Tap dance, as a dance form with distinctive rhythm and expressive power, has been widely used in dance professional courses. However, the traditional tap dance training methods have certain limitations, which make it difficult to provide accurate feedback on the quality of dancers’ movements and training effects. In this study, an innovative tap dance movement training path was designed by introducing a leg movement recognition technique that combines an improved particle swarm optimization algorithm with support vector machine (SVM). First, wavelet threshold denoising and time-domain feature extraction are performed on the sEMG signals, and the parameters of the support vector machine model are optimized by combining the time-frequency combination features in order to improve the recognition rate of tap dance leg movements. The experimental results show that the average recognition rate based on the WL-MPF time-frequency combination features is 97.71%, which is significantly higher than that of the traditional single-feature recognition methods (e.g., the recognition rate of WL features is 95.85%). In addition, the experimental group performed significantly better than the control group in tap dance training, and the difference in total course performance was statistically significant (P<0.05). By introducing the intelligent leg movement recognition technology, the training path proposed in this paper can not only improve the accuracy of training, but also enhance students' interest in dance and improve the learning effect of dance movements. The method has high application value in tap dance teaching.
