Research on the Algorithm of Swimming Posture Recognition System Based on MEMS Inertial Sensor

Aiming at the key issues of digital recognition of swimming postures, a research on swimming posture recognition based on MEMS inertial sensors and algorithm decision-making is proposed. Firstly, MEMS inertial sensors are used to monitor the three-dimensional acceleration and angular velocity data of swimming in real time, and a 60-dimensional statistical feature set is extracted based on the collected raw data. Then, the data from the experimental training is used for left-and-leave cross-validation and the algorithm decisions are continuously optimized and updated. Finally, the swimming postures were recognized by using the algorithm decision tree. The experimental results show that this algorithm can accurately recognize four swimming postures: butterfly, backstroke, frog and free-swimming. The accuracy rates of posture recognition for training and testing can reach 87.925% and 82.468% respectively.