摘要： 针对均值移动算法鲁棒性差以及粒子滤波算法计算量大、难以满足实时跟踪的特点，提出2种先均值移动后粒子滤波的融合算法，分别为粒子数目保持恒定的融合算法和粒子数目自适应的融合算法。实验结果证明，与已有算法相比，2种算法在实时性提高的同时，跟踪准确性和抗干扰能力没有明显下降。

关键词: 目标跟踪, 均值移动, 粒子滤波, 融合算法, 自适应, 实时跟踪

Abstract: This paper proposes two fusion algorithms based on poor robustness of Mean-shift algorithm and large computation of particle filtering algorithm. Both of the new algorithms take Mean-shift first and particle filtering later. The number of particles is constant and the number of particles is adaptive. Experimental results show that compared with existing algorithms, the real-time of new algorithms is improved, while tracking accuracy and anti-interference ability without significant decline.

Key words: object tracking, Mean-shift, particle filtering, fusion algorithm, self-adaptive, real-time tracking

中图分类号: 

• TP301.6