A Framework for Video Modeling

In recent years, research in video databases has increased greatly, but relatively little work has been done in the area of semantic content-based retrieval. In this paper, we present a framework for video modelling with emphasis on semantic content of video data. The video data model presented distinguishes four layers: the raw data layer, the feature layer, the object layer and the event layer. It supports automatic definition of high-level concepts, such as video objects and events, based on extracted features. We focus our attention on event descriptions in this paper and give two modelling examples in the medical and soccer domain to show that the proposed event grammar can be efficiently used in different domains