摘要
传统舞蹈辅助训练方法中摄像头捕捉人体动作的误差较大,因此提出基于动作捕捉传感器的自助舞蹈辅助训练方法。根据视频帧值建立动作数据库及二维人体骨架关节模型,使用矩阵BVH解析算法将二维模型转换为三维模型,选择传感器进行动作捕捉,对结果进行滤波归一化处理,根据捕捉到的动作数值,带入人体动作模型,与运动数据库中的标准动作进行点位距离对比,提示动作错误位置,进行辅助训练。实验结果表明,采用不同方法对训练者动作进行测试,设计的自助舞蹈辅助训练方法进行动作识别时角度和关节坐标定位准确度更高,满足设计需求。
In the traditional dance training methods,the accuracy of camera capturing human motion is low,therefore,a self-help dance training method based on motion capture sensor is proposed.According to the video frame,an action database and a two-dimensional human skeleton joint model are established.The two-dimensional model is then transformed into a three-dimensional model by using the matrix BVH analytic algorithm.The sensor is selected to capture the action,and the result is filtered and normalized,the captured of which is brought into the human action model,and is compared with the standard action in the motion database to prompt the position when doing the wrong action and carry out auxiliary training.The experiment results show that different methods are used to test the trainer's movements.The angle and joint coordinate positioning accuracy of the designed self-help dance training method for action recognition is higher,which meets the design requirements.
作者
李晓君
LI Xiao-jun(Baoji vocational and technical college Fengxiang normal college,Baoji 721000,Shaanxi Province,China)
出处
《信息技术》
2021年第6期120-124,共5页
Information Technology
