An Inertial Sensing-Based Approach to Swimming Pose Recognition and Data Analysis

Abstract

In this paper, inertial sensing is used to identify a swimming stance and analyze its swimming stance data. A wireless monitoring device based on a nine-axis microinertial sensor is designed for the characteristics of swimming motion, and measurement experiments are conducted for different intensities and stances of swimming motion. By comparing and analyzing the motion characteristics of various swimming stances, the basis for performing stroke identification is proposed, and the monitoring data characteristics of the experimental results match with it. The stance reconstruction technology is studied, PC-based OpenGL multithreaded data synchronization and stance following reconstruction are designed to reconstruct the joint association data of multiple nodes in a constrained set, and the reconstruction results are displayed through graphic image rendering. For the whole system, each key technology is organically integrated to design a wearable wireless sensing network-based pose resolution analysis and reconstruction recognition system. Inertial sensors inevitably suffer from drift after a long period of position trajectory tracking. The proposed fusion algorithm corrects the drift of position estimation using the measurement of the visual sensor, and the measurement of the inertial sensor complements the missing measurement of the visual sensor for the case of occlusion of the visual sensor and fast movement of the upper limb. An experimental platform for upper-limb position estimation based on the fusion of inertial and visual sensors is built to verify the effectiveness of the proposed method. Finally, the full paper is summarized, and an outlook for further research is provided.