Multimedia Video Information Retrieval Based on MapReduce under Cloud Computing

With the development of network and digital multimedia technology, in the Internet, the application of a variety of videos popularizes increasingly, and the number of online videos explodes rapidly. How to discover and retrieve the vast data of online videos effectively? It has been a problem in emergency to be studied and solved in the field of research and digital media sector. In this paper, from the selection and algorithm analysis of massive data-oriented video features, stream processing can be conducted in multimedia video data by Hadoop's MapReduce distributed computing framework, then users could assign implementation of tasks just by compiling Map function and Reduce function, and the underlying run-time system will conducts parallel computing by scheduling large-scale cluster automatically, so as to retrieve multimedia video information effectively.