Camera pose estimation from lines: a fast, robust and general method

Ping Wang, Guili Xu, Yuehua Cheng, Qida Yu

Research output: Contribution to journalArticlepeer-review

26 Scopus citations

Abstract

In this paper, we revisit the perspective-n-line problem and propose a closed-form solution that is fast, robust and generally applicable. Our main idea is to formulate the pose estimation problem into an optimal problem. Our method only needs to solve a fifteenth-order and a fourth-order univariate polynomial, respectively, which makes the processes more easily understood and significantly improves the performance. Experiment results show that our method offers accuracy and precision comparable or better than existing state-of-the-art methods, but with significantly lower computational cost. This superior computational efficiency is particularly suitable for real applications.

Original languageEnglish
Pages (from-to)603-614
Number of pages12
JournalMachine Vision and Applications
Volume30
Issue number4
DOIs
StatePublished - 1 Jun 2019
Externally publishedYes

Keywords

  • Absolute position and orientation
  • Camera pose estimation
  • Computer vision
  • Perspective-n-line problem (PnL)

Fingerprint

Dive into the research topics of 'Camera pose estimation from lines: a fast, robust and general method'. Together they form a unique fingerprint.

Cite this