Video event mining and content management system using shot ontology description

Since the mass growing amount of sports video has been produced, how to analysis and to make event mining in video content management issues are become more and more important. In this paper, we developed a shot ontology description based for the basketball video. Shot ontology is inferred by shot manipulations those included: shot detection, shot type classification, score board detection and motion statistics. This video content management system provided event feature manipulations at multiple levels: signal, structural, or semantic in order to meet user preferences while striking the overall utility of the video. The experiment results showed that our proposed methodologies could correctly detect interested events, long shots, and close-up shots and also achieved the purpose of video indexing and weaving for what user preferences.