Abstract:

Slow motion is one of the most important semantic contents in sports video. This paper studied the logo transition based slow motion replay detection of tennis video. Firstly, the candidate logo shots from the video streaming are detected. Secondly, the logo template is generated. Thirdly, the similarity matching between video frames and logo template in the candidate logo shots is done, and, according to the matching results, the size of the adaptive threshold is determined and the boundaries of slow motion is located roughly. Finally, the brightness changes during the logo transition process are used to accurately locate the boundaries of slowmotion. Experiments results show that，unlike previous approaches, this method has great improvement in both precision rate and positioning accuracy of the boundaries,and can be used for other sports videos which also have logo transition.

Key words: tennis video；slow motion；logo；adaptive threshold