Moving Object Tracking Technology In Sports Video

In recent years, With the development of science and technology, Sports science playes an important part in nationwide body - building plan and sports training. Introducing the digital video technology into sports training to improve the efficiency and science is a new research subject.with the improvement of physical exercise or sports, in the past training manner which is used in competitive sports by coach's intuition is unable to improve athlete's sports standards. The development of computer vision provides a new tool on sports training, because computer vision is more precise anJ more memory abilily than human eyes, it can quickly capture moving object, and then register object's motion data, and these motion data provide a visual representation of athlete's action.This paper researches on precision and efficient and algorithm's degree of complication about object tracking and acquire some achievements. The main contributions are as follows:1. Target's feature modelTarget's feature model is the base of object tracking algorithm, this paper improve target's feature model. Use target's color histogram and target's spatial information as target's feature model, and avoid losing target in tracking procedure.2. Hybrid object tracking algorithmThis paper brings forward a new hybrid object tracking algorithm, we introduce four kinds of motion model in particle filter, according to sports video's feature. And then combine mean-shift algorithm to improve tracking algorithm's precision and robust.3. A new multi-object tracking method in sports videoMulti-object tracking is usually used in static background, and sports video is always dynamic background, so we bring a new method to track spotters. In this paper, we use k-mean cluster algorithm to segment playground's color, and then use global motion estimation and video to playground map algorithm to acquire the...