匹配追踪下的篮球运动员投篮运动轨迹识别  

Recognition of Basketball Players’Shooting Trajectory Under Matching Tacking

作　　者：牛磊 NIU Lei(Bozhou University,Bozhou Anhui 236800,China)

机构地区：[1]亳州学院体育系,安徽亳州236800

出　　处：《铜陵学院学报》2022年第4期102-105,109,共5页Journal of Tongling University

基　　金：亳州学院校级质量工程项目“篮球运动”(2020XXKC01)。

摘　　要：以提升投篮运动轨迹识别水平为目的,提出基于匹配追踪算法的篮球运动员投篮运动轨迹识别方法。去除篮球运动员比赛视频内的复杂背景后,提取投篮运动轨迹特征;将其作为输入构建隐马尔可夫模型,利用该模型计算篮球运动员投篮运动轨迹状态概率后,得到待识别轨迹样本序列;使用Biterbi算法计算该数组内最大似然数值并对其进行排序处理后,以前若干位最大似然数值为约束条件,使用数组索引方式描述分类结果,该结果即为投篮运动轨迹识别结果。实验结果表明:该方法识别篮球运动员投篮运动轨迹可靠性较高,且识别结果与实际轨迹重合度极高。In order to improve the recognition level of shooting trajectory,a basketball player shooting trajectory recognition method based on matching tracking algorithm is proposed.After removing the complex background in the basketball player’s game video,the characteristics of shooting trajectory are extracted;taking it as the input,a hidden Markov model is constructed.After using the model to calculate the state probability of basketball players’shooting trajectory,the trajectory sample sequence to be identified is obtained;after using the Biterbi algorithm to calculate the maximum likelihood value in the array and sort it,the maximum likelihood value of the previous several bits is the constraint condition,and the array index is used to describe the classification result,which is the recognition result of shooting trajectory.The experimental results show that this method has high reliability in identifying basketball players’shooting trajectory,and the coincidence degree between the recognition result and the actual trajectory is very high.

关 键 词：匹配追踪算法 篮球运动员 运动轨迹 隐马尔可夫 背景差法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]