基于机器视觉的目标跟踪算法研究

Research of Object Tracking Algorithm Based on Machine Vision

针对遮挡情况下目标跟踪产生漂移的问题,提出一种基于MOSSE和Kalman滤波的目标跟踪算法以改善跟踪效果。算法使用MOSSE滤波器与目标图像的相关性度量来判断遮挡情况。无遮挡时,以MOSSE滤波器为主跟踪器,获取目标位置,更新MOSSE滤波器,并修正Kalman滤波器;遇到遮挡时,将Kalman滤波器设置为主跟踪器,预测目标位置,并保持MOSSE滤波器不变。实验从跟踪速度、精度、成功率等角度进行定性与定量分析,结果表明算法在遮挡情况下,实现目标的快速有效的跟踪。与同类算法比较,能有效改善遮挡情况下的跟踪效果,具有很强的鲁棒性。

In order to solve the problem of tracking drift in occlusion,an object tracking algorithm(MOSKal)based on MOSSE and Kalman filter is proposed to improve the tracking effect.The algorithm uses the correlation measurement between the object and the MOSSE filter to judge whether there is occlusion or not in every frame.When there is no occlusion,the MOSSE filter is used as the main tracker,the MOSSE filter is updated and the Kalman filter is modified according to the obtained position;when there is occlusion,the Kalman filter is set as the main tracker to predict position,and the MOSSE filter is kept unchanged.Qualitative and quantitative analysis of experiments were made from the aspects of tracking speed,accuracy and success rate.The results show that the algorithm can track target quickly and effectively with or without occlusion.Compared with other similar algorithms,the algorithm can effectively improve the tracking effect under partial and complete occlusion,and has strong robustness.