In the current teaching of physical education and dance in general colleges and universities, there are problems such as insufficient standardized training of movements and poor integration of aesthetics, which affect the teaching effect and the cultivation of students’ interests. Kinect 3D sensing technology provides a new way for movement detection, and the application of mathematical models can optimize the evaluation of training effect. In this study, through the world coordinate system conversion and dynamic sampling light projection algorithm, we realized the construction of three-dimensional model of human movement and assisted training based on joint coordinates and joint angle data. The experiment used random groups to conduct a semester-long comparative teaching experiment on 50 college students. The results showed that the students in the experimental group improved their sport dance motivation scores from 18.98±4.36 to 24.56±4.36 (P<0.01), and their skill learning scores from 18.12±3.42 to 23.73±2.55 (P<0.01), whereas the students in the control group did not have a significant improvement in motivation and skill learning. In terms of specialized technical performance, the experimental group was significantly higher than the control group in technical quality (82.48±7.36), music processing (81.14±6.85) and choreography and performance (79.76±4.51) (P<0.01). The study shows that the Kinect-based sports dance movement detection combined with aesthetic training can effectively enhance students' learning interest and special skills, providing a new optimization path for sports dance teaching in general colleges and universities.
