Design And Implementation Of Human Motion Capture System Based On Multi-sensor Nodes

As a new type of human-computer interaction techniques motion capturetechnology has an wide application prospect in the fields such as medical rehabilitation,film, animation, games and so on. With the rapid development of microelectronicstechnology, chip function has been highly integrated. Human motion capture based onMEMS sensor chip is moving in size small, simplified structure forward. This paperanalyzes the methods and principles of the technology framework, proposed the systemfunctional requirements, according to the requirements of the module design andprogramming, to achieve human motion capture. The main research consists of threeparts.1. Based on the STM32microprocessor and LSM303DLH, L3GD20sensor chip, aprogram is designed to obtain the human motion measurement data. An Extend KalmanFilter is designed for the multi-sensor data fusion. Real-time calibration procedure isdesigned, through it calibration parameters can write into FLASH memory of nodemodule anytime.2. Analysed the data communication requirements, a communication protocolbased on the STM32microcontroller SPI interface is designed to achieve the data’s timesynchronization of multi-nodes. A serial communication program is designed to uploadthe motion data to PC. In order to provieds the function of communication for other PCapplication, a serial communication program has been designed as a dynamic linklibrary.3. Diretc3D character models animation program is designed. It use serialcommunication, support by the dynamic link library, to read real-time data of humanmotion, driving action figure model’s transform and display it in screen. The humanbody posture directional calibration procedure is designed for the initialization ofsystem. Finally, we verify that the algorithm to capture the human movement is correct,and the performance is reliable by the experiment of human’s arm movement capture.