| Segmentation and tracking are a hot problem in video processing. This paper presents a method to implement quick segmentation and tracking of players from a soccer video according to the characteristics of it. First, the RGB and HSI color models are selected to be complementary to each other, we can segment the moving object in soccer video based on a united color space. Then correlation of two normalized color histograms is calculated to identify which team a player belongs to. Finally, this method has used the context features and correlated-template matching method to track the players. Experiment results show that this method deals well with occlusion of two players belonging to different teams and can track them robustly. |
PDF Full Text Request