A Saliency-based Framework for 2D-3D Registration

Mark Brown, Jean-Yves Guillemaut, David Windridge

2014

Abstract

Here we propose a saliency-based filtering approach to the problem of registering an untextured 3D object to a single monocular image. The principle of saliency can be applied to a range of modalities and domains to find intrinsically descriptive entities from amongst detected entities, making it a rigorous approach to multi-modal registration. We build on the Kadir-Brady saliency framework due to its principled information-theoretic approach which enables us to naturally extend it to the 3D domain. The salient points from each domain are initially aligned using the SoftPosit algorithm. This is subsequently refined by aligning the silhouette with contours extracted from the image. Whereas other point based registration algorithms focus on corners or straight lines, our saliency-based approach is more general as it is more widely applicable e.g. to curved surfaces where a corner detector would fail. We compare our salient point detector to the Harris corner and SIFT keypoint detectors and show it generally achieves superior registration accuracy.

Paper Citation

in Harvard Style

Brown M., Guillemaut J. and Windridge D. (2014). A Saliency-based Framework for 2D-3D Registration . In Proceedings of the 9th International Conference on Computer Vision Theory and Applications - Volume 1: VISAPP, (VISIGRAPP 2014) ISBN 978-989-758-003-1, pages 265-273. DOI: 10.5220/0004675402650273

in Bibtex Style

@conference{visapp14,
author={Mark Brown and Jean-Yves Guillemaut and David Windridge},
title={A Saliency-based Framework for 2D-3D Registration},
booktitle={Proceedings of the 9th International Conference on Computer Vision Theory and Applications - Volume 1: VISAPP, (VISIGRAPP 2014)},
year={2014},
pages={265-273},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0004675402650273},
isbn={978-989-758-003-1},
}

in EndNote Style

TY - CONF
JO - Proceedings of the 9th International Conference on Computer Vision Theory and Applications - Volume 1: VISAPP, (VISIGRAPP 2014)
TI - A Saliency-based Framework for 2D-3D Registration
SN - 978-989-758-003-1
AU - Brown M.
AU - Guillemaut J.
AU - Windridge D.
PY - 2014
SP - 265
EP - 273
DO - 10.5220/0004675402650273