Modeling of video objects in a video databases

In this paper, we present an efficient video data model to represent moving trajectories of video objects and spatiotemporal relationships among the video objects. A video clip is segmented into a set of common appearance intervals (CAIs). A CAI is a time interval that video objects appear together. Transitions among CAIs record the appearance/disappearance of video objects. Depending on the properties of video objects, they are classified as foreground and background video objects. Foreground video objects are further divided into moving video objects and static video objects. Different models are designed to capture these video objects and spatio-temporal relationships among foreground video objects.