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

Image retrieval method based on feature point clustering

An image retrieval and feature point technology, applied in digital data information retrieval, special data processing applications, instruments, etc., can solve the problems of retrieval failure, multiple interference content of images, etc., and achieve the effect of retrieval and reduction of interference.

Inactive Publication Date: 2020-02-18
珩鑫科技(北京)有限公司
View PDF0 Cites 2 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 an image retrieval method based on feature point clustering, which aims to overcome the above-mentioned deficiencies in the prior art, and it can effectively solve the problem of retrieval failure due to the existence of more disturbing content in the image

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
  • Image retrieval method based on feature point clustering
  • Image retrieval method based on feature point clustering
  • Image retrieval method based on feature point clustering

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0042] An image retrieval method based on feature point clustering, comprising the following steps:

[0043] Step 1: Extract local feature points of all images in the image database;

[0044] Local feature point in described step 1 can be SIFT feature point, adopts SIFT feature extraction operator vl_sift in MATLAB, extracts the SIFT feature point of all images in the image database;

[0045] Step 2: Use the K-means clustering algorithm to cluster all the local feature points extracted in step 1 to obtain K cluster centers, where the value of K is K=1-1000;

[0046] Step 3: For each image in the database, calculate its local feature aggregation descriptor based on the local feature points of the image extracted in step 1 and the K cluster centers obtained in step 2;

[0047] Step 4: For the retrieved image, extract its local feature points, and use the K-means clustering algorithm to cluster the local feature points to generate K' classes and get K' class centers, where the v...

Embodiment 2

[0057] The difference from embodiment 1 is that the local feature points in step 1 may also be SURF feature points, ORB feature points, HOG feature points, FAST feature points, BRISK feature points or LBP feature points.

Embodiment 3

[0059] On the basis of embodiment 1. The specific method of step 3 is as follows:

[0060] Step 3-1: Calculate the class number of each feature point in the image:

[0061] (Formula 1)

[0062] in, Indicates the tth feature point of the image, , n represents the number of image feature points, Indicates the jth cluster center, , i means the obtained Class number;

[0063] Step 3-2: Calculate the residual vector for each cluster:

[0064] (Formula 2)

[0065] in, Indicates the i-th cluster center, Indicates the kth feature point belonging to the i-th cluster in the image, and m indicates the total number of feature points in the image belonging to the i-th cluster; Represents the residual vector of the i-th cluster;

[0066] Step 3-3: Combine the k residual vectors obtained in step 3-2 into a one-dimensional vector using the following formula:

[0067] (Formula 3)

[0068] Step 3-4: Perform power-law normalization on each component in the one-dime...

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 an image retrieval method based on feature point clustering, and the method comprises the following steps: extracting local feature points of all images in an image database, and carrying out the clustering of all extracted local feature points; calculating a local feature aggregation descriptor; for a retrieval image, extracting local feature points of the retrieval image,and clustering the local feature points by using a clustering algorithm; calculating the shortest distance dmin among all class centers of the retrieval image; setting a threshold value T, and if dmin is smaller than or equal to T, combining the two classes corresponding to the shortest distance; repeating the previous step until dmin is greater than T; selecting image feature points in the classwith the maximum number of the retrieved images, and calculating a local feature aggregation descriptor; and calculating the distances between the local feature aggregation descriptors obtained in the previous step and the local feature aggregation descriptors of all the images in the database, wherein the database image corresponding to the minimum distance is a retrieval result. Interference ofsurrounding background image content can be effectively reduced, and successful retrieval of the image is achieved.

Description

technical field [0001] The invention relates to an image retrieval method, in particular to an image retrieval method based on feature point clustering. Background technique [0002] With the rapid development of computers and the Internet, image resources are becoming more and more abundant. How to accurately retrieve the images that users need from large-scale image resources has become a key problem that needs to be solved urgently. Therefore, to establish an accurate image retrieval method has become a current research hotspot. [0003] In the field of image retrieval, the Bag of Word (BoW) model is one of the most commonly used retrieval methods. This method first clusters all the features of the image in the image database, each cluster center is a visual vocabulary, and all the cluster centers together form a visual codebook; then, all the features of each image are mapped to the visual codebook In , a word frequency vector corresponding to the visual codebook is ob...

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): G06F16/583G06K9/62
CPCG06F18/23213
Inventor 史凌波刘文龙
Owner 珩鑫科技(北京)有限公司
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