Abstract: A hierarchical semantic associative video model was proposed which described video information in hierarchical concept tree,scene network and semantic object net. The concept tree could be constructed under different applications. Time information of video was represented by scene network. The contents of video shots were represented by semantic objects and their relations, and the semantic correlations between shots were represented by semantic relations among objects which were in different shots.Video can be retrieved by time information and other semantic information,and the results are convergented.

Key words: hierarchical semantic associative model, semantic network, video retrieval, content-based retrieval

摘要： 提出一种视频的分层语义联想模型,构造三个层次的信息:概念层次树,场景网络和语义对象网络。利用概念层次树来适应不同的应用环境,场景网络表示视频的时间信息,而语义对象及其关系用来表示视频镜头的内容,通过分属不同镜头的语义对象的关系来表示镜头间的语义相关度。该模型采用基于时间和语义关系的检索方法,搜索结果是收敛的。

关键词: 分层语义联想模型, 语义网络, 视频检索, 基于内容检索