Robust stereo matching using probabilistic laplacian surface propagation

Seungryong Kim, Bumsub Ham, Seungchul Ryu, Seon Joo Kim, Kwanghoon Sohn

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

3 Citations (Scopus)

Abstract

This paper describes a probabilistic Laplacian surface propagation (PLSP) framework for a robust stereo matching under severe radiometric variations. We discover that a progressive scheme overcomes an inherent limitation for this task, while most conventional efforts have been focusing on designing a robust cost function. We propose the ground control surfaces (GCSs) designed as progressive unit, which alleviates the problems of conventional progressive methods and superpixel based methods, simultaneously. Moreover, we introduce a novel confidence measure for stereo pairs taken under radiometric variations based on the probability of correspondences. Specifically, the PLSP estimates the GCSs from initial sparse disparity maps using a weighted least-square. The GCSs are then propagated on a superpixel graph with a surface confidence weighting. Experimental results show that the PLSP outperforms state-of-the-art robust cost function based methods and other propagation methods for the stereo matching under radiometric variations.

Original languageEnglish
Title of host publicationComputer Vision - ACCV 2014 - 12th Asian Conference on Computer Vision, Revised Selected Papers
EditorsIan Reid, Ming-Hsuan Yang, Hideo Saito, Daniel Cremers
PublisherSpringer Verlag
Pages368-383
Number of pages16
ISBN (Electronic)9783319168647
DOIs
Publication statusPublished - 2015 Jan 1
Event12th Asian Conference on Computer Vision, ACCV 2014 - Singapore, Singapore
Duration: 2014 Nov 12014 Nov 5

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9003
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other12th Asian Conference on Computer Vision, ACCV 2014
CountrySingapore
CitySingapore
Period14/11/114/11/5

Fingerprint

Stereo Matching
Control surfaces
Propagation
Cost functions
Cost Function
Confidence Measure
Weighted Least Squares
Confidence
Weighting
Correspondence
Unit
Experimental Results
Graph in graph theory
Estimate

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Kim, S., Ham, B., Ryu, S., Kim, S. J., & Sohn, K. (2015). Robust stereo matching using probabilistic laplacian surface propagation. In I. Reid, M-H. Yang, H. Saito, & D. Cremers (Eds.), Computer Vision - ACCV 2014 - 12th Asian Conference on Computer Vision, Revised Selected Papers (pp. 368-383). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9003). Springer Verlag. https://doi.org/10.1007/978-3-319-16865-4_24
Kim, Seungryong ; Ham, Bumsub ; Ryu, Seungchul ; Kim, Seon Joo ; Sohn, Kwanghoon. / Robust stereo matching using probabilistic laplacian surface propagation. Computer Vision - ACCV 2014 - 12th Asian Conference on Computer Vision, Revised Selected Papers. editor / Ian Reid ; Ming-Hsuan Yang ; Hideo Saito ; Daniel Cremers. Springer Verlag, 2015. pp. 368-383 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{e07bb580d339433aaf75d63f5b29448d,
title = "Robust stereo matching using probabilistic laplacian surface propagation",
abstract = "This paper describes a probabilistic Laplacian surface propagation (PLSP) framework for a robust stereo matching under severe radiometric variations. We discover that a progressive scheme overcomes an inherent limitation for this task, while most conventional efforts have been focusing on designing a robust cost function. We propose the ground control surfaces (GCSs) designed as progressive unit, which alleviates the problems of conventional progressive methods and superpixel based methods, simultaneously. Moreover, we introduce a novel confidence measure for stereo pairs taken under radiometric variations based on the probability of correspondences. Specifically, the PLSP estimates the GCSs from initial sparse disparity maps using a weighted least-square. The GCSs are then propagated on a superpixel graph with a surface confidence weighting. Experimental results show that the PLSP outperforms state-of-the-art robust cost function based methods and other propagation methods for the stereo matching under radiometric variations.",
author = "Seungryong Kim and Bumsub Ham and Seungchul Ryu and Kim, {Seon Joo} and Kwanghoon Sohn",
year = "2015",
month = "1",
day = "1",
doi = "10.1007/978-3-319-16865-4_24",
language = "English",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "368--383",
editor = "Ian Reid and Ming-Hsuan Yang and Hideo Saito and Daniel Cremers",
booktitle = "Computer Vision - ACCV 2014 - 12th Asian Conference on Computer Vision, Revised Selected Papers",
address = "Germany",

