Tennis Player Pose Recognition Algorithm Based on IMU Sensor and KNN

Tennis players’ posture recognition is of great significance for improving their technical level and preventing injuries. Therefore, in order to master the posture of tennis players, a tennis players’ posture recognition algorithm based on inertial measurement unit (IMU) sensor and K nearest neighbor (KNN) is proposed. The algorithm uses the IMU sensor to sense various key data of athletes, and calibrates the three-axis accelerometer, three-axis gyroscope and three-axis magnetometer of the sensor to improve the data perception effect. On this basis, a two-stage Kalman filter is used to process a variety of key sensed data. After reducing the noise in the data, the time-domain features in the data are extracted, and the extracted feature results are input into the KNN algorithm. The algorithm calculates the distance between feature vectors according to the mixed distance, and determines the category frequency of KNN points in each data sample according to the voting principle, the category corresponding to the maximum frequency is the result of tennis player pose classification and recognition of input samples. The test results show that the algorithm can better complete the perception of multiple key data of athletes. After calibration, the error range of three-axis gyroscope and three-axis accelerometer in different directions is between 0.15 and −0.10[Formula: see text]g and 0.17 and −0.25[Formula: see text]g, respectively, and the center of the ellipsoid should be more consistent with the coordinate origin. The average error of roll angle, pitch angle and yaw angle perception is within [Formula: see text], and the median error is not more than [Formula: see text].