The Research Of Image Retrieval Base On Visual Word Tree

With the development of computer technology, technology, content-based imageretrieval technology has been widespread attention and application, and the effect of themore significant applications in the field of image retrieval. But from the point of viewof efficiency of retrieval, retrieve the increasing amount of data. The drawbacks oftraditional content-based image retrieval methods to form revealed. The inefficienciesresulting retrieve, accuracy is not higher. How to improve the efficiency of imageretrieval are those loans to solve the critical problem of content-based video retrieval.This article is against this background, the image retrieval method based on visual wordtree. The image feature into the visual vocabulary in accordance with the treerelationship stored in the retrieval process, in accordance with the tree traversal queryeffectively reduce the system resources occupied in the system in the retrieval process,and improve the overall efficiency of the image retrieval. Tangible improvements in theefficiency of image content-based video retrieval.In the process of this research, the paper first discussed the key technology forcontent-based image retrieval, and the basis of the technical discussion, focusing on thevisual vocabulary modeling and learning methods. Using SIFT feature extractionmethods of the image feature extraction, the features extracted from the visual word treeby k-means clustering method and SVM method of visual word similarity judgment,and achieve tangible image retrieval purposes. Finally, the use of the VC platform forthe realization of the functionality of the system, effectively verification In this paper,the design of the efficiency of the image retrieval system based on visual vocabulary.Reference and the reference value for improving the retrieval efficiency of the imageretrieval system played.