基于多特征融合的羽毛球扣杀动作轨迹跟踪  

Badminton smash trajectory tracking based on multi-feature fusion

作　　者：何佳佳 薛诗乐 何建强[3] HE Jia-jia;XUE Shi-yue;HE Jian-qiang(Physical Education Research Department,Shangluo University,Shangluo 726000,Shaanxi Province,China;Department of Physical Education,Xi’an International Studies University,Xi’an 710000,China;School of Telecommunications,Shangluo University,Shangluo 726000,Shaanxi Province,China)

机构地区：[1]商洛学院体育教学研究部,陕西商洛726000 [2]西安外国语大学体育系,西安710000 [3]商洛学院电信学院,陕西商洛726000

出　　处：《信息技术》2023年第11期68-72,共5页Information Technology

基　　金：商洛学院教改项目(20jyjx128);陕西省体育局项目(20-20233);陕西省教育厅专项科研计划项目(19JK0261);商洛学院科研项目(19FK002)。

摘　　要：羽毛球扣杀动作专业性较强、动作轨迹较为复杂,难以捕捉,因此提出一种基于多特征融合的羽毛球扣杀动作轨迹跟踪方法。构建轨迹观测模型,提取羽毛球扣杀动作的目标特征和距离特征;计算实际轨迹观测值与估计轨迹观测值间的巴氏距离,采用分布序列函数表现形式,将复杂的跟踪轨迹转换为基于贝叶斯状态的运动曲线推理问题;计算每个时刻下跟踪轨迹的状态参数,凭借帧数连续性实时状态分布结果完成轨迹跟踪。仿真实验证明,所提方法轨迹融合效果较好,跟踪曲线稳定性较强,可实现羽毛球扣杀动作的有效跟踪。Badminton smash action is highly professional and difficult to capture.Therefore,a badminton smash trajectory tracking method based on multi-feature fusion is proposed.The trajectory observation model is constructed to extract the target feature and distance feature of badminton smash.The Bhattacharyya Distance between the actual trajectory observations and the estimated trajectory observations is calculated,and the complex tracking trajectory is transformed into a motion curve reasoning problem based on Bayesian state by using the expression form of distribution sequence function.The state parameters of the tracking track at each time is calculated to complete the track tracking according to the real-time state distribution results of the continuity of the number of frames.Simulation results show that the proposed method has good trajectory fusion effect and strong tracking curve stability,and can realize the effective tracking of badminton smash.

关 键 词：多特征融合 加权数值 巴氏距离 高斯分布 分布序列函数 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]