Research on Sports Video Tracking Based on Clustering Extraction Algorithm

Introducing digital video technology into physical training to improve the scientificity and efficiency of physical training is a new research content. In actual tracking, the target is moving and sometimes interfered by the background, so the accuracy of tracking the target is greatly reduced. With the improvement of sports competition level, the past training mode that only relied on coaches' intuition has been powerless to improve the competition level. Because machine vision has better accuracy and memory than human eyes, it can quickly capture moving targets and record various motion data of targets. In order to accomplish this task better, it is necessary to use existing technologies and methods to separate the target from the background, and then extract the target name, so as to achieve efficient tracking and capture of moving targets. In this paper, the improved background update difference method is used to detect the moving target, and then the clustering extraction algorithm is used to track the moving target, so as to realize the tracking analysis of sports video.