A Logarithmic Quantization-Based Image Watermarking Using Information Entropy in the Wavelet Domain.

Area of Science:

• Digital Image Processing
• Information Security
• Computer Vision

Background:

• Conventional quantization watermarking is vulnerable to estimation and introduces perceptible distortions.
• Existing methods often neglect host signal's visual perceptual characteristics.

Purpose of the Study:

• To develop a robust and imperceptible image watermarking method using logarithmic quantization and wavelet transform.
• To improve watermark robustness by embedding data in high-entropy image regions.
• To effectively balance invisibility and robustness using an entropy masking model and distortion compensation.

Main Methods:

• Image watermarking using logarithmic quantization index modulation (LQIM) and wavelet transform.
• Embedding watermark data into image blocks with high entropy values.
• Utilizing Watson's entropy masking model and a distortion-compensated scheme.
• Modeling wavelet coefficients with generalized Gaussian distribution (GGD) for bit error probability calculation.

Main Results:

• The proposed method demonstrates superior imperceptibility and robustness compared to conventional techniques.
• Effective trade-off between invisibility and robustness achieved through optimal quantization parameter selection.
• High resilience against various attacks including JPEG compression, AWGN, Gaussian filtering, and geometric transformations.

Conclusions:

• The developed logarithmic quantization-based watermarking method offers significant advantages in terms of imperceptibility and robustness.
• The approach effectively addresses the trade-off between watermark invisibility and robustness.
• The method is validated through simulations on real images, showing strong performance against diverse attacks.