Video semantic concept detection using ontology

Abstract

Semantic concept detection in video is a challenge for video semantic content analysis. The performance of semantic concept detection methods depends on representing the video semantic content exactly. In this paper, perception concept and semantic concept are defined to abstract and model video semantic content. Furthermore, semantic concept detection using ontology is proposed, in which the context is modeled by ontology, and the semantic concepts are detected combining with both low-level features and context information. Finally, the linear fusion strategy is used to fuse the matching results and detect the semantic concepts. The proposed method is demonstrated in a news video domain and shows promising results.