A rapid increase in the volume of digital video can be seen. This creates a strong demand for facilities to search and retrieve video footage from archived material. Current research activities in the area of content based representation and retrieval of video are strongly expanding and mainly application driven. Representations used in different approaches arc often based on very similar models. hut differentiated through specific semantics ot their vocabulary or used data models.
The proposed research project will develop a classification scheme and describe Video Object Foundation Classes (VOFCs) for the semantic modelling of moving image sequences. VOFCs are part of a framework for the development and comparison ot content based representations of video sequences. With the help of VOFCs the syntax and semantics of existing annotation data can be described. and in consequence the interchange of annotation information between different application domains can he realised. VOFCs will be described by formal description methods and modelled with data structures using object oriented design methodologies. The work will be carried out in an interdisciplinary programme (linguistics, film and media studies, applied computer science ) and will be directly related to industry (Sentient Machine Research. several television archives) and ESPRIT and ACTS projects.