视频镜头边界检测中奇异值分解算法

Algorithm of Shot Boundary Detection Based on Singular Value Decomposition

在海量的视频数据中准确、快速地检测出镜头边界非常重要。对这一问题,基于现有的视频镜头边界检测方法的基础上,提出基于奇异值分解的视频镜头边界检测方法。实验结果显示,该算法是有效的,其不但能准确地检测出突变边界而且对渐变边界检测也有较好的效果,并且在新闻视频中可有效识别闪光。

In the vastness of the video data retrieval,accurate and rapid detection of a shot boundary is very important. This article proposes a new framework by using singular value decomposition and existing methods of video shot boundary detection in video sequences. The experimental results show that this algorithm and framework are efficient because not only sharp cuts are detected precisely and transition cuts are detected efficiently ,but also the flash frame can be singled out.