Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Segmenting textured 3D surfaces using the space/frequency representation

J Krumm1, S A Shafer

  • 1Robotics Institute, Carnegie Mellon University, Pittsburgh, PA 15213.

Spatial Vision
|January 1, 1994
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Emission spectrum of nitric oxide between 5 microm and 7 microm.

Applied optics·2010
Same author

Genomic organization, 5' flanking enhancer region, and chromosomal assignment of the cell cycle gene, p55Cdc.

Molecular genetics and metabolism·1998
Same author

Automatic multiparameter fluorescence imaging for determining lymphocyte phenotype and activation status in melanoma tissue sections.

Cytometry·1996
Same author

Developmental training for the cerebral palsied.

Cerebral palsy review·1965

A new algorithm segments 3D textured surfaces by analyzing local surface normals and frontalized power spectra. This overcomes limitations of previous methods, enabling reliable segmentation of non-frontally viewed textures.

Area of Science:

  • Computer Vision
  • Image Processing
  • Computational Geometry

Background:

  • Current 3D textured surface segmentation methods fail due to assumptions of flat, frontally viewed textures.
  • Existing shape-from-texture algorithms require pre-segmented textures, creating a research deadlock.

Purpose of the Study:

  • To develop a novel algorithm for segmenting 3D textured surfaces, specifically addressing non-frontally viewed, planar, periodic textures.
  • To overcome the limitations of existing texture segmentation and shape-from-texture algorithms.

Main Methods:

  • Utilizes spectrograms (local power spectra) to compute local surface normals from image regions.
  • Computes a 'frontalized' power spectrum based on estimated surface normals.
  • Merges neighboring regions with similar frontalized power spectra using a description length criterion.

Related Experiment Videos

Main Results:

  • Successfully segments images containing non-frontally viewed, planar, periodic textures.
  • The algorithm avoids unreliable image feature detection.
  • Demonstrated effectiveness on real-world textured images.

Conclusions:

  • This work presents the first algorithm capable of segmenting 3D textured surfaces by explicitly accounting for 3D shape effects.
  • The developed method offers a significant advancement in general image understanding for 3D environments.