Object-orientation and description logics techniques for implementing the indexing of surveillance video content

Retrieving video based on content is a challenging task that has became a central focus for many professional groups including knowledge management, database and image processing communities. Benefiting from automatic feature extraction and labeling, current systems are still inherently limited as they only operate at the simple features level. This paper presents a model for content and event-based indexing and retrieval of surveillance video captured by a camera in a car park environment. The paper focuses chiefly on modeling the semantic video knowledge in description logics and its implementation in an object-oriented database environment using Objectivity/DB.