Design And Implementation Of Distributed Retrieval System Based On Hadoop For Massive Power Video

With the rapid development of video acquisition and video storage technology,and the rapid development and gradually perfection of smart grids,massive power video is being produced all the time,the power industries are also putting forward higher requirements for video processing.How to store and retrieve these video data effectively becomes the biggest problem we face.Cloud computing platform Hadoop,which has great advantage of mass data storage and computing,high reliability and high expansibility,provides a new research idea to solve the problem of massive power video retrieval.This paper mainly introduces the design and implementation of distributed retrieval system based on Hadoop for massive power video,includes:(1)After analyzing the characteristics of power video,combining with the distributed storage and computing characteristics of Hadoop large data platform,a new retrieval framework based on Hadoop is designed;(2)Using the histogram method to extract the video key frame,proposing video key frame distributed extraction algorithm based on MapReduce,combined with MapReduce computing framework;(3)Proposing a similar image retrieval method based on Lire and Hadoop,combined on Lire image retrieval and Hadoop large data platform;(4)Designs and implements Web-based power video large data distributed retrieval system.The system presents the retrieval results to users through the Web retrieval interface,constitute a three-tier structure CBIR Web application.The system introduced in this paper,extracts the key frames and builds the indexes from the video data.As long as users upload the picture in the Web interface,the server will retrieve image-related video data based on the pictures uploaded by users.This system can store and operate massive power video,realize quick retrieval.