Video summarization by Contourlet Transform and structural similarity

Abstract

Video summarization is the main aspect in video content management system, by which users can easily search the video content for a particular data or scene. Video summarization is the process of selecting a set of significant frames called key frames to represent original video in the form of a short video clip. In this work, individual frames of the video represented using Contourlet Transform are analyzed structurally to detect the scene changes, which will result in clustering of frames in the video. Finally Renyi Entropy can be used to extract most relevant frames from clusters to construct full motion summarized video.