Research on Badminton Motion Recognition Based on Hidden Markov Model

Abstract

In order to improve problems such as uncoordinated movements of exercisers and sports injuries, a badminton motion recognition method based on hidden Markov model is proposed, which uses non-visual sensing motion recognition based on multiple sensors to find out the problems existing in the technical movement during the movement and obtain the best motion effect. Firstly, four wearable inertial sensors were placed in several important parts of the experimental object, and each sensor collected signals of triaxial acceleration and angular velocity signals in three-dimensional space. Secondly, the HMM model is established through data acquisition, preprocessing, window segmentation, feature extraction and selection, classification and recognition. Finally, through three comparative experiments, the recognition rate was raised from 91.99% and 91.60% to 98.1%, effectively improving the recognition rate of badminton movements. The results of this study can provide scientific solutions to the technical movement problems existing in exercitation and promote the health management of the whole life cycle.