Analysis of the characteristics of skill-based street dance movements based on the improved K-means algorithm

Abstract

This paper first analyzes the K-mean algorithm from the core idea, algorithm process and advantages and disadvantages, then further improves the K-mean algorithm by using Gaussian mixture distribution and constructs the skill-based street dance movement recognition model based on the improved algorithm. Finally, the street dance teaching video is used as an example for dance movement acquisition and data pre-processing, and the recognition accuracy analysis of the street dance movement dataset is conducted based on the improved K-mean algorithm. The average recognition rates of the recognition model in the four data sets of the data set were 72.34%, 74.65%, 73.15% and 86.70%, respectively. This shows that analyzing the characteristics of street dance movements using the improved K-mean algorithm is beneficial for optimizing and improving existing street dance movements.

Keywords

• k-means algorithm
• gaussian mixture distribution
• action recognition model
• skill-based street dance
• data preprocessing
• 97p70