摘要
通过研究基于注释的视频检索和基于内容的视频检索的优缺点以及认知学关于人对视频相似性的理解特征,对新一代基于语义的视频检索进行了探索。以视频数据的一个大类——新闻视频为研究对象,采用层次处理思想和综合利用新闻视频各组成媒质的内容特征及其所表达的丰富语义,设计实现了一个基于语义的新闻视频检索原型系统。实验表明该系统自动化程度高、语义客观、检索灵活。
In this paper, a research about new theories of integrated semantic-based video retrieval is presented by exploring merits and defects of note -based or content-based video retrieval and apprehend characteristics about video comparability in cognitive science. A semantic-based video retrieval original system for video news, a main kind of video is designed and realized by applying an idea of multi-layer processing and integrating content features and rich semantic of news video's media components. The system is experimentally proved automatic high, semantic impersonal and retrieval flexible.
出处
《计算机工程》
CAS
CSCD
北大核心
2004年第16期155-157,共3页
Computer Engineering
关键词
新闻视频检索
视频分段
语义提取
多粒度反馈检索
News video retrieval
Video parsing
Semantic extraction
Multi-granularity teedback retrieval
