Image sorting method and device

Abstract

The present invention relates to a picture sorting method and device. The method includes: performing picture search by using initial keywords, calculating an anchor text concept set of the search results according to the search results of the initial keywords, and calculating the set of anchor text concepts The weight of the degree of association between the anchor text concepts in the anchor text concept set, the anchor text concept in the anchor text concept set is the vertex, and the weight of the degree of association between the anchor text concepts is the weight between the vertices The weight of the edge constitutes the anchor text concept graph structure ACG; using the anchor text concept to obtain a training positive sample, and using the training positive sample to train a classifier; using the ACG and the classifier to obtain a concept projection vector, and calculate The ACG distances between the pictures in the search results of the initial keyword, and the pictures are sorted according to the ACG distances. The invention makes the sorting of the search results of pictures more accurate, and more in line with the user's search intention.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a picture sorting method and device. Background technique [0002] With the rapid development of Internet technology, network image data is rapidly increasing at an alarming rate. If ordinary Internet users want to use such massive data resources, they must search for pictures. Using keywords to search for images usually results in a lot of images that are somewhat related to the keywords, but usually also results in a lot of images that are little or not at all related to the results required by the user. [0003] In recent years, search engine operators represented by Google, Bing, and Baidu have provided image search functions to provide services for ordinary users to search massive network image data. There are currently two ways to retrieve images: image retrieval using keywords and image retrieval based on content. Using keywords to retrieve images is currently...

AI Technical Summary

Problems solved by technology

However, the diversity of the visual information of the picture itself and the existence of the "semantic gap" make the retrieval effect of CBIR still far from the actual use.

[0004] All in all, the current image retrieval method cannot bring satisfactory results to users. According to the specific information provided by users themselves, the obtained images can be reorganized and sorted, so that users can see more images that meet the requirements with as little effort as possible. became an important requirement

Images