An Efficient and Consistent Solution to the PnP Problem

Xiaoyan Zhou, Zhengfeng Xie, Qida Yu, Yuan Zong, Yiru Wang

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

Abstract

In this paper, we present a novel non-iterative algorithm for solving the pose estimation problem from a set of 3D-to-2D point correspondences, known as the Perspective-n-Point (PnP) problem. The presented algorithm is capable of achieving both geometrical and statistical optimality by exploring the geometrical constraints of the PnP problem through a nonlinear least-squares fashion, as well as accounting for observation uncertainty in the solution process. In addition, to further improve the accuracy of the presented algorithm, we introduce a method that is able to eliminate the bias of solution caused by the propagation of uncertainty, resulting in a consistent estimate. Experimental tests on synthetic data and real images (i.e., TempleRing dataset) show that the presented algorithm can well adapt to different levels of noise, and out-perform state-of-the-art (SOTA) PnP algorithms in terms of accuracy and computational cost. This makes the presented algorithm eminently suitable for a wide range of application scenarios.

Original languageEnglish
Title of host publicationPattern Recognition and Computer Vision - 6th Chinese Conference, PRCV 2023, Proceedings
EditorsQingshan Liu, Hanzi Wang, Rongrong Ji, Zhanyu Ma, Weishi Zheng, Hongbin Zha, Xilin Chen, Liang Wang
PublisherSpringer Science and Business Media Deutschland GmbH
Pages207-220
Number of pages14
ISBN (Print)9789819984312
DOIs
StatePublished - 2024
Externally publishedYes
Event6th Chinese Conference on Pattern Recognition and Computer Vision, PRCV 2023 - Xiamen, China
Duration: 13 Oct 202315 Oct 2023

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume14426 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference6th Chinese Conference on Pattern Recognition and Computer Vision, PRCV 2023
Country/TerritoryChina
CityXiamen
Period13/10/2315/10/23

Keywords

  • PnP Problem
  • Pose Estimation
  • Statistically Optimal

Fingerprint

Dive into the research topics of 'An Efficient and Consistent Solution to the PnP Problem'. Together they form a unique fingerprint.

Cite this