摘要
对篮球运动员跳投动作视频三维监测,可以提高运动员跳投动作的精度。在进行篮球运动员跳投动作精度视频三维监测时,需要对不同角度的篮球运动员跳投动作图像进行采集和分析。传统监测方法只能从单一角度对跳投动作进行分析,导致不能获取完成的跳投动作视频图像,降低了篮球运动员跳投动作精度视频三维监测精度。提出改进KLT特征点的篮球运动员跳投动作三维重构方法。上述方法先利用KLT特征点算法采集不同角度的篮球运动员跳投动作图像,组建摄像机的投影矩阵,求得摄像机之间的外参数,可计算出每个图像匹配点所对应运动员跳投动作空间点的具体坐标,根据单视频图像序列中篮球运动员原地跳投过程的关节用力特征点二维数据,获取三维信息,完成篮球运动员跳投动作精度视频三维监测。仿真结果表明,所提算法可以精确地建立篮球运动员跳投动作三维模型,提高篮球运动员跳投动作精度。
Video three- dimensional monitoring can improve the basketball players＇ jump shot motion precision.It needs to collect and analyze jump shot motion image from different angles during monitoring. Traditional monitoring method can only analyze jump shot motion from single angle,it cannot obtain complete video image and reduce the video 3D monitoring accuracy of basketball players＇ jump shot motion. In this paper,a 3D reconstruction method for jump shot motion is proposed based on modified KLT feature point. Firstly,the jump shot images with different angles are collected and the projection matrix of camera to acquire external parameters between cameras is built using the KLT feature point algorithm. Then the spatial point specific coordinate of jump shot motion is worked out corresponding to every image feature point. Finally,the 3D information is obtained according to the feature point 2D data of stress on articulation during the standing shot process in image sequence of single video,and the video 3D monitoring on jump shot motion is completed. The simulation results show that the algorithm mentioned above can build 3D model of basketball players＇ jump shot motion accurately and improve the motion precision.
出处
《计算机仿真》
CSCD
北大核心
2016年第10期183-186,420,共5页
Computer Simulation
关键词
篮球运动员
跳投动作
三维监测
Basketball player
Jump shot motion
3D monitoring
