Abstract: In order to avoid the tracking failure based on single feature under the conditions of cluttered backgrounds and illumination changes, a robust tracking algorithm was proposed based on multi-feature fusion and particle filter. Multi-block color histogram based on HSV was used to describe the overall distribution characteristics of the target and histogram of oriented gradients containing some construction information. The two features were fused in the frame of particle filter. Meanwhile, the weighs of fusing strategy, template and noise distribution parameters were updated online, and the particle number of the features was adjusted dynamically. The experimental results show that the proposed method is of higher robustness and provides more accurate result.

Key words: particle filter, multi-feature fusion, color histogram, histogram of oriented gradients, template updating

摘要： 为克服基于单一特征的跟踪方法在复杂环境和光照变化下易导致跟踪失败的缺点，提出了一种基于多特征融合的粒子滤波跟踪算法。通过基于HSV的多块颜色直方图来表征目标的总体分布，而方向梯度直方图又包含了一定的结构信息，两者互为补充，将两者融合于粒子滤波的框架中。同时，自适应更新融合权重、模板和噪声分布参数，动态调节粒子数目，在环境干扰较大(如遮挡)时，分配较多的粒子。实验结果表明，算法鲁棒性较高，同时提高了跟踪的精度。

关键词: 粒子滤波, 多特征融合, 颜色直方图, 方向梯度直方图, 模块更新