Abstract
The incorporation of intelligent video processing algorithms into digital surveillance systems has been examined in this work. In particular, the use of the latest standard in multi-media feature extraction and matching is discussed. The use of such technology makes a system very different to current surveillance systems which store text-based meta-data. In our system, descriptions based upon shape and colour are extracted in real-time from two sequences of video recorded from a real-life scenario. The stored database of descriptions can then be searched using a query description constructed by the operator; this query is then compared with every description stored for the video sequence. We show examples of the fast and accurate search made possible with this latest technology for multimedia content description applied to a video surveillance database.