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

Multiscale fragile watermarking based on the Gaussian mixture model.

Hua Yuan1, Xiao-Ping Zhang

  • 1Department of Electrical and Computer Engineering, Ryerson University, Toronto, Canada. hyuan@ee.ryerson.ca

IEEE Transactions on Image Processing : a Publication of the IEEE Signal Processing Society
|October 7, 2006
PubMed
Summary

This study introduces a novel fragile watermarking technique using a Gaussian mixture model (GMM) for secure image authentication. The method embeds watermarks robustly, ensuring imperceptible distortion and accurate tamper detection.

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

Heat-triggered phospholipid flipping stabilizes plasma membrane fluidity.

Nature·2026
Same author

Radiographic Evaluation of Peri-Implant Bone Thickness in Fibular Mandible Reconstruction With Preoperative Occlusion-Driven Design.

The Journal of craniofacial surgery·2026
Same author

Global Snow-free Leaf Area Index Dataset 1985-2020 for Earth System Modeling.

Scientific data·2026
Same author

Orbital hybridization-mediated nanozymes reverse cellular senescence for aged bone regeneration.

Biomaterials·2026
Same author

Achieving Superior Mechanical Properties in CoCrNiCuFe High-Entropy Alloys via Synergistic Design of Composition and Gradient Nanostructure.

Langmuir : the ACS journal of surfaces and colloids·2026
Same author

The impact of clinical internship experiences on professional identity among nursing undergraduates: A qualitative study based on control-value theory.

Nurse education today·2026

Area of Science:

  • Digital Image Processing
  • Information Security
  • Statistical Modeling

Background:

  • Digital image authentication is crucial for verifying image integrity.
  • Existing fragile watermarking methods face challenges in robustness and imperceptibility.
  • Statistical models offer potential for advanced watermarking techniques.

Purpose of the Study:

  • To present a new multiscale fragile watermarking scheme.
  • To enhance image authentication and tamper detection capabilities.
  • To develop a robust and imperceptible watermarking method.

Main Methods:

  • Development of a Gaussian Mixture Model (GMM) to capture image statistical characteristics in the wavelet domain.
  • Utilizing the expectation-maximization algorithm for GMM parameter estimation.

Related Experiment Videos

  • Dividing wavelet multiscale subspaces into watermarking blocks and adjusting GMM parameters for secure embedding.
  • Implementing an optimal watermark embedding strategy for minimal distortion.
  • Designing a secret embedding key for robust watermark security.

    • Main Results:

    • Secure embedding of message bit streams (e.g., signatures, logos) as fragile watermarks.
    • Achieved imperceptible distortion due to minimal modification of image data.
    • Effective detection and localization of image tampering by spreading watermark bits across the image.
    • Distinguishing normal image operations (e.g., JPEG compression) from malicious attacks, enabling semi-fragile watermarking.

    Conclusions:

    • The proposed GMM-based fragile watermarking scheme provides a secure and robust solution for image authentication.
    • The method offers superior imperceptibility and tamper detection capabilities compared to conventional techniques.
    • The multiscale implementation enhances the ability to differentiate between benign and malicious image modifications.