Skip to Main content Skip to Navigation

La structuration sémantique des contenus des documents audiovisuels selon les points de vue de la production

Thi Minh Phung Bui 1
1 ASIM - Architecture des Systèmes intégrés et Micro électronique
LIP6 - Laboratoire d'Informatique de Paris 6
Abstract : Video content based indexing is highly related in one hand to the progress of new technologies, and image, audio and multimedia processes. In another hand, it relies on the power and skill of video and multimedia representation standards and domain knowledge organisation. Presently, video content representation is expecting the extension of standards for representing media contents in order to capture the semantics of media in deeper structures and provide formal and rich semantic description. Our work proposes a flexible and high level semantic video content description in the frame of the Production Shot Description Schema (PSDS) from the semiotic approach and the points of view of audiovisual production professionals. We analyse the video content into three semantic layers – sensory, narrative and semiotic – and each level provides its particular ontological description. The semiotic completes the technical and narrative dimensions in explaining the way the dynamic structures of the filmic text manages to produce theses two first semantic interpretations in terms of structural organisation. Semiotic approach facilitates to link the representation of the sensory level with the significance of media structures from one's semiotic points of view and audiovisual professionals classification reasoning. Using both media content automatic detection/recognition process and models, theories, discourses, and terminology of the cinema world, this approach can provide a natural and simple translation between the different semantic levels. We use an object-oriented, multi-points of view, video content description which implies to build the ontological tree of the described domain by selecting significant and essential information and their features. Represented as concepts, they constitute a semantic network where users can browse and retrieve information related with the described contents. Each concept is a node of the concerned domain semantic network. The knowledge associated in the PSDS is then structured by Mpeg-7 and XML Schema formalisms and can be used as the basis to build interactive environments for images edition. Video is an information stream in which data can be marked, annotated, analysed, and edited. Our description metadata format involves information about different steps of the production ((preproduction, production et post-production) in order to facilitate management and handling of video objects as well as representation of their significance. They can be reused in many access frameworks such as content based indexation, retrieval, filtering, analysis and movie apprehension.
Document type :
Complete list of metadatas

Cited literature [141 references]  Display  Hide  Download
Contributor : Minh Phung Bui Thi <>
Submitted on : Wednesday, February 27, 2008 - 12:34:25 PM
Last modification on : Friday, January 8, 2021 - 5:32:05 PM
Long-term archiving on: : Thursday, May 20, 2010 - 11:39:32 PM


  • HAL Id : tel-00259261, version 1


Thi Minh Phung Bui. La structuration sémantique des contenus des documents audiovisuels selon les points de vue de la production. Informatique [cs]. Université Paris VIII Vincennes-Saint Denis, 2003. Français. ⟨tel-00259261⟩



Record views


Files downloads