Fast and Robust Image Matching Based on Depth-Wise Convolution Features and Unique Nearest Neighbour Similarity

Li Ruan; Yuanjie Jiang; Chang Yang; Yiyang Xing; Limin Xiao; Xiangwen Qu

doi:10.1007/978-3-030-70665-4_100

Fast and Robust Image Matching Based on Depth-Wise Convolution Features and Unique Nearest Neighbour Similarity

Li Ruan^*
, Yuanjie Jiang
, Chang Yang
, Yiyang Xing
, Limin Xiao
, Xiangwen Qu

^*Corresponding author for this work

School of Computer Science and Engineering

Beihang University

Research output: Chapter in Book/Report/Conference proceeding › Chapter › peer-review

Abstract

Finding a template patch in a target image, including object detection and tracking, is a core component in computer vision applications. This paper introduces a fast and robust image matching based on depth-wise convolution features and unique nearest neighbour similarity. Experiments with its results show that our algorithm can effectively solve the problem of image mapping with complex deformation, and the matching performance is better than the existing optimal feature matching algorithms by improving the matching accuracy about 10.68%.

Original language	English
Title of host publication	Lecture Notes on Data Engineering and Communications Technologies
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	933-940
Number of pages	8
DOIs	https://doi.org/10.1007/978-3-030-70665-4_100
State	Published - 2021

Publication series

Name	Lecture Notes on Data Engineering and Communications Technologies
Volume	88
ISSN (Print)	2367-4512
ISSN (Electronic)	2367-4520

Keywords

Deep learning
Depth-wise convolution
Image matching

Access to Document

10.1007/978-3-030-70665-4_100

Cite this

Ruan, L., Jiang, Y., Yang, C., Xing, Y., Xiao, L., & Qu, X. (2021). Fast and Robust Image Matching Based on Depth-Wise Convolution Features and Unique Nearest Neighbour Similarity. In Lecture Notes on Data Engineering and Communications Technologies (pp. 933-940). (Lecture Notes on Data Engineering and Communications Technologies; Vol. 88). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-030-70665-4_100

@inbook{b839fb885e8c4b138fef8bb4f978b847,

title = "Fast and Robust Image Matching Based on Depth-Wise Convolution Features and Unique Nearest Neighbour Similarity",

abstract = "Finding a template patch in a target image, including object detection and tracking, is a core component in computer vision applications. This paper introduces a fast and robust image matching based on depth-wise convolution features and unique nearest neighbour similarity. Experiments with its results show that our algorithm can effectively solve the problem of image mapping with complex deformation, and the matching performance is better than the existing optimal feature matching algorithms by improving the matching accuracy about 10.68\%.",

keywords = "Deep learning, Depth-wise convolution, Image matching",

author = "Li Ruan and Yuanjie Jiang and Chang Yang and Yiyang Xing and Limin Xiao and Xiangwen Qu",

note = "Publisher Copyright: {\textcopyright} 2021, The Author(s), under exclusive license to Springer Nature Switzerland AG.",

year = "2021",

doi = "10.1007/978-3-030-70665-4\_100",

language = "英语",

series = "Lecture Notes on Data Engineering and Communications Technologies",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "933--940",

booktitle = "Lecture Notes on Data Engineering and Communications Technologies",

address = "德国",

}

Ruan, L, Jiang, Y, Yang, C, Xing, Y, Xiao, L & Qu, X 2021, Fast and Robust Image Matching Based on Depth-Wise Convolution Features and Unique Nearest Neighbour Similarity. in Lecture Notes on Data Engineering and Communications Technologies. Lecture Notes on Data Engineering and Communications Technologies, vol. 88, Springer Science and Business Media Deutschland GmbH, pp. 933-940. https://doi.org/10.1007/978-3-030-70665-4_100

Fast and Robust Image Matching Based on Depth-Wise Convolution Features and Unique Nearest Neighbour Similarity. / Ruan, Li; Jiang, Yuanjie; Yang, Chang et al.
Lecture Notes on Data Engineering and Communications Technologies. Springer Science and Business Media Deutschland GmbH, 2021. p. 933-940 (Lecture Notes on Data Engineering and Communications Technologies; Vol. 88).

Research output: Chapter in Book/Report/Conference proceeding › Chapter › peer-review

TY - CHAP

T1 - Fast and Robust Image Matching Based on Depth-Wise Convolution Features and Unique Nearest Neighbour Similarity

AU - Ruan, Li

AU - Jiang, Yuanjie

AU - Yang, Chang

AU - Xing, Yiyang

AU - Xiao, Limin

AU - Qu, Xiangwen

PY - 2021

Y1 - 2021

N2 - Finding a template patch in a target image, including object detection and tracking, is a core component in computer vision applications. This paper introduces a fast and robust image matching based on depth-wise convolution features and unique nearest neighbour similarity. Experiments with its results show that our algorithm can effectively solve the problem of image mapping with complex deformation, and the matching performance is better than the existing optimal feature matching algorithms by improving the matching accuracy about 10.68%.

AB - Finding a template patch in a target image, including object detection and tracking, is a core component in computer vision applications. This paper introduces a fast and robust image matching based on depth-wise convolution features and unique nearest neighbour similarity. Experiments with its results show that our algorithm can effectively solve the problem of image mapping with complex deformation, and the matching performance is better than the existing optimal feature matching algorithms by improving the matching accuracy about 10.68%.

KW - Deep learning

KW - Depth-wise convolution

KW - Image matching

UR - https://www.scopus.com/pages/publications/85116875875

U2 - 10.1007/978-3-030-70665-4_100

DO - 10.1007/978-3-030-70665-4_100

M3 - 章节

AN - SCOPUS:85116875875

T3 - Lecture Notes on Data Engineering and Communications Technologies

SP - 933

EP - 940

BT - Lecture Notes on Data Engineering and Communications Technologies

PB - Springer Science and Business Media Deutschland GmbH

ER -

Ruan L, Jiang Y, Yang C, Xing Y, Xiao L, Qu X. Fast and Robust Image Matching Based on Depth-Wise Convolution Features and Unique Nearest Neighbour Similarity. In Lecture Notes on Data Engineering and Communications Technologies. Springer Science and Business Media Deutschland GmbH. 2021. p. 933-940. (Lecture Notes on Data Engineering and Communications Technologies). doi: 10.1007/978-3-030-70665-4_100

Fast and Robust Image Matching Based on Depth-Wise Convolution Features and Unique Nearest Neighbour Similarity

Abstract

Publication series

Keywords

Access to Document

Other files and links

Fingerprint

Cite this