摘要： 为对烟包PV带进行准确的伺服跟踪并完成剪带任务，提出一种基于运动估计的视觉跟踪算法。利用图像矩得到运动目标质心的坐标，根据质心的图像坐标计算出目标的速度、加速度。采用卡尔曼滤波算法预测运动目标在下一时刻的位置，并结合物体运动的速度、加速度作为伺服控制的依据，控制伺服电机的运动。实验结果表明，运动目标的速度误差在2 pixel/s以内，位置误差在5 pixel以内，验证了该算法的准确性和实时性。
Abstract: In order to achieve serve tracking of cigarette package PV belt and shear the PV belt more accurately and efficiently, a visual servo tracking algorithm based on motion estimation is proposed. The centroid of the moving object is obtained by using image moments, the speed and accelerator of the moving object can be obtained, and the location of the moving object can be predicted in the next time through Kalman filter. The operation of the servo motor depends on the speed, accelerator and location. Experimental results show that the speed error is less than 2 pixel/s, and the position error is less than 5 pixel. So the algorithm has the virtue of accuracy and real-time property through analyzing the speed and the location.
moving object tracking,