A video summarization method based on key frames extracted by TMOF

In this paper, we propose a video summarization method based on the Temporally Maximum Occurrence Frame (TMOF). First, the key frames are extracted from the video and then they are clustered by calculating the distance between their feature vectors; the TMOF is constructed in the clustered collection. Finally, the video summarization is formed by the frames with the smallest distance from the TMOF. Taking a news video as example, the experiment result shows that the algorithm of video summarization meets the video semantic well.