An efficient summarization algorithm based on clustering and bitstream extraction

Visualizing video is a very time consuming task. Video summaries are compact representations and very useful in video management systems. However, in some applications, in which the summary must be generated on demand, it usually implies a high delay for the user. Video summarization is a very demanding task in terms of processing resources, and it is very difficult to obtain a summary of a large video with low delay. In this paper a flexible summarization framework is presented which can create storyboards and video skims using bitstream extraction to generate the summary after an appropriate analysis algorithm. This technique is very efficient and achieves very low delay while keeping a reasonable quality of video summaries.