跟踪窗自适应的Mean Shift目标跟踪算法

针对Mean Shift算法不能很好地跟踪尺度变化目标的缺点，提出一种将Mean Shift算法和目标几何特征相结合的目标跟踪算法。该方法在Mean Shift框架下提取目标的几何特征，根据目标的面积和型心坐标对跟踪窗的位置和大小进行修正，最后更新Mean Shift的目标模板。通过大量实验仿真证明，改进的算法能很好地跟踪尺度变化的目标，对目标的仿射变化和非刚性的形态变化具有有效性和鲁棒性。

Algorithm of target tracking based on Mean Shift with adaptive tracking window

Aiming at the deficiency of the Mean Shift algorithm in tracking target scale motions,a Mean Shift tracking algorithm combined with the target geometrical characteristic is proposed.In the framework of Mean Shift,the proposed algorithm extracts the geometric characteristics of the target and modifies the position and size of tracking window according to target area and core coordinate.Finally,the Mean Shift target template is updated.A large number of simulation experiments show that the proposed algorithm is good at tracking target scale motions,and is robust and effective to target affine motions and non-rigid shape changes.