基于规则分析的新闻视频口播帧检测

新闻视频的主体内容是一系列的新闻单元,准确地对新闻单元进行分割是新闻视频自动索引的重要依据,同时也是对新闻视频进行基于内容的视频分析的过程中所要解决的一个基本问题。由于新闻主持人口播镜头的开始通常是一个新闻单元的开始,所以检测新闻主持人口播帧是定位和分割新闻单元的有效途径。论文提出的基于规则分析的口播帧检测方法(DBRA,DetectionBasedonRuleAnalysis)是一种全新的口播帧检测方法,该方法通过分析新闻视频口播帧具有的一些规则(亦可称其为特点),由计算机系统完全独立地对新闻视频进行分析、并自动检测出口播帧。该检测方法与现有的检测方法相比,不仅使用较为方便,而且具有一定的智能性。

Detection Anchorperson Frame Based on Rule Analysis in News Video

In the news video,the main content is a series of news items.It is very important for content-based analysis of news video to exactly segment the news items.Because the beginning point of anchorperson shot is generally the starting point of one new item,detecting anchorperson frame is an effective approach to position every news item.In this paper,a completely new method to detect the anchorperson frame is proposed.By analyzing some rules(or characteris-tics)of the anchorperson frame ,computer system automatically detects the anchorperson.Comparing with other detection methods,the detection method based on rule analysis is not only very easy to use but also intelligent in certain degree.