自适应阈值新闻视频镜头分割算法  

News video shot boundary detection by automated threshold

作　　者：王国营[1] 寇红召[1] 李涛[1] 

机构地区：[1]解放军信息工程大学电子技术学院,河南郑州450004

出　　处：《计算机工程与设计》2011年第5期1734-1737,共4页Computer Engineering and Design

摘　　要：针对新闻视频镜头的转换基本上都是切变的特点,提出了一种自适应阈值新闻视频镜头分割算法。通过计算视频中相邻两帧的颜色直方图并进行比较得到帧间差,用帧间差与自适应阈值比较来判定镜头是否发生切换。该算法的最大优势是能够根据镜头内容的复杂程度自动确定阈值,在一定程度上避免了固定阈值算法适应性不强的问题,同时算法还考虑了如何消除新闻视频中常见的闪光灯对镜头检测的影响。实验结果表明,该算法对新闻视频镜头分割具有较好的效果。A new algorithm to news video shot boundary detection is presented according to the peculiarity of news video shots. Firstly, the gray-scale image histogram of every frame is calculated. By computing neighbor frame difference of histogram and comparing with the threshold, the algorithm can effectively detect the video shot boundary. The threshold is computed dynamically by the content of video shots, so the algorithm has a higher adaptability then those with fixed threshold. The influence of lamp light is also take into account in the algorithm, the experiment shows that the algorithm can get better result.

关 键 词：新闻视频 视频镜头 分割算法 直方图 自适应阈值 闪光检测 

分 类 号：TP311.134[自动化与计算机技术—计算机软件与理论]