一种基于镜头的视频场景检测方法

在数字视频的分析、浏览、检索中,现有的以镜头为基础的方法由于镜头粒度信息太小而不能表达视频语义上的联系,因此有必要将视频内容按照高层语义单元——场景进行组织。从分析视频剪辑的基本原则入手,给出了一种视频场景构造方法;首先使用改进了的像素匹配二次差分法结合双阈值法进行镜头突变和渐变的检测,然后对镜头内的帧间距离进行判断来提取关键帧;提出了一种基于双滑动镜头窗口的聚类方法,将内容相似语义相同的镜头聚合在一起形成新的场景。试验表明,该方法是有效的。

A Method of Video Scene Detection Based on Shots

In digital video analysis,browse and retrieval,existing method based on shots can not convey meaningful semantics because shots are too numerous to handle,so it's necessary to organize video content with the scene structure,which is a more meaningful and high-level semantic video unit.This article begins with analyzing the film grammar and then introduces a method of video scene construction.The pixel-matching two-difference algorithm and two histogram comparison are used for shot detection,a method of extraction shot key frame based on distance of adjacent frame is presented.And then a cluster scene based on double slide shot window that has similar content and semantic is formed.The experiment result shows that this method is efficient.