Riemannian Filters for Multi-variate Mesh Signals

Teodor Cioaca, Bogdan Dumitrescu, Mihai Sorin Stupariu

2017

Abstract

Designing filters over irregular non-Euclidean domains requires algorithms that take into account the intrinsic curvature of these domains. We propose a new filtering method based on Riemannian weighted averages. The resulting filters are non-Euclidean adaptations of the mean shift and blurring mean shift algorithms. We also introduce a hybrid, efficient computing strategy by combining these iterative filtering methods with wavelet multi-resolution editing. The applications of our filters include multi-variate mesh data smoothing, denoising, attribute enhancement and curvature filtering.

Paper Citation

in Harvard Style

Cioaca T., Dumitrescu B. and Stupariu M. (2017). Riemannian Filters for Multi-variate Mesh Signals . In Proceedings of the 12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 1: GRAPP, (VISIGRAPP 2017) ISBN 978-989-758-224-0, pages 228-235. DOI: 10.5220/0006128602280235

in Bibtex Style

@conference{grapp17,
author={Teodor Cioaca and Bogdan Dumitrescu and Mihai Sorin Stupariu},
title={Riemannian Filters for Multi-variate Mesh Signals},
booktitle={Proceedings of the 12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 1: GRAPP, (VISIGRAPP 2017)},
year={2017},
pages={228-235},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0006128602280235},
isbn={978-989-758-224-0},
}

in EndNote Style

TY - CONF
JO - Proceedings of the 12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 1: GRAPP, (VISIGRAPP 2017)
TI - Riemannian Filters for Multi-variate Mesh Signals
SN - 978-989-758-224-0
AU - Cioaca T.
AU - Dumitrescu B.
AU - Stupariu M.
PY - 2017
SP - 228
EP - 235
DO - 10.5220/0006128602280235