%0 Journal Article
%T Statistical Skimming of Feature Films
%A Sergio Benini
%A Pierangelo Migliorati
%A Riccardo Leonardi
%J International Journal of Digital Multimedia Broadcasting
%D 2010
%I Hindawi Publishing Corporation
%R 10.1155/2010/709161
%X We present a statistical framework based on Hidden Markov Models (HMMs) for skimming feature films. A chain of HMMs is used to model subsequent story units: HMM states represent different visual-concepts, transitions model the temporal dependencies in each story unit, and stochastic observations are given by single shots. The skim is generated as an observation sequence, where, in order to privilege more informative segments for entering the skim, shots are assigned higher probability of observation if endowed with salient features related to specific film genres. The effectiveness of the method is demonstrated by skimming the first thirty minutes of a wide set of action and dramatic movies, in order to create previews for users useful for assessing whether they would like to see that movie or not, but without revealing the movie central part and plot details. Results are evaluated and compared through extensive user tests in terms of metrics that estimate the content representational value of the obtained video skims and their utility for assessing the user's interest in the observed movie. ¡°I took a speed reading course and read ¡°War and Peace¡± in 20 minutes. It involves Russia.¡± Woody Allen. 1. Introduction In the last years, with the proliferation of digital TV broadcasting, dedicated internet websites, and private recording of home video, a large amount of video information has been made available to end-users. Nevertheless, this massive proliferation in the availability of digital video has not been accompanied by a parallel increase in its accessibility. In this scenario, video summarization techniques may represent a key component of a practical video-content management system. By watching a condensed video, a viewer may be able to assess the relevance of a programme before committing time, thus facilitating typical tasks such as browsing, organizing, and searching video-content. For unscripted-content videos such as sports and home-videos, where the events happen spontaneously and not according to a given script, previous work on video summarisation mainly focused on the extraction of highlights. Regarding scripted-content videos¡ªthose videos which are produced according to a script, such as feature films (e.g., Hollywood movies), news and cartoons¡ªtwo types of video abstracts have been investigated so far, namely, video static summarization and video skimming. The first one is a process that selects a set of salient key-frames to represent content in a compact form and present it to the user as a static programme preview. Video skimming
%U http://www.hindawi.com/journals/ijdmb/2010/709161/