一种改进的粒子滤波跟踪算法

Improved Particle Filter Tracking Algorithm

传统粒子滤波跟踪算法的退化现象和巨大的计算量不利于其应用,尤其在实时性要求较高的视频监控场合。引入均值漂移算法进行粒子的采样调整,采用积分直方图加快每个粒子的直方图计算速度,以改进传统粒子滤波跟踪算法的速度和跟踪效果,满足实时跟踪需要。实验结果证明了改进算法的有效性。

The degeneracy problem and the huge computational cost limit the usage of the traditional particle filter tracking algorithm, especially in video surveillance occasions which required higher real-time processing. The mean-shift algorithm is introduced to make particles herd to the nearby local maximum position, and the integral histogram can speed up the computing of the histogram of each particle. The speeds and effects of the traditional particle tracking algorithm are improved, and can meet the needs of real-time tracking. Experimental results prove the effectiveness of the improved algorithm.