基于均值移位的篮球轨迹跟踪算法

Basketball trajectory tracking algorithm based on mean shift

针对当前跟踪篮球轨迹时存在跟踪精度低、关联率低的问题,提出基于均值移位的篮球轨迹跟踪算法。通过自适应全变差模型消除篮球图像中存在的Gibbs伪影和高斯噪声,采用均值移位算法对篮球轨迹进行跟踪,确定候选区域,通过均值漂移向量对搜索中心进行更改,利用巴氏系数定位目标位置,实现篮球轨迹的跟踪。实验结果表明,该方法轨迹跟踪的准确率较好,x、y方向位移误差在0.1m以内,关联率达90%以上。

To solve the problems of low tracking accuracy and low correlation rate when tracking basketball trajectory,a basketball trajectory tracking algorithm based on mean shift is proposed.The Gibbs artifact and Gaussian noise in the basketball image are eliminated through the adaptive total variation model,the mean shift algorithm is used to track the basketball trajectory and determine the candidate area.The search center is changed through the mean shift vector,and the Bap coefficient is adopted to locate the target position,achieving the tracking of the basketball trajectory.Experiment results show that the accuracy of trajectory tracking by this method is better,the displacement error in the x and y directions is within 0.1m,and the correlation rate is over 90%.