基于Camshift的改进人脸跟踪算法

Improved face tracking algorithm based on Camshift

文中利用目标加速度运动位移方程,预测下一时刻目标可能移动的位置,使用预测位置误差方程,估测运动目标搜索范围,并且通过启动多个Camshift跟踪器的方法,改进Camshift算法。仿真实验表明,该方法有效地克服了Camshift算法自身的缺陷,即使是加速运动的目标,也可准确地预测运动目标的位置,并且有效提高了对遮挡目标跟踪和多个人脸目标跟踪的鲁棒性。

In this paper, the target acceleration motion displacement equation is used to predict where the target may move to at the next time, while the prediction position error equation is used to estimate the search range of the moving target. And the paper improves the Camshift algorithm by starting multiple Camshift trackers. The simulation results show that the Camshift algorithm after improving in this paper effectively overcomes the shortcomings of its own. Even if the movement of the goal is accelerating, the method can also accurately predict the location of moving targets. And the algorithm effectively improves the robustness of the block target tacking and multi-face tracking.