Solving Dense Image Matching in Real-Time using Discrete-Continuous Optimization

Oleksandr Shekhovtsov, Christian Reinbacher, Gottfried Graber, Thomas Pock

Research output: Contribution to conferencePaperResearchpeer-review

Abstract

Dense image matching is a fundamental low-level problem in Computer Vision, which has received tremendous attention from both discrete and continuous optimization communities. The goal of this paper is to combine the advantages of discrete and continuous optimization in a coherent framework. We devise a model based on energy minimization, to be optimized by both discrete and continuous algorithms in a consistent way. In the discrete setting, we propose a novel optimization algorithm that can be massively parallelized. In the continuous setting we tackle the problem of non-convex regularizers by a formulation based on differences of convex functions. The resulting hybrid discrete-continuous algorithm can be efficiently accelerated by modern GPUs and we demonstrate its real-time performance for the applications of dense stereo matching and optical flow.
Original languageEnglish
Publication statusPublished - 23 Jan 2016
EventComputer Vision Winter Workshop - Lasko, Slovenia
Duration: 3 Feb 20165 Feb 2016

Conference

ConferenceComputer Vision Winter Workshop
Abbreviated titleCVWW 2016
CountrySlovenia
CityLasko
Period3/02/165/02/16

Fingerprint

Image matching
Optical flows
Computer vision

Keywords

  • cs.CV

Cite this

Shekhovtsov, O., Reinbacher, C., Graber, G., & Pock, T. (2016). Solving Dense Image Matching in Real-Time using Discrete-Continuous Optimization. Paper presented at Computer Vision Winter Workshop, Lasko, Slovenia.

Solving Dense Image Matching in Real-Time using Discrete-Continuous Optimization. / Shekhovtsov, Oleksandr; Reinbacher, Christian; Graber, Gottfried; Pock, Thomas.

2016. Paper presented at Computer Vision Winter Workshop, Lasko, Slovenia.

Research output: Contribution to conferencePaperResearchpeer-review

Shekhovtsov, O, Reinbacher, C, Graber, G & Pock, T 2016, 'Solving Dense Image Matching in Real-Time using Discrete-Continuous Optimization' Paper presented at Computer Vision Winter Workshop, Lasko, Slovenia, 3/02/16 - 5/02/16, .
Shekhovtsov O, Reinbacher C, Graber G, Pock T. Solving Dense Image Matching in Real-Time using Discrete-Continuous Optimization. 2016. Paper presented at Computer Vision Winter Workshop, Lasko, Slovenia.
Shekhovtsov, Oleksandr ; Reinbacher, Christian ; Graber, Gottfried ; Pock, Thomas. / Solving Dense Image Matching in Real-Time using Discrete-Continuous Optimization. Paper presented at Computer Vision Winter Workshop, Lasko, Slovenia.
@conference{bded7ac9ae7747bd8f2cb6255597e451,
title = "Solving Dense Image Matching in Real-Time using Discrete-Continuous Optimization",
abstract = "Dense image matching is a fundamental low-level problem in Computer Vision, which has received tremendous attention from both discrete and continuous optimization communities. The goal of this paper is to combine the advantages of discrete and continuous optimization in a coherent framework. We devise a model based on energy minimization, to be optimized by both discrete and continuous algorithms in a consistent way. In the discrete setting, we propose a novel optimization algorithm that can be massively parallelized. In the continuous setting we tackle the problem of non-convex regularizers by a formulation based on differences of convex functions. The resulting hybrid discrete-continuous algorithm can be efficiently accelerated by modern GPUs and we demonstrate its real-time performance for the applications of dense stereo matching and optical flow.",
keywords = "cs.CV",
author = "Oleksandr Shekhovtsov and Christian Reinbacher and Gottfried Graber and Thomas Pock",
year = "2016",
month = "1",
day = "23",
language = "English",
note = "Computer Vision Winter Workshop, CVWW 2016 ; Conference date: 03-02-2016 Through 05-02-2016",

}

TY - CONF

T1 - Solving Dense Image Matching in Real-Time using Discrete-Continuous Optimization

AU - Shekhovtsov, Oleksandr

AU - Reinbacher, Christian

AU - Graber, Gottfried

AU - Pock, Thomas

PY - 2016/1/23

Y1 - 2016/1/23

N2 - Dense image matching is a fundamental low-level problem in Computer Vision, which has received tremendous attention from both discrete and continuous optimization communities. The goal of this paper is to combine the advantages of discrete and continuous optimization in a coherent framework. We devise a model based on energy minimization, to be optimized by both discrete and continuous algorithms in a consistent way. In the discrete setting, we propose a novel optimization algorithm that can be massively parallelized. In the continuous setting we tackle the problem of non-convex regularizers by a formulation based on differences of convex functions. The resulting hybrid discrete-continuous algorithm can be efficiently accelerated by modern GPUs and we demonstrate its real-time performance for the applications of dense stereo matching and optical flow.

AB - Dense image matching is a fundamental low-level problem in Computer Vision, which has received tremendous attention from both discrete and continuous optimization communities. The goal of this paper is to combine the advantages of discrete and continuous optimization in a coherent framework. We devise a model based on energy minimization, to be optimized by both discrete and continuous algorithms in a consistent way. In the discrete setting, we propose a novel optimization algorithm that can be massively parallelized. In the continuous setting we tackle the problem of non-convex regularizers by a formulation based on differences of convex functions. The resulting hybrid discrete-continuous algorithm can be efficiently accelerated by modern GPUs and we demonstrate its real-time performance for the applications of dense stereo matching and optical flow.

KW - cs.CV

M3 - Paper

ER -