ICF: A Shape-based 3D Segmentation Method

Koji Kobayashi, Koichi Ito, Takafumi Aoki

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

Abstract

As a new method that can contribute to 3D shape analysis, we propose Incremental Contour Flow (ICF), which divides a 3D object into segments separated by boundary surfaces at narrow parts. ICF utilizes a property of distance transform that generates local maxima in a center of swollen part and bottleneck-like structures in a narrow part. Core of segment is defined as a local maximum layer that has a higher value than surrounding layers in a layer structure generated by converting the distance values to integers. Voxels in a surrounding layer of a core are added to the core incrementally, thereby 3D voxels are grouped to as many segments as the cores. As shown in an example to be discussed, ICF can generates combined soap bubble-like boundary surfaces from a 3D object made from merged three spheres. Since human organs tend to be distinguished by its shape in medical 3D images, ICF can be useful in the medical imaging applications.

Original languageEnglish
Title of host publicationACM SIGGRAPH 2020 Posters, SIGGRAPH 2020
PublisherAssociation for Computing Machinery
ISBN (Electronic)9781450379731
DOIs
Publication statusPublished - 2020 Aug 17
EventACM SIGGRAPH 2020 Posters - International Conference on Computer Graphics and Interactive Techniques, SIGGRAPH 2020 - Virtual, Online, United States
Duration: 2020 Aug 17 → …

Publication series

NameACM SIGGRAPH 2020 Posters, SIGGRAPH 2020

Conference

ConferenceACM SIGGRAPH 2020 Posters - International Conference on Computer Graphics and Interactive Techniques, SIGGRAPH 2020
CountryUnited States
CityVirtual, Online
Period20/8/17 → …

Keywords

  • 3D shape
  • ICF
  • distance transform
  • segmentation

ASJC Scopus subject areas

  • Software
  • Computer Graphics and Computer-Aided Design

Fingerprint Dive into the research topics of 'ICF: A Shape-based 3D Segmentation Method'. Together they form a unique fingerprint.

Cite this