Marker Detection for Augmented Reality Applications

Martin Hirzer

Research output: Other contributionResearch

Abstract

In this paper we present a fast marker detection front end for Augmented Reality (AR) applications. The proposed algorithm is inspired by the ARTag system and designed to be robust against changing illumination conditions and occlusion of markers. In order to achieve this robustness we use an edge based approach. Edge pixels found by an edge detector are linked into lines by a RANSAC-grouper. These lines in turn are grouped into quadrangles. By detecting edge pixels only on a very coarse sampling grid the runtime of our algorithm is reduced significantly, so that we attain real time performance. Several experiments have been conducted on various images and video sequences. The obtained results demonstrate that our marker detection front end is fast and robust in case of changing lighting conditions and occlusions.
Original languageEnglish
TypeTechnical Report (ICG-TR-08/05)
Publication statusPublished - 27 Oct 2008

Fingerprint

Augmented reality
Lighting
Pixels
Sampling
Detectors
Experiments

Cite this

Marker Detection for Augmented Reality Applications. / Hirzer, Martin.

2008, Technical Report (ICG-TR-08/05).

Research output: Other contributionResearch

@misc{12d6848c177e4bc98c9a13bc5387c7e5,
title = "Marker Detection for Augmented Reality Applications",
abstract = "In this paper we present a fast marker detection front end for Augmented Reality (AR) applications. The proposed algorithm is inspired by the ARTag system and designed to be robust against changing illumination conditions and occlusion of markers. In order to achieve this robustness we use an edge based approach. Edge pixels found by an edge detector are linked into lines by a RANSAC-grouper. These lines in turn are grouped into quadrangles. By detecting edge pixels only on a very coarse sampling grid the runtime of our algorithm is reduced significantly, so that we attain real time performance. Several experiments have been conducted on various images and video sequences. The obtained results demonstrate that our marker detection front end is fast and robust in case of changing lighting conditions and occlusions.",
author = "Martin Hirzer",
year = "2008",
month = "10",
day = "27",
language = "English",
type = "Other",

}

TY - GEN

T1 - Marker Detection for Augmented Reality Applications

AU - Hirzer, Martin

PY - 2008/10/27

Y1 - 2008/10/27

N2 - In this paper we present a fast marker detection front end for Augmented Reality (AR) applications. The proposed algorithm is inspired by the ARTag system and designed to be robust against changing illumination conditions and occlusion of markers. In order to achieve this robustness we use an edge based approach. Edge pixels found by an edge detector are linked into lines by a RANSAC-grouper. These lines in turn are grouped into quadrangles. By detecting edge pixels only on a very coarse sampling grid the runtime of our algorithm is reduced significantly, so that we attain real time performance. Several experiments have been conducted on various images and video sequences. The obtained results demonstrate that our marker detection front end is fast and robust in case of changing lighting conditions and occlusions.

AB - In this paper we present a fast marker detection front end for Augmented Reality (AR) applications. The proposed algorithm is inspired by the ARTag system and designed to be robust against changing illumination conditions and occlusion of markers. In order to achieve this robustness we use an edge based approach. Edge pixels found by an edge detector are linked into lines by a RANSAC-grouper. These lines in turn are grouped into quadrangles. By detecting edge pixels only on a very coarse sampling grid the runtime of our algorithm is reduced significantly, so that we attain real time performance. Several experiments have been conducted on various images and video sequences. The obtained results demonstrate that our marker detection front end is fast and robust in case of changing lighting conditions and occlusions.

M3 - Other contribution

ER -