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

Robust global motion estimation oriented to video object segmentation.

Bin Qi1, Mohammed Ghazal, Aishy Amer

  • 1Electrical and Computer Engineering Department, Concordia University, Montréal, Canada. b_qi@ece.concordia.ca

IEEE Transactions on Image Processing : a Publication of the IEEE Signal Processing Society
|May 17, 2008
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

Analysis of the differential components of five processed products of <i>Cuscuta chinensis</i> based on HPLC-Q Exactive Orbitrap/MS and chemometrics.

Natural product research·2026
Same author

Animal gut microbes and microbiomes in the 21st century and beyond.

Science China. Life sciences·2026
Same author

Correction: Multiclass Classification of Renal Tumor Subtypes: Addressing Diagnostic Challenges Using a Texture-Informed Deep Hybrid CNN-Transformer.

Journal of imaging informatics in medicine·2026
Same author

3D Adversarial Segmentation of Kidney-Transplant Across Multiple MRI Sequences Using Probabilistic and Anatomical Priors.

Diagnostics (Basel, Switzerland)·2026
Same author

Advanced computer-aided diagnosis for age-related macular degeneration: integrating segmentation and feature extraction for precise diagnosis.

Scientific reports·2026
Same author

AI-Driven Breast Cancer Diagnosis: A Systematic Review of Imaging Modalities, Deep Learning, and Explainability.

Cancers·2026
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

This study introduces a new hierarchical differential global motion estimation (GME) method specifically for video object segmentation. The approach enhances robustness and speed for accurate object tracking and segmentation tasks.

Area of Science:

  • Computer Vision
  • Image Processing
  • Video Analysis

Background:

  • Current global motion estimation (GME) methods primarily target video coding, often inadequately addressing the needs of video object segmentation (VOS).
  • Existing VOS methods either ignore global motion (GM) or employ unsuitable coding-oriented GME techniques.
  • A gap exists for efficient and robust GM estimation tailored for VOS applications.

Purpose of the Study:

  • To propose a novel hierarchical differential GME method optimized for video object segmentation.
  • To improve the efficiency and robustness of GM estimation in the context of VOS.
  • To develop advanced outlier rejection mechanisms for accurate motion compensation.

Main Methods:

  • A hierarchical differential GME approach specifically designed for video object segmentation.

Related Experiment Videos

  • An efficient initial estimation scheme combining three-step search and motion parameter prediction.
  • A robust estimator utilizing object information to discard local motion outliers.
  • A robust estimator for the initial frame that analyzes outlier distribution in local neighborhoods to reject errors.

    • Main Results:

    • The proposed method demonstrates superior robustness compared to existing techniques.
    • The GME method is significantly more oriented towards the requirements of video object segmentation.
    • The developed approach achieves faster processing speeds than referenced methods.
    • Both subjective and objective evaluations confirm the method's effectiveness.

    Conclusions:

    • The proposed hierarchical differential GME method offers a significant advancement for video object segmentation.
    • The method provides a more robust, accurate, and efficient solution for handling global motion in VOS.
    • Future work could explore further optimizations and applications of this GME technique in related video analysis tasks.