Research Of Massive Image Retrieval Based On MapReduce

With information technology developing rapidly,variety and quantity of image datais increasing fast.How to retrieve desired images among massive imagesstorage isgetting to be an urgent problem.As the rising of cloud computing,Many new programming models genermed,thatmakes handling large scale data sets easier.MapReduce,a programming modeldeveloped by Google,provides a scalable,flexible and QoS guaranteed IT infrastructureparticularly for embarrassingly parallel data oriented computational tasks．Users specifythe computation in terms of a map and a reduce function,and the underlying runtimesystem automatically parallelizes the computation across large-scale clusters ofmachines,handles machine failures,and schedules inter-machine communication tomake efficient use of the network and disks.In this paper,we established a Distributed Image Retrieval System,in which imagesare retrieved in a content based way,and the retrieval among massive image data storageis speeded up by utilizing MapReduce distributed computing model. Moreover,faulttolerance,ability to run in a heterogeneous environment and scalability are supported inour system.Experiments are carried out to verify the improvement of performance whenMapReduce model is utilized.Results have shown that image storage and imageretrieval based on MapReduce outperform that in centralized way greatly when totalnumber of images is large.