基于MPEG-7的视频语义描述方法  

A Representation Scheme of Video Semantics Based on MPEG-7

作　　者：朱华宇[1] 孙正兴[1] 王箭[1] 张福炎[1] 

机构地区：[1]南京大学计算机软件新技术国家重点实验室,南京大学多媒体技术研究所,南京210093

出　　处：《南京大学学报（自然科学版）》2002年第1期74-82,共9页Journal of Nanjing University（Natural Science）

基　　金：国家自然科学基金 (6 990 30 0 6 );教育部高等学校骨干教师资助计划 [教技司 (2 0 0 0 ) 6 5号 ];中国博士后科学基金 (中博基 [1997]11号 )

摘　　要：基于对视频语义信息的 3个层次划分 ,提出了一个基于MPEG 7的视频数据模型 ,并运用扩展标记语言 (XML) ,以实例阐述了视频内容的视频对象、视频事件和视频元数据构造和描述方法 .所提出的方法能支持不同抽象层次上复杂语义关系的描述 ,能够使用户更加灵活地访问数字视频库 。With the rapid development of digital information, especially video, Content-Based Visual Query (CBVQ) has emerged as a challenging research area in the past years, which allows users to search video based on a rich set of visual features and spatio-temporal relationships. However, users often find interested video clips based on the semantic information they convey. Therefore, semantics description plays an important role in video modeling. In this paper, we propose a semantic video model based upon MPEG-7 to represent complicated semantic relations in the video consistently at different levels of abstraction. To ensure maximum interoperability and flexibility, we use the extensible markup language (XML) to illustrate and exemplify the representation scheme. The representation scheme is composed of video object DS, video event DS and video metadata DS. Under the proposed scheme, a video is viewed as a set of objects, relations among objects and relevant events occurring among these objects that can be organized hierarchically. Object can have multiple features, which are categorized to physical and semantic features. Temporal, spatial and semantic relation types are used to describe the relations among objects and nested relationships support to represent complex relationships efficiently. Hierarchical abstraction of video events is the most natural way of defining composite events. The description scheme is so generic in the sense that it does not target any specific application, but users are able to acquire specialized content and functionality by extending it for their own applications.

关 键 词：多媒体内容描述接口 MEPG-7 视频语义描述 扩展标记语言 XML 数据模型 视觉信息查询 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]