A vote-and-verify strategy for fast spatial verification in image retrieval

Johannes L. Schönberger, True Price, Torsten Sattler, Jan Michael Frahm, Marc Pollefeys

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

  • 2 Citations

Abstract

Spatial verification is a crucial part of every image retrieval system, as it accounts for the fact that geometric feature configurations are typically ignored by the Bag-of-Words representation. Since spatial verification quickly becomes the bottleneck of the retrieval process, runtime efficiency is extremely important. At the same time, spatial verification should be able to reliably distinguish between related and unrelated images. While methods based on RANSAC’s hypothesize-and-verify framework achieve high accuracy, they are not particularly efficient. Conversely, verification approaches based on Hough voting are extremely efficient but not as accurate. In this paper, we develop a novel spatial verification approach that uses an efficient voting scheme to identify promising transformation hypotheses that are subsequently verified and refined. Through comprehensive experiments, we show that our method is able to achieve a verification accuracy similar to state-of-the-art hypothesize-and-verify approaches while providing faster runtimes than state-of-the-art voting-based methods.

LanguageEnglish (US)
Title of host publicationComputer Vision - ACCV 2016 - 13th Asian Conference on Computer Vision, Revised Selected Papers
PublisherSpringer Verlag
Pages321-337
Number of pages17
Volume10111 LNCS
ISBN (Print)9783319541808
DOIs
StatePublished - 2017
Event13th Asian Conference on Computer Vision, ACCV 2016 - Taipei, Taiwan, Province of China
Duration: Nov 20 2016Nov 24 2016

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume10111 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other13th Asian Conference on Computer Vision, ACCV 2016
CountryTaiwan, Province of China
City Taipei
Period11/20/1611/24/16

Fingerprint

Image retrieval
Vote
Image Retrieval
Verify
Voting
RANSAC
Strategy
High Accuracy
Retrieval
Configuration
Experiment
Experiments

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Schönberger, J. L., Price, T., Sattler, T., Frahm, J. M., & Pollefeys, M. (2017). A vote-and-verify strategy for fast spatial verification in image retrieval. In Computer Vision - ACCV 2016 - 13th Asian Conference on Computer Vision, Revised Selected Papers (Vol. 10111 LNCS, pp. 321-337). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10111 LNCS). Springer Verlag. DOI: 10.1007/978-3-319-54181-5_21

A vote-and-verify strategy for fast spatial verification in image retrieval. / Schönberger, Johannes L.; Price, True; Sattler, Torsten; Frahm, Jan Michael; Pollefeys, Marc.

Computer Vision - ACCV 2016 - 13th Asian Conference on Computer Vision, Revised Selected Papers. Vol. 10111 LNCS Springer Verlag, 2017. p. 321-337 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10111 LNCS).

