cross-modal Hash learning method based on an anchor point graph

A learning method and cross-modal technology, applied in other database retrieval based on metadata, character and pattern recognition, instruments, etc., can solve the problem of keeping beneficial information of feature data and not completely solving it

Active Publication Date: 2019-04-19
JIUJIANG UNIVERSITY
View PDF0 Cites 8 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to provide a cross-modal hash learning method based on anchor graphs, which solves the problem that existing cross-modal hash learning methods have not completely solved the problem of maintaining features based on graph structures on large-scale data sets. The problem of beneficial inf

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • cross-modal Hash learning method based on an anchor point graph
  • cross-modal Hash learning method based on an anchor point graph
  • cross-modal Hash learning method based on an anchor point graph

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0031] A cross-modal hash learning method based on anchor graphs, which establishes the characteristics of n objects in image mode and text mode as with in, with represent the feature vectors of the i-th object in image mode and text mode respectively, i=1,2,...,n,d 1 and d 2 represent the dimensions of the feature vectors of the image modality and the text modality respectively; at the same time, it is assumed that the feature vectors of the image modality and the text modality are preprocessed by zero centralization, that is, satisfy suppose with are the adjacency matrices of image modality and text modality samples respectively; matrix A (1) elements in and matrix A (2) elements in Represents the similarity between the i-th sample and the j-th sample in the image modality and the text modality respectively; suppose S∈{0,1} n×n is the semantic correlation matrix between samples in two modalities, where S ij Indicates the semantic correlation between the i...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

A cross-modal hash learning method based on anchor map, characterized in that the method comprises the following steps: (1) using an objective function designed based on anchor map technology Binaryhash encoding of objects in image modal and text modal and the projection matrix of the image modal and text modal; (2) In view of the non-convex nature of the objective function, solving the unknownvariable in the objective function by alternate updating; (3) projection matrix based on the image modal and text modal obtained by the solution and generating a binary hash code for the sample of the query and the samples in the retrieved sample set; (4) Calculating the Hamming distance of the query sample to each sample in the sample set based on the generated binary hash code; (5) performing the retrieval of the query sample using the cross-modal retriever based on the approximate nearest neighbor search. The method can quickly obtain an approximate matrix of the true similarity matrix based on the anchor map technique.

Description

technical field [0001] The invention relates to a cross-modal hash learning method based on an anchor graph. Background technique [0002] With the rapid development of information technology, human society has entered the era of big data, and massive data from different fields and different applications will be generated all the time. Facing the explosive growth of data, how to quickly retrieve the required information to ensure the effective use of data has become an urgent and very challenging problem to be solved in the era of big data. [0003] Nearest neighbor search, also known as similarity search, plays an important role in many applications such as document retrieval, object recognition, and approximate image detection. Among the methods for approximate nearest neighbor search, the hash-based search (retrieval) method has received more and more attention in recent years. The hash-based search method can map high-dimensional feature data into compact binary hash c...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
IPC IPC(8): G06F16/907G06K9/62
CPCG06F18/213G06F18/22
Inventor 董西伟邓安远胡芳贾海英周军孙丽杨茂保王海霞
Owner JIUJIANG UNIVERSITY
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap