一种基于传感器的人体上肢运动实时跟踪方法

实时跟踪人体运动是人机交互的重要研究课题，可以广泛应用于虚拟现实、虚拟人运动合成、聋人手语自动生成、计算机3D动画、机器人运动控制、远程人机交互等领域。文中介绍了一种基于传感器的人体上肢运动实时跟踪方法，给出了该方法的虚拟人模型、计算原理与校正方法，最后介绍了整个方法的实现以及在中国聋人手语自动合成中的应用。该方法具有使用传感器少，运动跟踪精度高、计算过程简单而且速度快等特点。

A Sensor Based Method to Capture Motion of Two Arms

Motion capture is a popular process for human computer interaction, and has many applications in virtual reality, virtual human motion synthesis, automatic generation of sign language, 3D computer animation, robot control, etc. This paper presents a sensor based method, in which only 3 sensors are used, to capture motion of two arms. A joint based virtual human model used in this method is given. The processes of dealing with the magnetic sensor information, including calculation of shoulder and elbow's joint angles, are described. At the same time, a sensor calibration method based on rotation axes of arms is detailed. Using this method, sensor information from a performer can be transformed rapidly and correctly into joint angles of a human body, which can be used to generating human animation. In fact, we have built a 5500 words motion library of Chinese Sign Language using this method. In this library, the motions of all words are recorded from real performers who are teachers of a deaf school, and can be understood by deaf people well because of its real and natural property. Based on this motion library, we developed a Chinese Sign Language synthesizer that can translate Chinese text into human animation of Chinese Sign Language for deaf people in education or other Internet applications.