基于镜头内容的视频检索系统  

Video retrieval system based on shot contents

作　　者：姜静[1] 张栋梁[1] 

机构地区：[1]周口师范学院计算机科学与技术学院,河南周口466001

出　　处：《山东理工大学学报（自然科学版）》2013年第3期21-24,共4页Journal of Shandong University of Technology:Natural Science Edition

基　　金：河南省科技厅攻关项目(122400450356)

摘　　要：针对已有检索系统的不足,设计并实现了一种基于镜头内容的视频检索系统.该系统在镜头边缘检测关键技术上,采用了一种基于相邻尺度小波变换乘积的镜头检测方法;在关键帧提取关键技术上,采用了一种基于多特征自适应阈值检测的关键帧提取方法,实现了对镜头内容的多层次描述,并且所选取的关键帧能够很好地反映视频内容.最后将研究方法通过检索平台进行了验证.Considering the shortcomings in existing retrieval systems,a video retrieval system is designed and realized based on shot contents.The retrieval system use a lens edge detection method based on the wavelet transform multiplication of adjacent scales.Furthermore,it uses a key-frame extraction method based on adaptive threshold detection of multi-features,in order to achieve the description of a multi-level of the lens contents and the selected key frames adequately reflecting the video content.Finally,these methods are tested and verified through this system.

关 键 词：视频检索 镜头边缘检测 小波变换 关键帧提取 阈值检测 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]