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

A Method of Image Feature Point Matching

A technology of image feature points and matching methods, which is applied in the field of image search, can solve problems such as not being able to adapt to the image library mode, and achieve the effects of retrieval accuracy and high efficiency, high retrieval accuracy, and reduced training time

Active Publication Date: 2020-06-30
UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF2 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Using the existing feature point matching method, it can no longer adapt to the existing fast-growing image library mode

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

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0032] A method for matching image feature points, comprising the following steps,

[0033] Feature point extraction of stored images: extract the features of stored images and form a stored feature vector, and reduce its dimension;

[0034] Vector warehousing: Divide the warehousing feature vector after dimension reduction, and do product quantization and then vector quantization for each divided part to form a product quantizer and vector quantizer, and establish a retrieval tree and a hash table;

[0035] Extraction of feature points of images to be matched: extract features of images to be matched and form feature vectors to be matched, and reduce their dimensions;

[0036] Vector matching: Divide the feature vectors to be matched after dimensionality reduction, find multiple cluster centers that are closer to the cluster centers of the product quantizer and vector quantizer, and find multiple cluster centers according to the search tree and hash table. The pictures corre...

Embodiment 2

[0039] Based on the ideas of the foregoing embodiments, this embodiment refines each step.

[0040] Regardless of whether it is a picture that is still waiting to be matched for a picture stored as a retrieval tree, feature points need to be extracted. There are many ways to extract feature points, such as convolutional neural networks, and the dimension of the output feature vector is a relatively large value. It can be set to n, and n may be 128, 256 or 512, etc. Its dimension is large, which will increase the amount of calculation in the matching process. We need to reduce its dimensionality. For dimensionality reduction, the principal component analysis method can be used to reduce the dimensionality of the output feature vector to d dimensions, where d is less than or equal to n, d 128 or 64 is desirable. The method of dimensionality reduction can not only remove the influence of noise, but also reduce the amount of calculation and the calculation time.

[0041] The spe...

Embodiment 2

[0071] Regarding embodiment 2, a detailed implementation manner is now disclosed.

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 method for matching image feature points, which comprises the following steps: extracting feature points of stored pictures: extracting features of stored images to form a stored feature vector, and performing dimension reduction on its dimension; vector storage: segmentation and dimension reduction The final input feature vector, and perform product quantization on each part after division, and then vector quantization to form a product quantizer and a vector quantizer, and establish a retrieval tree and hash table; feature point extraction of images to be matched: extraction to be Match the features of the image and form the feature vector to be matched, and reduce its dimension; Vector matching: segment the feature vector to be matched after dimensionality reduction, and find out the cluster center distance between the feature vector to be matched and the product quantizer and vector quantizer For multiple cluster centers in the front, find the pictures corresponding to multiple cluster centers according to the search tree and hash table and form a candidate set, and use floating point vectors to calculate the picture with the closest distance between the candidate set and the feature vector to be matched; Fast speed and high precision.

Description

technical field [0001] The invention relates to the technical field of image search, in particular to an image feature point matching method. Background technique [0002] In the field of image search, feature matching is a very important link, and the matching efficiency and accuracy of features determine the final search speed and accuracy. When searching existing images, it adopts the following steps: the first step is to train a transformation matrix through a large number of sample data, convert the binary code through the hash function, segment the binary code, generate multiple hash tables, and obtain the segmented binary code directly As the entry of the hash table. In the second step, when the vector to be queried arrives, it is converted into a binary code in the same way, and mapped to the corresponding hash table entry and other entries whose distance is r, and all pictures in the entry are used as candidate sets. The third step is to calculate the complete Ham...

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): G06F16/583G06F16/55G06F16/51G06K9/62
CPCG06F16/51G06F16/583G06F18/23213
Inventor 段翰聪赵子天谭春强文慧闵革勇陈超李博洋
Owner UNIV OF ELECTRONICS SCI & TECH OF CHINA
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