3D Correspondence Grouping with Compatibility Features

Jiaqi Yang, Jiahao Chen, Zhiqiang Huang, Zhiguo Cao, Yanning Zhang

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

5 Scopus citations

Abstract

We present a simple yet effective method for 3D correspondence grouping. The objective is to accurately classify initial correspondences obtained by matching local geometric descriptors into inliers and outliers. Although the spatial distribution of correspondences is irregular, inliers are expected to be geometrically compatible with each other. Based on such observation, we propose a novel feature representation for 3D correspondences, dubbed compatibility feature (CF), to describe the consistencies within inliers and inconsistencies within outliers. CF consists of top-ranked compatibility scores of a candidate to other correspondences, which purely relies on robust and rotation-invariant geometric constraints. We then formulate the grouping problem as a classification problem for CF features, which is accomplished via a simple multilayer perceptron (MLP) network. Comparisons with nine state-of-the-art methods on four benchmarks demonstrate that: 1) CF is distinctive, robust, and rotation-invariant; 2) our CF-based method achieves the best overall performance and holds good generalization ability.

Original languageEnglish
Title of host publicationPattern Recognition and Computer Vision - 4th Chinese Conference, PRCV 2021, Proceedings
EditorsHuimin Ma, Liang Wang, Changshui Zhang, Fei Wu, Tieniu Tan, Yaonan Wang, Jianhuang Lai, Yao Zhao
PublisherSpringer Science and Business Media Deutschland GmbH
Pages66-78
Number of pages13
ISBN (Print)9783030880064
DOIs
StatePublished - 2021
Event4th Chinese Conference on Pattern Recognition and Computer Vision, PRCV 2021 - Beijing, China
Duration: 29 Oct 20211 Nov 2021

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13020 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference4th Chinese Conference on Pattern Recognition and Computer Vision, PRCV 2021
Country/TerritoryChina
CityBeijing
Period29/10/211/11/21

Keywords

  • 3D correspondence grouping
  • 3D point cloud
  • Compatability feature

Fingerprint

Dive into the research topics of '3D Correspondence Grouping with Compatibility Features'. Together they form a unique fingerprint.

Cite this