一种改进的粒子滤波目标跟踪算法

Improved particle filter algorithm for object tracking

传统的Condensation跟踪算法使用状态转移分布作为采样粒子的建议分布函数,没有考虑当前的观测值,大量的粒子运算浪费在了那些具有小似然性的区域。针对该问题,提出一种基于Mean Shift以改进建议分布函数的粒子滤波跟踪方法。实验表明,由于有效地利用了当前观测值,改进的算法具有较强的鲁棒性和实时性。

The conventional Condensation tracking method uses the state transition distribution as the proposal distribution.However,the transition distribution does not take into account the current observations,thus many particles can be wasted in low likelihood regions.In this dissertation,this paper proposed a particle filter tracking method based on Mean Shift to improve the proposal distribution.Experimental results show that the improved particle filter algorithm has robustness and real-time property.