An ontology based approach for inferring multiple object events in surveillance domain

The ontology is an efficient tool that can bridge the semantic gap between the extracted information from the visual data and its interpretation in a given context. The ontology has been used in video surveillance applications to improve the accuracy of the indexing and retrieval system. However, these systems handle only one or two objects without considering events that involve multiple objects. In this paper, we propose to use OVIS (Ontology based Video surveillance Indexing and retrieval System) a system for indexing and retrieving videos in video surveillance application. We have applied OVIS to videos that contain multiple objects events (e.g. Group walking, Group splitting, Group formation, etc.).