A kind of image semantic retrieval method and system

Abstract

The invention belongs to the field of image retrieval, and discloses an image semantic retrieval method and system, which redefines the DEF item and expands the DEF item into the union of the basic sememe contained in the word and the relational sememe in the word; the sentence is established as a unit The document model before the concept is acquired; enter the sentence scanning to establish the vector space model; scan the words in the sentence one by one to find out whether there is a corresponding DEF item and a word with the same meaning as the DEF item. In the present invention, the router logs in the database server, queries through the information input module, the database server is connected to the semantic application server, and after the user sends a retrieval command, the semantic application server converts the query through the query conversion submodule Instructions, through the decision-making implementation module to call the database server to complete image semantic retrieval, through the dynamic generation and release sub-module of the scheme to return data to the user terminal.

Description

technical field [0001] The invention belongs to the field of image retrieval, in particular to an image semantic retrieval method and system. Background technique [0002] With the rapid development of computer, multimedia and network technology, massive digital image information resources appear on the Internet. In recent years, image search is the fastest-growing category search application, and the number of image searches on several major search engines in the world has doubled. At present, image retrieval technology has become a research hotspot at home and abroad. [0003] In order to overcome the shortcomings of image retrieval methods based on simple visual features, people have proposed semantic-based image retrieval methods, which are different from queries based on low-level physical features. "Semantic feature queries are text-based queries" include natural language processing and traditional image retrieval techniques. . [0004] In summary, the problems in t...

AI Technical Summary

Problems solved by technology

[0005] When extracting image features, the traditional image cutting method has high time complexity and space complexity, and the number of segmentation blocks needs to be selected in advance; when the edge information is not obvious or the noise has a great influence on the edge, the method based on edge information, The problem of poor semantic retrieval

[0006] The semantic extraction method based on human interaction needs to manually label the image first, which often requires a lot of work; the semantic extraction method based on external information uses the external information source when the external information of the image is not easy to obtain or when the external information is very little. method will be difficult

Images