We propose a new method for structuring and analyzing sports videothrough clustering of video shots based on low-level visual content. Thedominant scenes of a sports video are first automatically extracted byusing the color features derived from its video shots. The video shotsbelonging to each dominant scene are then processed to identifyinteresting play events in the sports video. To reduce the influence ofthreshold selection on the results, different forms of shot colorfeatures, including shot color histogram and its principal component andsubspace linear discriminant representations, are also examined.Experimental results on various kinds of sports videos, such as tennis,volleyball, basketball and football videos, are presented to demonstratethe effectiveness of the proposed method
展开▼