基于二值掩码图像的粒子滤波目标跟踪快速算法

摘要

粒子滤波算法是进行运动目标跟踪的一种重要方法.针对传统粒子滤波算法在进行目标跟踪时存在的计算量大、实时性不足的问题,提出一种基于二值掩码图像的粒子滤波目标跟踪快速算法.该算法在传统粒子滤波算法的每个帧处理阶段产生二值掩码图像,再结合权重选择方法移除背景中权重较小的粒子,保留权重较大的重要粒子.提出的算法可以有效减少参与计算的粒子数目,节约算法的计算成本,从而提高目标跟踪的实时性.与传统粒子滤波算法进行比较,实验结果表明,提出的算法不仅能够有效地提高跟踪速度,而且跟踪结果的准确性和鲁棒性也有所增强.%Particle filter algorithm is an important method in the field of moving object tracking. A fast particle filter algo-rithm for object tracking based on binary mask image is proposed to solve the large calculation and poor real-time prob-lems in traditional particle filter tracking algorithm. In the proposed algorithm, a binary mask image is generated in each frame processing stage of traditional particle filter, then combined with a weight selection method, the particles with low-er weight in the background of the mask are removed and the important particles with larger weight are retained. The pro-posed algorithm can effectively save the calculation cost by reducing the number of particles involved in calculation, and improve the real-time performance of object tracking. The experimental results compared with the traditional particle fil-ter algorithm show that the proposed algorithm can not only improve the tracking speed effectively, but also improve the tracking accuracy and robustness.