A direct method for estimating planar projective transform

Yu Tseh Chi, Jeffrey Ho, Ming Hsuan Yang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

5 Citations (Scopus)

Abstract

Estimating planar projective transform (homography) from a pair of images is a classical problem in computer vision. In this paper, we propose a novel algorithm for direct registering two point sets in using projective transform without using intensity values. In this very general context, there is no easily established correspondences that can be used to estimate the projective transform, and most of the existing techniques become either inadequate or inappropriate. While the planar projective transforms form an eight-dimensional Lie group, we show that for registering 2D point sets, the search space for the homographies can be effectively reduced to a three-dimensional space. To further improve on the running time without significantly reducing the accuracy of the registration, we propose a matching cost function constructed using local polynomial moments of the point sets and a coarse to fine approach. The resulting registration algorithm has linear time complexity with respect to the number of input points. We have validated the algorithm using points sets collected from real images. Preliminary experimental results are encouraging and they show that the proposed method is both efficient and accurate.

Original languageEnglish
Title of host publicationComputer Vision, ACCV 2010 - 10th Asian Conference on Computer Vision, Revised Selected Papers
Pages268-281
Number of pages14
EditionPART 2
DOIs
Publication statusPublished - 2011 Mar 16
Event10th Asian Conference on Computer Vision, ACCV 2010 - Queenstown, New Zealand
Duration: 2010 Nov 82010 Nov 12

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
NumberPART 2
Volume6493 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference10th Asian Conference on Computer Vision, ACCV 2010
CountryNew Zealand
CityQueenstown
Period10/11/810/11/12

Fingerprint

Direct Method
Point Sets
Transform
Registration
Lie groups
Homography
Cost functions
Local Polynomial
Computer vision
Linear Complexity
Polynomials
Computer Vision
Search Space
Time Complexity
Cost Function
Correspondence
Moment
Three-dimensional
Experimental Results
Estimate

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Chi, Y. T., Ho, J., & Yang, M. H. (2011). A direct method for estimating planar projective transform. In Computer Vision, ACCV 2010 - 10th Asian Conference on Computer Vision, Revised Selected Papers (PART 2 ed., pp. 268-281). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6493 LNCS, No. PART 2). https://doi.org/10.1007/978-3-642-19309-5_21
Chi, Yu Tseh ; Ho, Jeffrey ; Yang, Ming Hsuan. / A direct method for estimating planar projective transform. Computer Vision, ACCV 2010 - 10th Asian Conference on Computer Vision, Revised Selected Papers. PART 2. ed. 2011. pp. 268-281 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2).
@inproceedings{646e16f34ac64307ba12c597274daf24,
title = "A direct method for estimating planar projective transform",
abstract = "Estimating planar projective transform (homography) from a pair of images is a classical problem in computer vision. In this paper, we propose a novel algorithm for direct registering two point sets in using projective transform without using intensity values. In this very general context, there is no easily established correspondences that can be used to estimate the projective transform, and most of the existing techniques become either inadequate or inappropriate. While the planar projective transforms form an eight-dimensional Lie group, we show that for registering 2D point sets, the search space for the homographies can be effectively reduced to a three-dimensional space. To further improve on the running time without significantly reducing the accuracy of the registration, we propose a matching cost function constructed using local polynomial moments of the point sets and a coarse to fine approach. The resulting registration algorithm has linear time complexity with respect to the number of input points. We have validated the algorithm using points sets collected from real images. Preliminary experimental results are encouraging and they show that the proposed method is both efficient and accurate.",
author = "Chi, {Yu Tseh} and Jeffrey Ho and Yang, {Ming Hsuan}",
year = "2011",
month = "3",
day = "16",
doi = "10.1007/978-3-642-19309-5_21",
language = "English",
isbn = "9783642193088",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
number = "PART 2",
pages = "268--281",
booktitle = "Computer Vision, ACCV 2010 - 10th Asian Conference on Computer Vision, Revised Selected Papers",
edition = "PART 2",

}

Chi, YT, Ho, J & Yang, MH 2011, A direct method for estimating planar projective transform. in Computer Vision, ACCV 2010 - 10th Asian Conference on Computer Vision, Revised Selected Papers. PART 2 edn, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 2, vol. 6493 LNCS, pp. 268-281, 10th Asian Conference on Computer Vision, ACCV 2010, Queenstown, New Zealand, 10/11/8. https://doi.org/10.1007/978-3-642-19309-5_21

A direct method for estimating planar projective transform. / Chi, Yu Tseh; Ho, Jeffrey; Yang, Ming Hsuan.

Computer Vision, ACCV 2010 - 10th Asian Conference on Computer Vision, Revised Selected Papers. PART 2. ed. 2011. p. 268-281 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6493 LNCS, No. PART 2).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - A direct method for estimating planar projective transform

AU - Chi, Yu Tseh

AU - Ho, Jeffrey

AU - Yang, Ming Hsuan

PY - 2011/3/16

Y1 - 2011/3/16

N2 - Estimating planar projective transform (homography) from a pair of images is a classical problem in computer vision. In this paper, we propose a novel algorithm for direct registering two point sets in using projective transform without using intensity values. In this very general context, there is no easily established correspondences that can be used to estimate the projective transform, and most of the existing techniques become either inadequate or inappropriate. While the planar projective transforms form an eight-dimensional Lie group, we show that for registering 2D point sets, the search space for the homographies can be effectively reduced to a three-dimensional space. To further improve on the running time without significantly reducing the accuracy of the registration, we propose a matching cost function constructed using local polynomial moments of the point sets and a coarse to fine approach. The resulting registration algorithm has linear time complexity with respect to the number of input points. We have validated the algorithm using points sets collected from real images. Preliminary experimental results are encouraging and they show that the proposed method is both efficient and accurate.

AB - Estimating planar projective transform (homography) from a pair of images is a classical problem in computer vision. In this paper, we propose a novel algorithm for direct registering two point sets in using projective transform without using intensity values. In this very general context, there is no easily established correspondences that can be used to estimate the projective transform, and most of the existing techniques become either inadequate or inappropriate. While the planar projective transforms form an eight-dimensional Lie group, we show that for registering 2D point sets, the search space for the homographies can be effectively reduced to a three-dimensional space. To further improve on the running time without significantly reducing the accuracy of the registration, we propose a matching cost function constructed using local polynomial moments of the point sets and a coarse to fine approach. The resulting registration algorithm has linear time complexity with respect to the number of input points. We have validated the algorithm using points sets collected from real images. Preliminary experimental results are encouraging and they show that the proposed method is both efficient and accurate.

UR - http://www.scopus.com/inward/record.url?scp=79952519861&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79952519861&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-19309-5_21

DO - 10.1007/978-3-642-19309-5_21

M3 - Conference contribution

AN - SCOPUS:79952519861

SN - 9783642193088

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 268

EP - 281

BT - Computer Vision, ACCV 2010 - 10th Asian Conference on Computer Vision, Revised Selected Papers

ER -

Chi YT, Ho J, Yang MH. A direct method for estimating planar projective transform. In Computer Vision, ACCV 2010 - 10th Asian Conference on Computer Vision, Revised Selected Papers. PART 2 ed. 2011. p. 268-281. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2). https://doi.org/10.1007/978-3-642-19309-5_21