Abstract:

Video Structured Description (VSD) is a technique that extracting and refining information from video data, to structure it to be a text information which can be read by both human and machine according to information’s intrinsic semantic relations. To distill knowledge from video, typical methods could be employed, such as spatialtemporal segmentation, feature extraction, object recognition and so on. This paper introduces a VSDbased indoor scene description system which implements scene description, description storage and dispatching. Experiments show the VSDbased system is efficient for video content application.

Key words: VSD(Video Structured Description), video retrieval, video encoding, media streaming, XML-based DOM, Windows media