Recently, there has been research to utilize spatial metadata of videos for the management of large numbers of videos. For example, the location of a camera from GPS and the viewing direction from an embedded digital compass are captured at the recording time so that they form the field of view (FOV) model for the coverage of a viewable scene. These spatial metadata effectively convert the challenging video search problem to well known spatial database problems, which can greatly enhance the utility of collected videos in many emerging applications. However, there are still many open fundamental research questions in utilizing spatial metadata for data collection, indexing, searching, etc., especially in the presence of a large video dataset, as well as more sophisticated questions such as whether a video database search can accommodate human friendly views.
展开▼
