基于关节点检测的引体向上评测系统

Pull-Up Evaluation System Based on Joint Point Detection

传统的引体向上测试主要通过裁判员进行人工测量计数,依赖裁判员的专业素质和业务能力,主观性较强且评判标准不统一,测试人员较多时会加大裁判员的工作负担,影响测试效率。针对上述问题,将物联网技术和3D体感技术应用于引体向上评测系统的设计,采用奥比中光摄像机获得人体实时骨骼图像数据流,通过分析引体向上的状态机,设计多个状态机之间的转换匹配算法,实现对引体向上的运动识别。系统以微软基础类(Microsoft Foundation Classes,MFC)为编程框架开发了上位机,综合引体向上识别和人脸识别功能,并使用网站开发技术将测试人员的个人信息和运动数据显示在Web端。经过多组实验测试和数据分析,系统对引体向上的识别率可以达到98%,能够满足体能训练和运动考核的引体向上测量计数任务需求。

The traditional pull-up test is mainly measured and counted manually by the referee,this method depends on the professional quality and professional ability of the referee,the subjectivity is strong and the evaluation standard is not unified,when there are more testers,it will increase the work burden of the referee and affect the testing efficiency.In order to solve the above problems,the 3D somatosensory technology is applied to the design of the pull-up evaluation system,and the real-time human bone image data stream is obtained by using the Orbbec camera.By analyzing the pull-up state machine,a template matching algorithm for the conversion between multiple state machines is designed to realize the motion recognition of pull-up.The system uses Microsoft Foundation Classes(MFC) as the framework to develop the host computer,integrates pull-up recognition and face recognition functions,and uses website development technology to display the tester's personal information and motion data on the web side.After several groups of experimental tests and data analysis,the recognition rate of the system for pull-up can reach 98%,which can meet the human body pull-up measurement and counting task of physical fitness training and exercise assessment.