文章基本信息

标题：An Unsupervised Method for Real Time Video Shot Segmentation
本地全文：下载
作者：Hrishikesh Bhaumik ; Siddhartha Bhattacharyya ; SusantaChakraborty 等
期刊名称：Computer Science & Information Technology
电子版ISSN：2231-5403
出版年度：2014
卷号：4
期号：5
页码：307-318
DOI：10.5121/csit.2014.4531
出版社：Academy & Industry Research Collaboration Center (AIRCC)
摘要：Segmentation of a video into its constituent shots is a fundamental task for indexing andanalysis in content based video retrieval systems. In this paper, a novel approach is presentedfor accurately detecting the shot boundaries in real time video streams, without any a prioriknowledge about the content or type of the video. The edges of objects in a video frame aredetected using a spatio-temporal fuzzy hostility index. These edges are treated as features of theframe. The correlation between the features is computed for successive incoming frames of thevideo. The mean and standard deviation of the correlation values obtained are updated as newvideo frames are streamed in. This is done to dynamically set the threshold value using thethree-sigma rule for detecting the shot boundary (abrupt transition). A look back mechanismforms an important part of the proposed algorithm to detect any missed hard cuts, especiallyduring the start of the video. The proposed method is shown to be applicable for online videoanalysis and summarization systems. In an experimental evaluation on a heterogeneous test set,consisting of videos from sports, movie songs and music albums, the proposed method achieves99.24% recall and 99.35% precision on the average.
关键词：Real time video segmentation; spatio-temporal fuzzy hostility index; image correlation; threesigma;rule