A Cross-media Retrieval Method Based on Locality Sensitive Hash Algorithm and Neural Network

Abstract

The invention discloses a cross-media retrieval method based on local sensitive hash algorithm and neural network, which relates to the technical field of cross-media retrieval. The method includes two stages of local sensitive hash and hash function learning. stage, image data is mapped to m hash tables G=[g 1 , g 2 ,..., g m ]∈R k×m In the hash bucket, where G is a collection of m hash tables, g j Indicates the jth hash table, k is the length of the hash code corresponding to the hash bucket; in the hash function learning stage, the text data is mapped to the corresponding hash buckets in the m hash tables through neural network algorithm learning Hash function Ht=(Ht (1) , Ht (2) ,...,Ht (m) ), Ht (j) , (1≤j≤m) represents the learned hash function Ht corresponding to the jth hash table. After obtaining the functions of these two stages, all images and documents are further coded and indexed for more accurate retrieval.

Description

technical field [0001] The invention relates to the technical field of cross-media retrieval, in particular to a cross-media retrieval method based on local sensitive hash algorithm and neural network. Background technique [0002] In the era of cross-media big data, massive multi-modal information generated all the time has brought huge cross-media retrieval requirements, such as using text to search for images or videos, and vice versa. For example, an entry on Wikipedia usually contains text descriptions and example images, and the retrieval of such information requires the construction of cross-media indexes and learning methods. Compared with traditional single-media retrieval, the core issue of cross-media retrieval is how to mine the associations between the same or related semantic objects represented by different media. [0003] At present, many solutions to the core problem of cross-media retrieval have been proposed around the world. The existing cross-media ret...

AI Technical Summary

Problems solved by technology

[0014] The existing cross-media retrieval methods all have the same technical defect, that is, only considering the cross-media retrieval method itself and ignoring some feasible optimization processing of the document set, because there are a large number of irrelevant documents in the document set, so Preprocessing the document set before precise query, increasing the proportion of relevant documents in the document set is of great significance for improving retrieval efficiency

Images