Semantic characterization, indexing and universal customized access to audiovisual content under controlled environments TIN2004-07860-C02
Jesús Bescós, Fernando Jaureguizar · 2007
The main objective of this project is to research in the design of algorithms and the development of tools both oriented to automatic feature extraction in image sequences of controlled environments, additionally the further indexing, storage and customized access to the previously extracted and properly coded information out of those sequences. The project consists of to main stages: Information acquisition and Management of that information to offer services to the final user. The information acquisition out of images from several cameras performs the automatic extraction of features of people present in those images, segmenting in real time the individual persons out of the background, while building a 3D model of the scene thanks to self-calibration techniques. Finally, this information is described according to the MPEG-7 standard to allow for efficient access and storage. The second stage is mainly concerned with the management of the obtained information for its use in the development of applications. Its design is based on the user and terminal/network profiles (MPEG-21 DIA) and on the management of alert levels/options for the delivery of information relevant to each user. According to these profiles, the available coders with the proper parameters are run for the purpose of adaptation and customization of the information during each session.