Research And Inclement At Ion Of An Index System Of Massive Image Data Based On Cloud Computing Platform

With the rapid development of multimedia technology, the number of digitalimages all over the world is increasing at an astonishing rate recently. How to utilize theuseful information that contained in these massive images effectively is critical in imageretrieval. At present, the content-based image retrieval (CBIR) systems use featureextraction technologies to transform an image in the database into a point inside thefeature space, and the image features are very large-scaled for the massive imagecollections. Unfortunately, the current indexing mechanisms are not so efficient, whichposes a great emphasis on designing and implementing an appropriate indexingmechanism to assist and accelerate the retrieval process.In order to solve the mentioned problems above, this paper has proposed anindexing system of massive images based on cloud computing platform. The systemimplements the feature extraction, image storage and feature indexing in the cloudcomputing platform. Two kinds of clustering algorithms are used to cluster the differentfeature vectors and build index tables for feature indexing. Besides, our system alsosupports different images queries, such as union, intersection and subtraction operationsamong images, users can adjust many metrics such as color, texture and shape for asingle image. Some practical experiments have proven the effectiveness and efficiencyof the proposed system.