Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

A Fast and Optimized Deep Hash Image Coding Method and Target Image Retrieval Method

A technology of image coding and hash coding, which is applied in the field of information retrieval, can solve problems such as limited application scope and slow network training, and achieve the effects of ensuring accuracy, fast training speed, and excellent retrieval performance

Active Publication Date: 2022-03-25
PEKING UNIV
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Although the retrieval performance has been improved, HashNet and DSDH still have problems such as slow network training and limited application range

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
  • A Fast and Optimized Deep Hash Image Coding Method and Target Image Retrieval Method
  • A Fast and Optimized Deep Hash Image Coding Method and Target Image Retrieval Method
  • A Fast and Optimized Deep Hash Image Coding Method and Target Image Retrieval Method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0071] Below in conjunction with the accompanying drawings, the present invention is further described by means of embodiments, but the scope of the present invention is not limited in any way.

[0072] The present invention provides a fast optimization depth hash image coding method and target image retrieval method based on greedy strategy. The greedy strategy is used to solve the discrete optimization problem of depth hash, and the approximate optimal solution that satisfies the discrete constraints in the current situation is iteratively obtained. to update the network for fast and efficient training. By designing a new deep hash coding module, the sign function is strictly used during forward propagation to keep the discrete constraints always established, avoiding the problem of quantization error, and the gradient is completely returned to the front-layer network during back-propagation. At the same time as the problem of gradient disappearance, each coding bit is updat...

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 fast optimized deep hash image coding method and a target image retrieval method. Based on a greedy strategy, a hash image coding model is established for a large image data set, and all images are generated through the optimized deep hash coding network binary code of . When performing target image retrieval, similar images of the same type to the query image can be quickly obtained by calculating the Hamming distance between the query image code and the database image code. The method of the present invention combines the neural network to better solve the problems of gradient disappearance and quantization error, and the coding performance is better; the training process of the deep network is completed with fewer iterations, and the training speed is faster; it can be applied to various systems with discrete constraints problems, and a wider range of applications; further improving the optimization speed of the deep neural network and the retrieval performance of the generated image codes, effectively improving the retrieval accuracy of large image databases.

Description

technical field [0001] The invention belongs to the technical field of information retrieval, relates to image processing and fast image retrieval technologies, and in particular relates to a fast optimized deep hash image coding method based on a greedy strategy and a target image retrieval method. Background technique [0002] With the advent of the era of big data, data in various fields has exploded. In such a large wave of data, how to retrieve the information you need has become an important and urgent research topic. The hash algorithm is an algorithm for quickly completing target image retrieval on large image data sets. As the feature representation of the image), the Hamming distance is obtained through the fast XOR operation between the binary codes, so as to complete the approximate nearest neighbor image retrieval after sorting (that is, find the image closest to the query image from the image database) . This representation of binary image features can bring ...

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 Patents(China)
IPC IPC(8): G06F16/53G06N3/04G06N3/08G06T9/00
CPCG06F16/53G06N3/084G06T9/00G06N3/045
Inventor 张超苏树鹏韩凯田永鸿
Owner PEKING UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap 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