Skip to main content

Research Repository

Advanced Search

Using outlier elimination to assess learning-based correspondence matching methods

Ding, Xintao; Luo, Yonglong; Jie, Biao; Li, Qingde; Cheng, Yongqiang

Authors

Xintao Ding

Yonglong Luo

Biao Jie

Yongqiang Cheng



Abstract

Recently, deep learning (DL) technology has been widely used in correspondence matching. The learning-based models are usually trained on benign image pairs with partial overlaps. Since DL model is usually data-dependent, non-overlapping images may be used as poison samples to fool the model and produce false registrations. In this study, we propose an outlier elimination-based assessment method (OEAM) to assess the registrations of learning-based correspondence matching method on partially overlapping and non-overlapping images. OEAM first eliminates outliers based on spatial paradox. Then OEAM implements registration assessment in two streams using the obtained core correspondence set. If the cardinality of the core set is sufficiently small, the input registration is assessed as a low-quality registration. Otherwise, it is assessed to be of high quality, and OEAM improves its registration performance using the core set. OEAM is a post-processing technique imposed on learning-based method. The comparison experiments are implemented on outdoor (YFCC100M) and indoor (SUN3D) datasets using four deep learning-based methods. The experimental results on registrations of partially overlapping images show that OEAM can reliably infer low-quality registrations and improve performance on high-quality registrations. The experiments on registrations of non-overlapping images demonstrate that learning-based methods are vulnerable to poisoning attacks launched by non-overlapping images, and OEAM is robust against poisoning attacks crafted by non-overlapping images.

Citation

Ding, X., Luo, Y., Jie, B., Li, Q., & Cheng, Y. (2024). Using outlier elimination to assess learning-based correspondence matching methods. Information Sciences, 659, Article 120056. https://doi.org/10.1016/j.ins.2023.120056

Journal Article Type Article
Acceptance Date Dec 25, 2023
Online Publication Date Jan 2, 2024
Publication Date 2024-02
Deposit Date Jan 12, 2024
Publicly Available Date Jan 3, 2025
Journal Information Sciences
Print ISSN 0020-0255
Publisher Elsevier
Peer Reviewed Peer Reviewed
Volume 659
Article Number 120056
DOI https://doi.org/10.1016/j.ins.2023.120056
Keywords Correspondence matching; Multi-scale; Outer neighborhood; Poisoning attacks; Registration assessment
Public URL https://hull-repository.worktribe.com/output/4511485

Files

This file is under embargo until Jan 3, 2025 due to copyright reasons.

Contact Q.Li@hull.ac.uk to request a copy for personal use.




You might also like



Downloadable Citations