基于改进镜头边界检测算法的视频检索系统  

Video Retrieval System based on improved detection algorithm of shot border

作　　者：邹晓燕[1] 殷建[1] 

机构地区：[1]山东大学威海分校信息工程学院,威海264209

出　　处：《铁路计算机应用》2011年第6期8-10,14,共4页Railway Computer Application

摘　　要：随着多媒体信息技术的不断发展,视频信息大量涌现,如何在海量的数据中快速准确地检索出所需要的资料就显得至关重要。于是,基于内容的视频检索(CBVR)技术研究引起了人们广泛的关注。本文首先对视频的特征进行分析并提取出颜色和纹理直方图作为代表特征信息,然后使用一种改进的自适应滑窗的双阈值镜头检测算法对视频进行镜头分割。最后,通过实验对3种算法进行对比,结果表明,文中提出的算法在复杂度没有明显增加的情况下对镜头分割的准确度优于传统算法。文章的最后一部分使用本文提出的算法实现了一个视频检索系统,结果表明,本算法的准确度比传统算法有了一定的改进。With the rapid development of multimedia technical,there were more and more digital video resources on the Internet.How to retrive the information in enormous amout of data fast,effectively and accurately became the focus of attention.As a result,content-based video retrieval（CBVR） had aroused widespread concern.This paper first analyzed the characteristics of the video,extracted the color and texture histogram as the main feature information,and then used an improved adaptive sliding-window double-threshold shot detection algorithm for video shot segmentation.Finally,three kinds of algorithms were compared through experiments.The results showed that the algorithm proposed in this paper was superior to the traditional in accuracy when the complexity had no significant increase.The last part of this article used the algorithm proposed in this paper implemented a retrieval system,and the results showed that the accuracy of this algorithm was better than the traditional algorithm to some extent.

关 键 词：镜头分割 视频检索 滑动窗口 双阈值 

分 类 号：U285.17[交通运输工程—交通信息工程及控制] TP391[交通运输工程—道路与铁道工程]