The Design And Implementation Of Image Retrieval System Based On Hadoop Technology

As computer technology and information multimedea technology’s fast developing, image information now is a popular nomal data storage form increasing astonishingly, how to find out more valuable, more meanings infos from such huge of image infos now is main appeal in image infos using.This article text analyzes image infos technology developements and results,researchs and discusses image retrieval technology based on text, image retrieval technology based on content retrieval technology (CBIR) and based on high level semantic. This subject analyzes and researchs image retrieval’s algorithm theory,focus on researching that large scale data image retrieval system’s priciple design and function implementation, designs massive image retrieval system structure based on Hadoop technology, and describes system implementation principle and content. The image as an important index of user information, combined with the IT system, the image information to the user seamless integration of existing business data, through this information management way to fully support business work.System utlizes Mapreduce computational framework to achieve large data image retrieval’s function meeds, proposes retrieval type that specific application scenarios and algorithm(algprithm module) establishs correlation relationships, and adapt to all kinds of fields consumers needs; builds uniform algorithm interface specifications, reduce coupling degree, improves scalability, conveniently put new algorithm into computational system frameworks, introduces "relative feedback" method, experience data feedbacks through accumulations, reversely adjust retrieval parameters, improving the whole systems retrieval accuracy.In the subject the writer’s main works are as follows:1, Theoretical results. Analyzing and studying the main theory and method of image retrieval system, the implementation and application of image retrieval system in China and abroad are studied. The combination of large data processing technology and traditional system is studied. Based on the Hadoop data processing capability and MapReduce model, the frame model of image retrieval system is realized.2, System design and implementation. On the basis of theoretical research, design and implementation of image retrieval system based on Hadoop framework is verified, based on the above theoretical study, realize image retrieval technology feasibility, put forward the business work process of this system based on Image Retrieval3, Running system design experiment case, collect the experimental data, analysis of experimental data to draw a conclusion, and on this basis to verify the feasibility of the system.Image infos retrieval is a very important and meaningfull research field in scientific research and production and life, the subject find a specific practical andfeasible image infos retrieval method and and direction through principle research and system implementation,satisfys image retrieval appeals under certain data scale.