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

Variational optical flow computation in real time.

Andrés Bruhn1, Joachim Weickert, Christian Feddern

  • 1Mathematical Image Analysis Group, Faculty of Mathematics and Computer Science, Saarland University, 66041 Saarbrücken, Germany. bruhn@mia.uni-saarland.de

IEEE Transactions on Image Processing : a Publication of the IEEE Signal Processing Society
|May 13, 2005
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

Connecting image inpainting with denoising in the homogeneous diffusion setting.

Advances in continuous and discrete models·2025
Same author

Triaging mammography with artificial intelligence: an implementation study.

Breast cancer research and treatment·2025
Same author

Regularised Diffusion-Shock Inpainting.

Journal of mathematical imaging and vision·2024
Same author

Quantum State Assignment Flows.

Entropy (Basel, Switzerland)·2023
Same author

Learning system parameters from turing patterns.

Machine learning·2023
Same author

Enhancing the reliability and accuracy of AI-enabled diagnosis via complementarity-driven deferral to clinicians.

Nature medicine·2023
Same journal

Change-Prior-Guided Unsupervised Change Detection of Heterogeneous Remote Sensing Images.

IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2026
Same journal

AgonicDreamer: Enhancing Multi-View Consistency in Text-to-3D Generation via Rectified Score Distillation.

IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2026
Same journal

BiCM-Prompt: Bidirectional Cross-Modal Prompt Tuning for Class-Incremental Learning on Multisource Remote Sensing Images.

IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2026
Same journal

GoP-based Quality Enhancement on Video Compression.

IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2026
Same journal

Align then Tensorize: Multi-Level Consistent Anchor Graph Learning for Scalable Multi-View Clustering.

IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2026
Same journal

Beyond Fidelity: Diverse Image Synthesis via Retrieval-Augmented Diffusion.

IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2026
See all related articles

Bidirectional multigrid methods offer significant speedups for variational optical flow calculations. This approach enables real-time dense flow field computation on standard PCs, outperforming traditional solvers.

Area of Science:

  • Computer Vision
  • Numerical Analysis
  • Image Processing

Background:

  • Variational optical flow methods are crucial for motion estimation in computer vision.
  • Existing numerical schemes for optical flow are often computationally intensive, limiting real-time applications.
  • Bidirectional multigrid methods are known for their efficiency in solving equation systems but are underutilized in computer vision.

Purpose of the Study:

  • To investigate the applicability and efficiency of bidirectional multigrid methods for variational optical flow.
  • To demonstrate that these methods can achieve real-time performance for optical flow computations.
  • To adapt and implement bidirectional multigrid techniques for a combined local-global variational optical flow method.

Main Methods:

Related Experiment Videos

  • Implementation of bidirectional multigrid algorithms tailored for variational optical flow formulations.
  • Development of multigrid schemes utilizing nondyadic grid hierarchies via intergrid transfer operators.
  • Comparison of decoupled and coupled Gauss-Seidel solvers within the multigrid framework.
  • Adaptation of a noise-robust combined local-global optical flow method.

    • Main Results:

    • Achieved real-time performance for dense optical flow computation on standard PCs.
    • Demonstrated superior speed compared to unidirectional multigrid methods and nonhierarchical solvers.
    • Computed dense flow fields for the Yosemite sequence at 18 frames per second, a nearly 1000-fold improvement over Gauss-Seidel.

    Conclusions:

    • Bidirectional multigrid methods are highly effective for accelerating variational optical flow computations.
    • The proposed approach overcomes limitations of image size and grid hierarchy in multigrid methods.
    • This research paves the way for efficient, real-time dense optical flow estimation in computer vision applications.