Related Experiment Video
Updated: Jul 7, 2026
03:31
End-To-End Deep Neural Network for Salient Object Detection in Complex Environments
Published on: December 15, 2023
DCT-domain watermarking techniques for still images: detector performance analysis and a new structure
J R Hernández1, M Amado, F Pérez-González
1Departamento de Tecnologías de las Comunicaciones, E.T.S.I. Telecomunicación, Universidad de Vigo, Vigo 36200, Pontevedra, Spain. jhernan@tsc.uvigo.es
Summary
This study introduces a Discrete Cosine Transform (DCT) watermarking technique for digital image copyright protection. It improves ownership verification by modeling DCT coefficients with generalized Gaussian distributions, outperforming traditional methods.
Area of Science:
- Digital image processing
- Information security
- Computer vision
Background:
- Digital image copyright protection is crucial.
- Existing watermarking techniques often require the original image for verification.
- Discrete Cosine Transform (DCT) is a common image processing technique, used in JPEG compression.
Purpose of the Study:
- To analyze a spread-spectrum-like DCT domain watermarking technique for copyright protection.
- To develop a watermarking method that does not require the original image for ownership verification.
- To improve the performance of watermark detection and decoding.
Main Methods:
- Applying DCT in 8x8 blocks, similar to JPEG.
- Modeling original image DCT coefficients using generalized Gaussian distributions.
- Developing watermark decoding and detection tests for ownership verification.
Main Results:
- The proposed detector structures show significant performance improvements over correlation receivers.
- Analytical expressions for error probabilities in decoding and detection were derived.
- Experimental results validated the derived analytical performance measures.
Conclusions:
- The generalized Gaussian distribution model enhances DCT watermarking performance.
- The developed technique offers robust copyright protection without needing the original image.
- This method provides improved accuracy in watermark detection and decoding.