一种导向粒子滤波跟踪算法

Guided particle filter tracking algorithm

粒子滤波算法在运动目标跟踪方面有着广泛的使用,粒子滤波中的重采样是解决粒子退化的一种重要方法,但是重采样会导致粒子的多样性的丧失。针对这个问题,改进粒子滤波算法,改进过程中结合了导向滤波的基本思想,因此将这种方法称为导向粒子滤波跟踪算法。导向滤波是近几年提出的一种新的滤波方式,与传统滤波相比,它在滤波的时候会引入一幅指导图像,鉴于这个思想,我们在进行粒子滤波的时候,引入一种导向粒子作为一个指导量,来保留一些目标图像上的信息。实验证明了这种算法可以更好地对目标进行定位跟踪。

The particle filter algorithm has been widely used in the moving target tracking.In the particle filter,the resampling is an important way to solve the particle degeneration,but the resampling can lead to the loss of the particles＇ diversity.To solve this problem,the guided particle filter tracking algorithm is proposed.The guided filtering is a new filtering method.Compared with the traditional filtering methods,the guided filtering will introduce a guided picture when filtering the image.Based on this idea,the guided particle filter will introduce a guided particle as a guide to remain some informations of target images when filtering the image.The experiment proves that this algorithm can locate and track the target better.