Eureka AIR delivers breakthrough ideas for toughest innovation challenges, trusted by R&D personnel around the world.

Semi-supervised image retrieval method based on self-coding network and robust kernel hashing

A self-encoding network and image retrieval technology, which is applied in the field of large-scale image retrieval, can solve problems such as poor robustness and large amount of calculation, and achieve the effect of improving robustness and reducing computational complexity

Pending Publication Date: 2019-08-02
CENT SOUTH UNIV
View PDF0 Cites 6 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the KSH method has poor robustness and a large amount of calculation

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
  • Semi-supervised image retrieval method based on self-coding network and robust kernel hashing
  • Semi-supervised image retrieval method based on self-coding network and robust kernel hashing
  • Semi-supervised image retrieval method based on self-coding network and robust kernel hashing

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0044] The present invention first uses a self-encoding network structure to extract deep image features. The process of training the self-encoding network is to sequentially input the images in the public image data set into the network, and use the random gradient descent method to update the network parameters by calculating the loss of the network. When the training reaches the specified number of times Or the training is completed when the loss function becomes stable. After the training is completed, the network structure of the encoding stage is selected as the image depth feature extraction network. Since the images in the retrieval library are relatively fixed, in order to speed up the retrieval, all the images in the retrieval library can be pre-generated with corresponding depth features.

[0045] The self-encoding network is an encoding-decoding structure. In the encoding stage, the input feature map of each layer is represented by a three-dimensional array [h,w,d], wh...

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 semi-supervised image retrieval method based on a self-coding network and robust kernel hashing. Step 1, depth image features are extracted by adopting a self-coding networkstructure; step 2, a robust kernel Hash mechanism is taken as a retrieval mechanism; and step 3, the self-coding network and the robust kernel Hash are combined to complete the retrieval process, andcomplete image features are extracted by using the self-coding network, and the image does not need to be labeled manually, so that the method has good learning ability and expression ability for different data sets, and has a better effect for large-scale data sets. Meanwhile, a Robust Hashwith Kernels (RSH) retrieval mechanism is proposed to be used, the robustness of the KSH method is improved,the calculation amount in the retrieval process is reduced, and finally the retrieval precision is improved and the storage space and the calculation complexity are greatly reduced through the imageretrieval algorithm combining the self-encoding network and the robust kernel Hash.

Description

Technical field [0001] The invention relates to the technical field of computer image retrieval, in particular to large-scale image retrieval. Background technique [0002] The pros and cons of image retrieval algorithms mainly measure retrieval accuracy, retrieval efficiency and the size of storage space. Therefore, the two core issues of image retrieval are how to extract effective image features and how to design efficient retrieval mechanisms. At present, the commonly used method of extracting image features is to use shallow features such as SIFT, HOG, GIST, etc. These features cannot fully express images that contain rich visual information and have certain limitations. The mainstream retrieval mechanism is to use the hash method to generate multiple binary codes for the image. The representative algorithm has the supervised hash with kernels (KSH) based on the kernel function, which uses the characteristics of the kernel function in processing linear indivisible data. Con...

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
Patent Type & Authority Applications(China)
IPC IPC(8): G06F16/583
CPCG06F16/583
Inventor 王勇李仪万明阳刘星辰谢斌
Owner CENT SOUTH UNIV
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
Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com