Soccer Motion Track Recognition Based on Machine Vision and Image Processing

Abstract

Sports programs are widely popular in today's society because of their unique charm. As one of the most popular sports videos, the analysis and research of soccer videos are receiving more and more attention from researchers. Due to the diversity of conditions in soccer games such as venue and color of clothing, there is no universal tracker that can perfectly adapt to all scenarios. Machine vision is the main means of acquiring information and understanding the world using computers instead of human eyes and brain. In the case of video images, by acquiring information under time series and fully exploiting its internal features after certain image processing, it can be used not only to effectively identify target objects, but also to locate moving targets, predict possible future motion trajectories of targets, etc. In this paper, we propose the recognition of soccer ball motion trajectory based on machine vision and image processing. Based on the in-depth study of feature extraction method of soccer ball, soccer ball target recognition based on target shape analysis, etc., we propose a practical and effective analysis for Vi Be algorithm and edge detection of image under machine vision of ball in soccer video. Therefore, the algorithm studied in this paper can better accomplish the task of multi-target tracking in soccer game scenes and can adapt to different scenes with strong robustness to changes in conditions such as field and clothing color, which is a successful application of machine vision in soccer trajectory recognition processing.