The Three-dimensional Image Analysis of Value Evaluation of Waltz Square Step in Sports Dance by Big Data

The primary challenge in sports dance research is its long-standing reliance on subjective, experience-based instruction. This approach lacks effective tools for high-precision, quantitative analysis of technical movement details, which has limited the scientific improvement of teaching quality. To address this issue, this study introduces a markerless three-dimensional image analysis method that combines music-beat synchronization with an improved optical flow recovery model. For the first time, the complete movement cycle of the Waltz Square Step (WSS) was divided into seven refined phases according to musical rhythm (Beats Per Minute, BPM = 30). In addition, the Hanavan human body model was applied to enable a coordinated quantitative analysis of skeletal and muscular motion parameters. This approach overcomes major drawbacks of traditional motion capture technologies in teaching contexts, such as high costs and operational complexity. The results show that skilled dancers demonstrated clear technical advantages over average dancers when evaluated using this method. Their center-of-mass displacement fluctuation was reduced by 39% (Phase 3), knee joint angular variation decreased to 8.5 degrees, and the movement economy index improved to 0.92. The method also achieved high accuracy, with a center-of-mass displacement error of only 0.12 m, a joint angle error of 1.3 degrees, and a motion recognition accuracy of 94.6%. Overall, these findings provide unprecedented quantitative evidence for the teaching and training of the WSS. More broadly, they illustrate a research paradigm that can be applied to performance-oriented, skill-intensive disciplines.