Supervised and unsupervised classification post-processing for visual video summaries

Representation of the video content using a set of key frames is one of the most common techniques for video summarization. Summaries composed of key frames allow users to grasp the overall content of a video, and access specific sequences. The post-processing algorithm presented in this paper makes it possible to create visual video summaries that are exhaustive, but not redundant. In three steps the method removes meaningless key frames, groups the key frames into clusters to allow multi-level summary presentation, and determines the default summary level. The algorithm utilizes both supervised and unsupervised classification strategies to perform these tasks. It requires no previous knowledge about the video contents, nor is any assumption made about the input data.