Design And Realization Of Content-based Image Retrieval System

With the rapid development of multimedia technology,the number of digital images in the network has increased rapidly.How to quickly retrieve the target image in massive digital images has become an urgent problem to be solved in the academic field.Due to the limitation of traditional text based image retrieval,content-based image retrieval has become a hot research topic in the present research.Content based image retrieval gets rid of the limitations of manual understanding,because it mainly identifies the image from the inherent characteristics.At present,the main research of content based image retrieval is focused on feature extraction,feature matching,image retrieval.In this paper,we design and implement a distributed image retrieval system based on content,combining the image feature extraction,image retrieval,image matching technology,and distributed computing technology.Fast and efficient image retrieval is realized in this system.The specific work of this paper is as follows:Based on the research background and the current status of content-based image retrieval technology,we deeply analysis the relevant key technology of content-based image retrieval,including image feature extraction,image index,image similarity measure and distributed technology.I have also made a comparative analysis of the existing technologies?We design and implement a content based image retrieval system,using the SIFT feature descriptor for the feature extraction of the image then we can get and then process the feature vector.Due to the high dimension of the feature vector,we chose locality sensitive hashing(LSH)algorithm which use for the feature vector of high dimension for feature vectors of image to build index,effectively solving the retrieval problem of high dimensional vector.Moreover,we introduce the specific research and implementation of the related algorithm.This paper's image retrieval system gets rid of the traditional pattern limits of centralized computation and centralized storage,using distributed computing and distributed storage way to solve the problem of computing and storage for large-scale images data.The image feature extraction and indexing based on Spark platform are realized,improving the efficiency of image retrieval.