田径运动终点裁判的智能视觉分析研究

Track and Field at The End of the Referee Intelligent Visual Analysis and Research

在裁判田径比赛的名次的过程中,由于运动速度过快,视觉对运动员的终点冲刺进行分析时,容易产生对终点冲刺中的像素拖拉延迟问题,如采用高频采集的方式进行解决,以最小采样周期消除干扰,过小的采样频率无法建立帧差特征,导致根据帧差特征的判断方法失效。提出一种改进冗余小波变换域算法的田径运动终点裁判智能视觉判断方法。对运动员冲刺过程视频图像进行预处理,将各帧图像中运动员冲刺终点过程与背景有效的分离,构造运动员冲刺终点特征的不变量,利用相似函数计算相邻帧冲刺终点特征相似度。利用冗余小波变换在小波域提取终点冲刺的区域,采用mean-shif算法利用运动员冲刺终点的色彩分布信息描述其终点冲刺的变化过程。仿真结果证明,上述方法能对比赛名次作出准确的裁判。

In this paper, we proposed an intelligent vision judging method at the destination of track and field based on improving redundant wavelet transform domain algorithm. Firstly, we preprocessed video images during sprint of athlete. Then we separated background effectively from the sprint process of each frame image to build invariant of sprint destination features. The similarity function was utilized to calculate similarity of adjacency frame at sprint destination. We also used redundant wavelet transform to extract region of sprint in wavelet domain. Finally we used distribution information of color to describe changing progress of destination sprint adopting mean-shift algo-rithm. The simulation results show that the method mentioned above can judge competition ranking accurately.