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

Image false matching detection method based on augmented homogeneous coordinate matrix

A homogeneous coordinate and wrong matching technology, applied in image analysis, image data processing, instruments, etc., can solve the problems of sensitivity to matching number and noise interference, algorithm performance degradation, poor detection effect, etc. Fast, adaptable effects

Active Publication Date: 2018-03-23
PEKING UNIV
View PDF3 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0008] In the case of similar transformations and affine transformations between images, the first and second types of methods can effectively remove false matches by checking geometric consistency, but are not suitable for projective transformations
Although the third type of method is suitable for the case of projective transformation, the detection time of RANSAC, MLESAC and VFC is too long and the result is unstable. SparseVFC detection speed is fast, but it is sensitive to the number of matching between images and noise interference, especially the correct matching. When the number is small, the performance of the algorithm degrades seriously
[0009] Therefore, when there is a complex projective transformation between the search image and the target image, which brings large feature detection noise and a high proportion of false matches, the existing methods generally have the problems of long detection time and poor detection effect

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 false matching detection method based on augmented homogeneous coordinate matrix
  • Image false matching detection method based on augmented homogeneous coordinate matrix
  • Image false matching detection method based on augmented homogeneous coordinate matrix

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0120] Dataset: The typical Mikolajczyk and Schmid (document [19]) dataset is used as the retrieved dataset. The data set has 8 groups of pictures, and each group of pictures contains 6 pictures. There are similarities, affine or projective transformation relationships and lighting effects between the 2nd to 6th pictures and the 1st picture. This image set provides the transformation matrix between each group of images to test the reliability of different algorithms.

[0121] Evaluation indicators: This embodiment uses average F-score and average retrieval time to compare with other best methods in the industry. Average F-score is a comprehensive evaluation of recall rate (Recall) and precision rate (Precision), wherein recall rate (Recall )=retrieved matching images / total number of all images, precision (Precision)=retrieved matching images / all matching images.

[0122]F-score=2*Precision*Recall / (Precision+Recall).

[0123] Implementation steps:

[0124] We use ASIFT as a ...

Embodiment 2

[0130] Datasets: Popular datasets are used as the retrieved datasets, namely Holiday[7], GCDup[20] and Oxford5k[21] datasets. Among them, Holiday is a data set with a total of 500 groups of approximately repeated pictures containing 1491 images; GCDup is a data set with a total of 33 groups containing 1104 partially repeated pictures; Oxford5k is a data set with a total of 55 groups containing 5062 high-resolution photos and partially repeated data set. In order to make the example more realistic, this embodiment also downloaded the MIRflickr1M [22] data set from the Flickr website as an interference data set. The MIRflickr1M data set contains one million mutually irrelevant pictures.

[0131] In this embodiment, the first five images of each group in the three benchmark datasets are set as query images. Then, the remaining images in the same group were mixed into different numbers of MIRflickr1M datasets, and finally the retrieval results were checked by observing whether th...

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 false matching detection method based on an augmented homogeneous coordinate matrix. Based on correct matching in one iteration, the augmented homogeneous coordinate (AHC) matrix is constructed, more accurate anchor matching is selected till an Euler's distance between an estimated matching coordinate and an actual matching coordinate corresponding to each pair ofmatching of anchor matching is not greater than the set threshold, and the AHC matrix is constructed based on a final anchor point matching set to complete global verification. The method is advantaged in that a problem of difficulty in false matching during projective image search in the prior art can be effectively solved, high-efficient processing can be carried out even when two correspondingpoint sets has projective transformation and has strong characteristic detection noise and high-proportion false matching, and properties of fast calculation speed, high precision, wide application scope, strong robustness and insensitiveness to large-scale image calculation are realized.

Description

technical field [0001] The invention belongs to the technical field of image retrieval, and relates to a retrieval method for partially repeated images, in particular to a method for detecting a mismatch between images based on an augmented homogeneous coordinate matrix, which is suitable for detecting a mismatch between images in projective transformation, that is, in When there is a complex projective transformation between the original image and the image to be matched, and there may be a large number of mismatches and high noise, it can quickly and efficiently detect the mismatch between images. Background technique [0002] Partially repeated images mainly refer to pictures of the same scene taken at different angles or pictures before and after processing by image processing software. The geometric relationship between these images can be summarized as follows: similarity transformation is that there are only transformations such as displacement, scaling and rotation b...

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
IPC IPC(8): G06T7/00G06T7/33G06F17/30
CPCG06F16/583G06T7/0002G06T7/33
Inventor 林宙辰征妍
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