Campus pushes out the boundaries of video

A busy university professor is rushing across campus to his next lecture. Suddenly his personal digital assistant (PDA) beeps, a warning that a message has arrived.

Flipping open the screen he sees an e-mail warning that something he has been waiting for is running on television. He messages back, requesting that it be patched through to his PDA. Within seconds the familiar signature tune for Bob the Builder comes through, giving him a chance to see at least the start of the show before going into class.

The video age has arrived, at least on the campus of Dublin City University (DCU). The university's Centre for Digital Video Processing now has 1,000 users across campus attached to a video network. They can prerecord and play back television programmes on their personal computers, ask for warnings of when shows are on or have been played, and can see freeze-frame overviews of any show, allowing them to choose particular scenes or get a quick look at the end only.

The service is called Fischlar and while it makes viewing very convenient for its users, it has a much more serious intent, explained Prof Alan Smeaton, head of the School of Computer Applications at DCU. Fischlar serves as a permanent test bed for research into video signal processing and transmission.

New software and hardware systems are under development all of the time, he said, and these emerge as new features available to users. It also provides an insight into how people interact with the still evolving digital video services.

The centre is run jointly by Prof Smeaton's department and the School of Electronic Engineering. This illustrates the dual nature of the software/hardware combinations that drive existing information technology systems and the much newer video processing systems.

The centre was established in 1997 and "just sort of happened", Prof Smeaton explained. He provided expertise in information management and retrieval, while electronic engineering was doing a great deal of work on the digital video processing standard, MPG.

Both groups came together to see how compressed video images might be captured, processed and redistributed over a network in much the same way as text documents are over computer networks and the Web.

This by necessity involved research in several areas and, rather than just publish findings, the team decided to test their discoveries on a live, working video-distribution network. The result was Fischlar, a service that is positioned on the cutting edge of video-processing technology, a place that Prof Smeaton described as "scary".

"The goal isn't to develop a product from this," he said, although clearly if such a system were brought to market there would be takers for what it has to offer. In fact, members of the team are now behind a campus company that will move into DCU accommodation this month. Its goal is to decide what elements of the service might be commercial.

Fischlar offers a real world "demonstration system" where new ideas, software and artificial computer intelligence can be applied to video cataloging and processing.

The heart of the service is a £330,000 (€420,000) digital video streaming server, acquired with assistance from the Higher Education Authority and Sun Microsystems. "It does nothing else other than pumping out video signals," Prof Smeaton said, distributing them over an ethernet backbone in much the same way as digital text moves across an e-mail server.

Users only need their PCs and a video stream player, which comes as a piece of plug-in software. PDAs can also access the system over a wireless local area network, which is part of the Fischlar service.

The system as it now exists allows users to record programmes broadcast by any of the existing eight terrestrial stations. These can be played back at any time.

DCU has been able to team up with University College Dublin's "MyTV" service to support this feature. This profiles a user's preferences and sends out detailed notification for anything it thinks the user might like.

These features are impressive to see but are only the beginning, Prof Smeaton said. New computer-hungry services are on the way and will be delivered in a year or two.

One involves software driven "summarisation" of a programme that can automatically reduce a 90-minute football game to an action-packed five minutes. In the same way, it could reduce a 30-minute soap opera to just a few minutes of slanging matches and aggro, the parts many viewers most appreciate.

This automated reduction will be controlled by computer, Prof Smeaton said. Software will note changes in crowd noise or repeated use of replay or slow motion, preferentially editing out and patching together only these key segments for the viewer.

Equally a soap could be reduced by having the computer watch for key words in the dialogue or a change in speaking voice. The key word analysis is based on utilising the existing closed caption and subtitling data already being provided by many television broadcasters.

The goal, Prof Smeaton said, is to provide this level of control and video handling for each individual user. "The real world application tells us what facilities are worth developing and which aren't useful," he said. The Fischlar test bed provides an immediate consumer response.

Very powerful services and video-editing systems would emerge over time, he said. Some would allow video data on, say, a person or subject to be taken from an archive and streamed together into a single "programme".

And some of these services might have commercial applications to be exploited by the campus company, Aliope Ltd.