提出了一种将粒子滤波和CamShift 相结合的多特征视觉跟踪方法。通过CamShift 对粒子的位置和尺度同时进行优化,使得跟踪窗口能随着目标尺度的大小变化相应调整。同时采用自适应方式将颜色信息和运动信息在CamShift 优化的粒子滤波框架下有效结合起来。该方法使用CamShift 对粒子传播进行优化,每个粒子都收敛到目标附近,粒子的有效性得到提高。实验结果表明,使用10 个粒子的CamShift 优化的粒子滤波的跟踪误差小于100个粒子的传统粒子滤波的跟踪误差。并且由于多特征的使用,目标在受到背景相似物体干扰和场景光线发生显著变化等情况下仍能实现稳定的跟踪。用较少的粒子就能实现稳定的跟踪,减少了计算代价,提高了跟踪的鲁棒性。

A multi-feature vision tracking method by CamShift optimizing particle filter was proposed. The position and scale of the particle were optimized by CamShift, and the algorithm can select the proper size of the tracking window in the scenarios that the object scale varies. Meanwhile, by adopting an adaptive method, the color information was combined with motion information in the framework of CamShift optimizing particle filter. Because the propagation of the particle was optimized, which could make each particle converge on the true state region of target, the effectiveness of particle is improved. The experimental results show that the tracking error of CamShift optimizing particle filter with only 10 particles is less than that of the traditional particle filter with 100 particles. What's more, as a result of the use of multi-feature, it can still track the target accurately and solve the problems encountered in tracking an object with illumination variation and the background color clutter. Stable tracking can be realized by a few particles, which can reduce calculation cost and improve the robustness of tracking.