}

Kim, S, Ham, B, Ryu, S, Kim, SJ & Sohn, K 2015, Robust stereo matching using probabilistic laplacian surface propagation. in I Reid, M-H Yang, H Saito & D Cremers (eds), Computer Vision - ACCV 2014 - 12th Asian Conference on Computer Vision, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9003, Springer Verlag, pp. 368-383, 12th Asian Conference on Computer Vision, ACCV 2014, Singapore, Singapore, 14/11/1. https://doi.org/10.1007/978-3-319-16865-4_24

Robust stereo matching using probabilistic laplacian surface propagation. / Kim, Seungryong; Ham, Bumsub; Ryu, Seungchul; Kim, Seon Joo; Sohn, Kwanghoon.

Computer Vision - ACCV 2014 - 12th Asian Conference on Computer Vision, Revised Selected Papers. ed. / Ian Reid; Ming-Hsuan Yang; Hideo Saito; Daniel Cremers. Springer Verlag, 2015. p. 368-383 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9003).

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

TY - GEN

T1 - Robust stereo matching using probabilistic laplacian surface propagation

AU - Kim, Seungryong

AU - Ham, Bumsub

AU - Ryu, Seungchul

AU - Kim, Seon Joo

AU - Sohn, Kwanghoon

PY - 2015/1/1

Y1 - 2015/1/1

N2 - This paper describes a probabilistic Laplacian surface propagation (PLSP) framework for a robust stereo matching under severe radiometric variations. We discover that a progressive scheme overcomes an inherent limitation for this task, while most conventional efforts have been focusing on designing a robust cost function. We propose the ground control surfaces (GCSs) designed as progressive unit, which alleviates the problems of conventional progressive methods and superpixel based methods, simultaneously. Moreover, we introduce a novel confidence measure for stereo pairs taken under radiometric variations based on the probability of correspondences. Specifically, the PLSP estimates the GCSs from initial sparse disparity maps using a weighted least-square. The GCSs are then propagated on a superpixel graph with a surface confidence weighting. Experimental results show that the PLSP outperforms state-of-the-art robust cost function based methods and other propagation methods for the stereo matching under radiometric variations.

AB - This paper describes a probabilistic Laplacian surface propagation (PLSP) framework for a robust stereo matching under severe radiometric variations. We discover that a progressive scheme overcomes an inherent limitation for this task, while most conventional efforts have been focusing on designing a robust cost function. We propose the ground control surfaces (GCSs) designed as progressive unit, which alleviates the problems of conventional progressive methods and superpixel based methods, simultaneously. Moreover, we introduce a novel confidence measure for stereo pairs taken under radiometric variations based on the probability of correspondences. Specifically, the PLSP estimates the GCSs from initial sparse disparity maps using a weighted least-square. The GCSs are then propagated on a superpixel graph with a surface confidence weighting. Experimental results show that the PLSP outperforms state-of-the-art robust cost function based methods and other propagation methods for the stereo matching under radiometric variations.

UR - http://www.scopus.com/inward/record.url?scp=84938860671&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84938860671&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-16865-4_24

DO - 10.1007/978-3-319-16865-4_24

M3 - Conference contribution

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

SP - 368

EP - 383

BT - Computer Vision - ACCV 2014 - 12th Asian Conference on Computer Vision, Revised Selected Papers

A2 - Reid, Ian

A2 - Yang, Ming-Hsuan

A2 - Saito, Hideo

A2 - Cremers, Daniel

PB - Springer Verlag

ER -

Kim S, Ham B, Ryu S, Kim SJ, Sohn K. Robust stereo matching using probabilistic laplacian surface propagation. In Reid I, Yang M-H, Saito H, Cremers D, editors, Computer Vision - ACCV 2014 - 12th Asian Conference on Computer Vision, Revised Selected Papers. Springer Verlag. 2015. p. 368-383. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-16865-4_24