The first EUROMEDIA workshop on Image Processing and Computer Aided Video Retrieval

The research project EUROMEDIA - funded by the European Union in the Esprit framework - is familiar to most FIAT/IFTA members, since it was presented during our conference last year in Paris. The Documentation Commission has selected this project to be one of its major fields of interest and will have a close look at the results and their possible implementation.

EUROMEDIA aims at setting up a rationalised workflow for handling digital video images on a tapeless basis in an archive environment, using the most modern computer tools available, and connecting the video streams to textual descriptions in the databases well known to every media archivist. The general goal of the project is to facilitate the work of the documentalists cataloguing the images and to enhance the accessibility for all users of our large collections of television archive material.

'Cut Detection', 'Key Frame Extraction', 'Motion Analysis', 'Similarity Retrieval' - those are only some of the buzzwords describing advanced research topics that are relevant for reaching the goals set for EUROMEDIA.

To get a clear and common understanding of the latest developments in technology and their possible application to the project, the EUROMEDIA consortium organised a 3-day workshop, hosted by ORF, in Vienna, 3-5 March 1997.

The workshop served as a perfect platform for an in-depth discussion of image processing methods and computer aided video-retrieval among archive professionals, technical developers and researchers. The total attendance was about 50 people, among them several FIAT/IFTA colleagues, the partners from the EUROMEDIA consortium and three external experts: Prof. Desai Narasimhalu, University of Singapore, Prof. Wolfgang Effelsberg, University of Mannheim, and his assistant Mr.Reiner Lienhart.

All speakers gave presentations of their most current research results and developments - making the very complex contents easily understandable by visualising most of their arguments with examples on screen. Just to mention some of the highlights presented:

• automated retrieval of patterns within a video stream - e.g. individual human faces, buildings etc. that the system has already been taught
• automated retrieval of captions (subtitles)
• automated genre recognition (e.g. commercials)
• automated recognition of violent scenes in a programme

During the final discussion on the third day the workshop was considered a very fruitful exchange of ideas by all sides: the researchers gained an insight into the desires of the users, the implementers were able to get ideas on both the economic and technical feasibility of their ideas and, lastly, the users got to see what can be done and how it could be applied to their problems.

As it is not possible to go into the details of the workshop's contents here; We will use a medium that is more suited for large amounts of information and put a summary on our web site. This will also contain some of the most relevant sources of information concerning this topic on the net. Updated information on the EUROMEDIA project as a whole will be given in a presentation during our conference in Budapest in September - so see you there !

Herbert Hayduck
ORF