Hockey Trajectory Prediction Combining Visual Object Alignment Algorithm and Template Matching Model

ABSTRACT In order to improve the effectiveness of targeted training after hockey matches, a hockey trajectory prediction method combining visual object alignment algorithm and template matching model is proposed. The advantage of this method is that it effectively solves the impact of complex perspectives and background changes on trajectory prediction accuracy through visual object alignment algorithms, achieving high consistency in the competition scene and improving prediction stability. Meanwhile, the template‐matching model can accurately detect the rotation angle of the hockey, ensuring precise prediction of the motion trajectory. Compared with traditional methods, this method improves prediction accuracy by about 4.1% and increases computational efficiency by about 6.5% and 5.6%, respectively, when processing 30 and 50 frames of images. The experimental results show that this method not only exhibits good real‐time performance and stability in dynamic competition scenarios but also provides strong technical support for the coaching team to arrange post‐match training, which helps athletes to carry out targeted compensation training.