Design And Implementation Of Two-Level Image Retrieval System Based On Content

Image is one of the important ways to acquire,express and transmit information,and is the carrier of all visual information.With the increasing abundance of image resources,the orderly organization and effective utilization of images have become an urgent problem.In the shopping website,most of the merchandise is presented as an image except for the description of the text.The ability to retrieve images of goods that are completely consistent or highly relevant becomes the decisive factor for online shoppers to purchase the goods they need.Content-based image retrieval technology can make up for many shortcomings of text-based image retrieval technology and achieve more accurate retrieval.This paper applies content-based hierarchical retrieval technology and database technology to commodity image retrieval,and designs and implements a practical offline commodity image retrieval system.The main results are as follows:1.Based on image global features(color,shape or comprehensive features)and local features(Sift features),a hierarchical retrieval model was introduced to analyze the attributes and characteristics of commodity images,and a content-based two-layer commodity image retrieval system was designed.2.Apply the content based hierarchical retrieval technology and database technology in the field of commodity image retrieval,realizing a practical offline commodity image retrieval system.3.Using an image library to retrieve the experiment,through the index analysis and the results,it verified that the image retrieval system was able to improve the accuracy of the image retrieval,and to some extent,improved image retrieval,providing an effective solution to the image retrieval.