An improved slow-motion detection approach for soccer video

In this paper, an improved slow-motion detection approach for soccer video is presented. Unlike the existing zero-crossing measure that only rely on the related feature of frame difference, we also introduce a series of characters related to dominant color in the interest of identifying slow-motion replays generated by standard or high-speed camera. The experimental results on six soccer video clips confirmed that the proposed framework was very effective and could successfully detect all the types of slow-motion replays. From our comparative analysis, we obtained a better result than the existing zero-crossing measure, especially reached high recall rate 100%.