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

A Large-Scale Image Retrieval Method

An image retrieval, large-scale technology, applied in character and pattern recognition, special data processing applications, instruments, etc.

Active Publication Date: 2017-09-29
NANJING UNIV
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] Purpose of the invention: In order to solve the problems in the prior art, the present invention proposes a large-scale image retrieval method, thereby effectively solving the problem of fast and accurate encoding and retrieval of image features under large-scale data

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 Large-Scale Image Retrieval Method
  • A Large-Scale Image Retrieval Method
  • A Large-Scale Image Retrieval Method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0145] This embodiment includes the following parts:

[0146] 1. Image feature extraction

[0147] In this embodiment, the public image data set CIFAR-10 is used to learn a hash function, encode image features, and then perform retrieval. Specifically, for each image in CIFAR-10, an original image pixel gray value feature is extracted: first, the gray level images of all images are obtained through color space conversion, and the gray value of each gray level image is divided into rows Splicing to obtain image features, each image is represented by an image feature, and each image feature is a vector.

[0148] 2. Hash function projection vector learning:

[0149] CIFAR-10 has a total of 10 categories, and 100 image features are randomly selected from each category to form an image feature training set, with a total of 1000 image features.

[0150] Then, learn the hash function projection vector for each category. Taking the first category as an example, it is divided into t...

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 large-scale image retrieval method, comprising the following steps: image feature extraction; hash function projection vector learning; hash function offset learning; image feature dimensionality reduction; image feature encoding; image retrieval. The invention can quickly retrieve large-scale images. First, by learning a discriminative hash function, the discriminability between codes is improved, so as to better distinguish image features of different categories; second, the hash function is used to reduce the dimensionality and code of image features to reduce the storage of image features Computational overhead of the demand and retrieval process. The invention realizes efficient and accurate large-scale image retrieval, and thus has high use value.

Description

technical field [0001] The invention belongs to the field of computer image retrieval, in particular to a large-scale image retrieval method. Background technique [0002] With the rapid development of the Internet, various network resources are increasingly abundant, and the scale of network data is also growing exponentially. Among the various types of data on the Internet, images account for the majority, and have reached a massive scale: in 2010, the total number of images counted by the famous website Flickr exceeded 5 billion. Such data continues to grow at an alarming rate, and will reach an unimaginable scale in a few years. Undoubtedly, it is very important to quickly and accurately search the data needed by users from such a large database, but there are also huge difficulties. For example, given an image, how to quickly and accurately search for images similar to the given image in a large-scale database is a hot research topic at present. However, there are oft...

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): G06F17/30
CPCG06F16/5838G06F18/21
Inventor 杨育彬毛晓蛟
Owner NANJING 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