AN MPEG-7 BASED VIDEO SEMANTIC RETRIEVAL SYSTEM

Zheng Quan,Li Shaobo,Sun Zhijun,Yuan Jing,Wang Song
DOI: https://doi.org/10.3969/j.issn.1000-386X.2012.05.017
2012-01-01
Abstract:A distributed video retrieval system that supports semantics,called XuNet,is designed and implemented.It uses an improved video semantic processing tool,which is based on IBM VideoAnnEx annotation tool and equipped with shot semantic graph annotation and natural language processing functions to semantically parse and annotate videos and generate MPEG-7 description files that contain semantic information.Afterwards XuNet builds distributed indices for videos'MPEG-7 files,at the meantime distributedly storing video files.The system provides abundant web query interfaces,including Key words semantic expansion query,semantic graph query and natural language query.As soon as a user submits his semantic query intention,XuNet quickly retrieves videos or segments according to his interests or even allows him to browse or play videos on demand.Owing to the distributed architecture,the whole system is highly scalable and supports mass video information indexing and retrieval.
What problem does this paper attempt to address?