Movement Recognition of Sports Teaching and Training Based on Feature Reduction and Gauss Mixture Model

In order to overcome the shortcomings of current sports movement recognition methods and obtain better sports movement recognition effect, a sports movement recognition method based on feature dimensionality reduction and Gauss mixture model is proposed. Firstly, the video images of sports movements are collected, and the feature vectors of sports movements are extracted. Then, the feature vectors are reduced by using the random projection algorithm. Finally, the training samples after dimensionality reduction are learned by using the Gauss mixture model, the recognition model of sports movements is constructed, and the performance is tested by using various sports movement data sets. The results show that the method achieves ideal recognition results of sports movement, and the recognition accuracy is higher than other sports movement recognition methods.