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

Stereo using monocular cues within the tensor voting framework.

Philippos Mordohai1, Gérard Medioni

  • 1Department of Computer Science, University of North Carolina, Chapel Hill, NC 27599-3715, USA. mordohai@cs.unc.edu

IEEE Transactions on Pattern Analysis and Machine Intelligence
|May 27, 2006
PubMed
Summary

This study introduces a novel stereo matching method using perceptual organization to overcome occlusion and texture challenges. It accurately segments surfaces and refines disparity for improved image matching results.

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

Glissando-Net: Deep Single View Category Level Pose Estimation and 3D Reconstruction.

IEEE transactions on pattern analysis and machine intelligence·2025
Same author

Inlier Clustering based on the Residuals of Random Hypotheses.

Pattern recognition letters·2021
Same author

On the Synergies Between Machine Learning and Binocular Stereo for Depth Estimation From Images: A Survey.

IEEE transactions on pattern analysis and machine intelligence·2021
Same author

Learning Pose-Aware Models for Pose-Invariant Face Recognition in the Wild.

IEEE transactions on pattern analysis and machine intelligence·2018
Same author

Facial Landmark Detection with Tweaked Convolutional Neural Networks.

IEEE transactions on pattern analysis and machine intelligence·2018
Same author

Structured Time Series Analysis for Human Action Segmentation and Recognition.

IEEE transactions on pattern analysis and machine intelligence·2015

Area of Science:

  • Computer Vision
  • Image Processing
  • Computational Geometry

Background:

  • Stereo matching is crucial for 3D reconstruction but faces challenges with occlusions and textureless regions.
  • Existing methods often struggle with local ambiguities and scanline-dependent artifacts.

Purpose of the Study:

  • To develop a robust stereo matching algorithm addressing occlusion and lack of texture.
  • To improve surface segmentation and disparity estimation accuracy in static images.

Main Methods:

  • Utilizing a perceptual organization framework with binocular and monocular cues.
  • Generating matching candidates and embedding them in disparity space for 3D neighborhood analysis.
  • Achieving surface segmentation based on geometric properties and correcting occlusions via color consistency checks.

Related Experiment Videos

  • Employing tensor voting for propagating information and refining final disparity hypotheses.

    • Main Results:

    • Demonstrated accurate surface segmentation based on geometric properties, not just photometric information.
    • Successfully corrected surface overextensions caused by occlusions.
    • Improved disparity estimation for previously unmatched pixels through iterative refinement.
    • Presented effective results on widely used benchmark stereo pairs.

    Conclusions:

    • The proposed perceptual organization framework offers a robust solution for stereo matching challenges.
    • The method effectively segments surfaces and refines disparity, leading to enhanced 3D scene understanding.
    • This approach provides a significant advancement in handling occlusions and textureless areas in stereo vision.