Convolutional neural network and tree-hash combination indexing-based image retrieval method

A convolutional neural network and image retrieval technology, applied in the field of image retrieval based on convolutional neural network and combined tree and hash index, can solve problems such as poor retrieval time efficiency, improve retrieval effect, and speed up the time to find pictures Effect

Active Publication Date: 2017-09-19
SOUTH CHINA UNIV OF TECH
View PDF4 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Since the tree-based index method has poor retrieval time efficiency when the data dimension is high, the current research focus is mainly on the hash-based index method, among which the better methods are E2LSH and S2JSD-LSH

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
  • Convolutional neural network and tree-hash combination indexing-based image retrieval method
  • Convolutional neural network and tree-hash combination indexing-based image retrieval method
  • Convolutional neural network and tree-hash combination indexing-based image retrieval method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0037] Such as figure 1 As shown, the present invention provides an image retrieval method based on a convolutional neural network and a tree and hash index. The main steps include: collecting an image library, and training a convolutional neural network model for the image library; The model performs feature extraction to obtain the feature library; the index method based on the combination of tree and hash is used to index the feature library; the retrieval is performed according to the feature vector v in the index to obtain the Top-n result.

[0038] The specific implementation is divided into two parts, one is the offline part, and the other is the online part. The offline part is mainly to train the feature extraction model and construct the index based on the combination of tree and hash index method. The online part is the normal use process.

[0039] Such as Figure 7 As shown, first is the offline part. The first is to train a convolutional neural network model based on...

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

The invention discloses a convolutional neural network and tree-hash combination indexing-based image retrieval method. The method comprises the steps of S1, collecting an image library and training a convolutional neural network model for the image library; S2, performing feature extraction on the convolutional neural network model to obtain a feature library; S3, establishing indexes for the feature library by adopting a tree-hash combination indexing method; and S4, performing retrieval according to an eigenvector v in the indexes to obtain a Top-n result. The tree-hash combination indexing method is adopted, so that compared with a pure hash algorithm, the retrieval effect is better under the condition of ensuring the same retrieval time.

Description

Technical field [0001] The invention relates to the technical field of application of massive image retrieval in computers, and in particular to an image retrieval method based on convolutional neural network and a combination index of tree and hash. Background technique [0002] The feature extraction method based on convolutional neural network has been tested in practice, especially vgg and googlenet were among the best in the 2014 imagenet competition. The features extracted based on the convolutional neural network model have rotation invariance and strong robustness. [0003] Since features extracted based on convolutional neural networks are generally of relatively high dimensions, if it is a massive face image feature library, then if linear scanning is to be performed, the efficiency will be very low, so a method that can quickly access specific images is needed . So content-based image retrieval technology (CBIR) came into being. [0004] There are two main index methods...

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): G06F17/30G06N3/04
CPCG06F16/51G06F16/583G06F16/9014G06F16/9027G06N3/045
Inventor 文贵华梁倜
Owner SOUTH CHINA UNIV OF TECH
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