Realzeit-Szeneninterpretation mit ontologiebasierten Regeln

This thesis is concerned with the subject of “high-level scene interpretation”, which can be roughly defined as understanding images or video streams at abstraction levels above single objects. A generic framework for knowledge-based scene interpretation is presented, realised with an interpretation system called SCENIOR (SCENe Interpretation with Ontology-based Rules). Activity models are defined in an ontology using OWL (Web Ontology Language), extended by SWRL rules for describing constraints. This conceptual knowledge base is transformed into a high-level scene interpretation system based on Jess rules. Interpretation goals are transformed into hierarchical hypotheses structures associated with constraints. The incremental interpretation process is organised as a Beam Search with multiple parallel interpretation threads. At each step, a context-