Mesh median filter for smoothing 3-D polygonal surfaces

H. Yagou, A. Belyaev, Daming Wei

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

12 Citations (Scopus)

Abstract

In this paper we introduce a new mesh filtering method: a mesh median filter. This is an application of the median filter to smoothen 3-D noisy shapes given by triangle meshes. An algorithm of the mesh median filter is realized by applying the median filter to face normals on triangle meshes and updating mesh vertex positions to make them fit to the filtered normals. As an advanced modification of the mesh median filter we further introduce a weighted mesh median filter. The weighted mesh median filter has a reinforced feature preservation effect. The weighted mesh median filter with positive weighting has the smoothing effect, and the one with negative weighting has the enhancing effect. The two kinds of mesh median filters are compared with two conventional mesh filtering methods: the Laplacian smoothing flow and the mean curvature flow. Experimental results demonstrate that the mesh median filter does not induce oversmoothing.

Original languageEnglish
Title of host publication1st International Symposium on Cyber Worlds 2002
EditorsShietung Peng, Shuichi Yukita, Vladimir V. Savchenko
PublisherIEEE
Pages488-495
Number of pages8
ISBN (Electronic)9780769518626
DOIs
Publication statusPublished - 25 Feb 2003
Event1st International Symposium on Cyber Worlds 2002 - Tokyo, Japan
Duration: 6 Nov 20028 Nov 2002

Conference

Conference1st International Symposium on Cyber Worlds 2002
Abbreviated titleCW 2002
Country/TerritoryJapan
CityTokyo
Period6/11/028/11/02

Keywords

  • 3-D mesh smoothing
  • median filter
  • triangle meshes
  • triangulated surfaces
  • weighted median filter

ASJC Scopus subject areas

  • Computational Theory and Mathematics
  • Computer Graphics and Computer-Aided Design
  • Artificial Intelligence

Fingerprint

Dive into the research topics of 'Mesh median filter for smoothing 3-D polygonal surfaces'. Together they form a unique fingerprint.

Cite this