RVARM Algorithm for Classifying Event Based Video Sequences

Abstract

Multimedia based systems have gained importance with the advent of broadband networks and compression of videos. Visuals are an important element in multimedia based computing and communication environments and applications from broadcasting, education and military use multimedia information. Media intensive information has resulted in the need to manage streaming audio, video and digital TV contents. Visual media requires processing efficiency during video retrievals where indexing and retrieval by themselves have a wide spectrum of promising applications. Users are interested in querying videos at a very top level based on descriptions and contrary to this, video content storage and retrieval uses low level features. The low level features are not very user friendly and presents many challenges in the management of video based collections. A rapid video mining algorithm for association rules based on indices of descriptions is proposed in this paper. The annotated videos descriptions are transformed into a relational dataset and mined for associations.