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

Schönberger, JL, Price, T, Sattler, T, Frahm, JM & Pollefeys, M 2017, A vote-and-verify strategy for fast spatial verification in image retrieval. in Computer Vision - ACCV 2016 - 13th Asian Conference on Computer Vision, Revised Selected Papers. vol. 10111 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10111 LNCS, Springer Verlag, pp. 321-337, 13th Asian Conference on Computer Vision, ACCV 2016, Taipei, Taiwan, Province of China, 11/20/16. DOI: 10.1007/978-3-319-54181-5_21
Schönberger JL, Price T, Sattler T, Frahm JM, Pollefeys M. A vote-and-verify strategy for fast spatial verification in image retrieval. In Computer Vision - ACCV 2016 - 13th Asian Conference on Computer Vision, Revised Selected Papers. Vol. 10111 LNCS. Springer Verlag. 2017. p. 321-337. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). Available from, DOI: 10.1007/978-3-319-54181-5_21
Schönberger, Johannes L. ; Price, True ; Sattler, Torsten ; Frahm, Jan Michael ; Pollefeys, Marc. / A vote-and-verify strategy for fast spatial verification in image retrieval. Computer Vision - ACCV 2016 - 13th Asian Conference on Computer Vision, Revised Selected Papers. Vol. 10111 LNCS Springer Verlag, 2017. pp. 321-337 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{fcd2ad0805714257aed01c4a4d7e9ef5,
title = "A vote-and-verify strategy for fast spatial verification in image retrieval",
abstract = "Spatial verification is a crucial part of every image retrieval system, as it accounts for the fact that geometric feature configurations are typically ignored by the Bag-of-Words representation. Since spatial verification quickly becomes the bottleneck of the retrieval process, runtime efficiency is extremely important. At the same time, spatial verification should be able to reliably distinguish between related and unrelated images. While methods based on RANSAC’s hypothesize-and-verify framework achieve high accuracy, they are not particularly efficient. Conversely, verification approaches based on Hough voting are extremely efficient but not as accurate. In this paper, we develop a novel spatial verification approach that uses an efficient voting scheme to identify promising transformation hypotheses that are subsequently verified and refined. Through comprehensive experiments, we show that our method is able to achieve a verification accuracy similar to state-of-the-art hypothesize-and-verify approaches while providing faster runtimes than state-of-the-art voting-based methods.",
author = "Sch\{"o}nberger, {Johannes L.} and True Price and Torsten Sattler and Frahm, {Jan Michael} and Marc Pollefeys",
year = "2017",
doi = "10.1007/978-3-319-54181-5_21",
language = "English (US)",
isbn = "9783319541808",
volume = "10111 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "321--337",
booktitle = "Computer Vision - ACCV 2016 - 13th Asian Conference on Computer Vision, Revised Selected Papers",
address = "Germany",

}

TY - GEN

T1 - A vote-and-verify strategy for fast spatial verification in image retrieval

AU - Schönberger,Johannes L.

AU - Price,True

AU - Sattler,Torsten

AU - Frahm,Jan Michael

AU - Pollefeys,Marc

PY - 2017

Y1 - 2017

N2 - Spatial verification is a crucial part of every image retrieval system, as it accounts for the fact that geometric feature configurations are typically ignored by the Bag-of-Words representation. Since spatial verification quickly becomes the bottleneck of the retrieval process, runtime efficiency is extremely important. At the same time, spatial verification should be able to reliably distinguish between related and unrelated images. While methods based on RANSAC’s hypothesize-and-verify framework achieve high accuracy, they are not particularly efficient. Conversely, verification approaches based on Hough voting are extremely efficient but not as accurate. In this paper, we develop a novel spatial verification approach that uses an efficient voting scheme to identify promising transformation hypotheses that are subsequently verified and refined. Through comprehensive experiments, we show that our method is able to achieve a verification accuracy similar to state-of-the-art hypothesize-and-verify approaches while providing faster runtimes than state-of-the-art voting-based methods.

AB - Spatial verification is a crucial part of every image retrieval system, as it accounts for the fact that geometric feature configurations are typically ignored by the Bag-of-Words representation. Since spatial verification quickly becomes the bottleneck of the retrieval process, runtime efficiency is extremely important. At the same time, spatial verification should be able to reliably distinguish between related and unrelated images. While methods based on RANSAC’s hypothesize-and-verify framework achieve high accuracy, they are not particularly efficient. Conversely, verification approaches based on Hough voting are extremely efficient but not as accurate. In this paper, we develop a novel spatial verification approach that uses an efficient voting scheme to identify promising transformation hypotheses that are subsequently verified and refined. Through comprehensive experiments, we show that our method is able to achieve a verification accuracy similar to state-of-the-art hypothesize-and-verify approaches while providing faster runtimes than state-of-the-art voting-based methods.

UR - http://www.scopus.com/inward/record.url?scp=85016018827&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85016018827&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-54181-5_21

DO - 10.1007/978-3-319-54181-5_21

M3 - Conference contribution

SN - 9783319541808

VL - 10111 LNCS

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

SP - 321

EP - 337

BT - Computer Vision - ACCV 2016 - 13th Asian Conference on Computer Vision, Revised Selected Papers

PB - Springer Verlag

ER -