多目标视频跟踪算法

研究目标数变化的多目标视频跟踪问题.首先阐述了概率假设密度(PHD)滤波的基本原理;然后给出序列图像多目标跟踪系统的运动目标检测算法、状态方程、观测方程以及基于高斯混合概率假设密度(GM-PHD)的多目标视频跟踪算法的具体实现.该算法有效解决了新目标出现、目标合并、目标分裂及目标消失等多目标跟踪问题.实验结果表明,该算法在复杂场景下具有较强的鲁棒性,能有效实现目标数变化的多目标视频跟踪.

Probability hypothesis density filter based multi-target visual tracking

The issue of tracking a variable number of multiple targets is discussed in this paper. The theory in relation
to probability hypothesis density(PHD) filter is given firstly. Then the motion detection, dynamic equation, measurement
equation and visual multi-target tracking algorithm based on Gaussian mixture probability hypothesis density(GM-PHD) are
presented in details. The proposed method can track objects correctly when they appear, merge, split and disappear in the
field of view of a camera. Experimental results show that GM-PHD based multi-target visual tracking is robust in clutter and
can effectively track a varying number of targets.