基于姿态估计的运动辅助训练系统研究

Research on motion aided training system based on attitude estimation

文中针对羽毛球运动系统化、科学化和高效化训练的需求,研究了基于姿态估计的羽毛球运动辅助训练系统。该系统采用分部三维姿态估计的方法,将三维人体姿态估计问题简化为多个二维子问题。通过采用贝叶斯算法对人体各个部位进行特征识别,并利用人体运动约束完成了三维姿态合成。最后,通过整体姿态估计空间的调整进行优化。系统采用VB程序语言和MicrosoftOf?ficeAccess数据库进行开发,完成了客户端接口、数据库及中间连接层的数据访问接口的构建,同时利用Java语言实现了对运动分析结果的输出展示功能。本研究功能对于实现运动员的科学、合理、高效训练均具有重要意义。

Traditionally,aiming at the need of systematic,scientific and efficient badminton training, the badminton auxiliary training system based on attitude estimation is studied. The system simplifies the problem of three- dimensional human pose estimation into several two- dimensional sub- problems by using the method of piecewise three-dimensional attitude estimation. The feature recognition of each part of human body is carried out by using Bayesian algorithm,and the three-dimensional pose synthesis is completed by using human motion constraints. Finally,the optimization is carried out by adjusting the overall attitude estimation space. The system is developed by using VB programming language and Microsoft Office Access database. The client interface,database and intermediate connection layer constitute the data access interface. At the same time,the output display function of the motion analysis results is realized by using Java language. The research function of this paper is of great significance to the scientific,rational and efficient training of athletes.