基于背景差法的篮球投篮轨迹自动捕捉方法研究

Research on basketball shooting trajectory automatic capture method based on background difference method

针对原有篮球投篮轨迹捕反义捉方法数据聚类效果较差的问题,设计基于背景差法的篮球投篮轨迹自动捕捉方法。通过传感器对比结果选定惯性传感器,设计轨迹捕捉器并构建相应的轨迹捕捉网络构架。在运动员小臂处与篮筐篮板入放置捕捉器获取投篮动作数据并对其去噪处理。使用背景差法中的多高斯模型构建投篮轨迹模型。对模型内数据点进行聚类处理,剥离背景数据获取轨迹数据,最后完成轨迹数据的聚类捕捉。至此,基于背景差法的篮球投篮轨迹自动捕捉方法完成。构建对比实验,与原有方法相比,此方法对数据点的聚类效果更佳,原有方法聚类效果较为松散易造成数据缺失,轨迹捕捉精度较低。因而,此方法的轨迹捕捉效果更为优越。

Aiming at the problem that the data clustering effect of the original basketball shooting trajectory capture method is poor,an automatic basketball shooting trajectory capture method based on the background difference method is designed.The inertial sensors are selected to design the trajectory capture and the corresponding trajectory capture network architecture is constructed.A trap is placed at the forearm of the player and the backboard of the basket to obtain the shooting motion data and to de-noise it.The shot trajectory model is constructed by using the multi-gaussian model in the background difference method.The data points in the model are clustered,the background data are stripped to obtain the track data,and finally the track data is clustering captured.Thus,the basketball shooting trajectory automatic capture method based on background difference method is completed.Compared with the original method,this method has better clustering effect on logarithmic data points.The clustering effect of the original method is relatively loose,which can easily cause data loss and track capture accuracy is low.Therefore,the effect of trajectory capture is superior.