A New Approach to Commutative Watermarking-Encryption

Roland Schmitz, Shujun Li, Christos Grecos, Xinpeng ZhangCommunications and Multimedia Security 2012 · DOI

TL;DR

This paper introduces a novel approach to Commutative Watermarking-Encryption (CWE) by combining a permutation-based cipher with a histogram-based watermarking scheme. This method preserves the global image statistics, allowing watermarking to be performed on either the encrypted or unencrypted data without compromising the watermark's integrity or the encryption's security.

What Problem Does This Paper Address?

The paper addresses the challenge of securely protecting digital multimedia content by combining encryption and watermarking in a way that allows for operations to be commutative. This means that the order of encryption and watermarking should not affect the final outcome, and importantly, watermarking should be possible even on encrypted data. Existing CWE approaches often rely on homomorphic encryption, which can limit robustness, or partial encryption, which can compromise security. This paper aims to bridge this gap by proposing a method that offers both strong security and robust watermarking.

Methods

The core of the proposed approach lies in using a permutation cipher for encryption. Unlike traditional ciphers that alter pixel values, a permutation cipher merely shuffles pixel locations. This crucial detail ensures that the global statistics of the image, such as its histogram, remain intact. This preservation of global statistics is key because many watermarking techniques, particularly histogram-based ones, rely on these properties.

The paper then integrates this with a modified histogram-based watermarking scheme. The modification addresses limitations of previous histogram-based methods, specifically their small key space and limited capacity. The new scheme uses a stream cipher to select bin pairs randomly, increasing the key space, and employs a technique of dividing the image into sub-images to enhance embedding capacity.

The commutative property is achieved because the permutation cipher, by preserving global statistics, allows the histogram-based watermarking algorithm to function correctly on both the original and encrypted data. The paper also details the encryption process using Arnold's cat map, a well-known discrete 2D chaotic map, for generating the permutation matrix.

Datasets

The paper evaluates its proposed scheme using the Kodak true-color image database. They also use test images such as "baboon" and "parrots" for specific demonstrations of encryption and watermarking. No new benchmarks or datasets were introduced.

Key Insights and Ideas

  • Permutation Ciphers as a Bridge: The key insight is that permutation ciphers, by preserving global image statistics, are ideal for CWE. This allows watermarking algorithms that rely on these statistics to operate seamlessly on encrypted data.
  • Histogram Preservation is Crucial: The paper highlights that many watermarking techniques are sensitive to changes in global image properties like histograms. By using a permutation cipher, these properties are maintained, enabling commutative watermarking.
  • Addressing Limitations of Existing CWE: The work directly tackles the robustness issues of homomorphic encryption and the security concerns of partial encryption by offering a solution that is both secure and robust.
  • Enhanced Histogram Watermarking: The modifications to the histogram-based watermarking scheme (random bin selection, sub-image division) are significant for improving key space and capacity, making it more practical.

Key Results

The paper demonstrates that the proposed CWE scheme successfully embeds and extracts watermarks from both encrypted and unencrypted images, satisfying the four properties of CWE. Experimental results show that the watermarked images maintain good visual quality, with PSNR values around 36 dB and SSIM values above 0.87 for the tested images. The modified watermarking scheme shows comparable visual quality to the original scheme but with a smaller variance, attributed to the random bin selection. The paper also provides theoretical bounds for the key space of the watermarking part, showing it to be sufficiently large for practical security.

Limitations and Future Work

The authors acknowledge that histogram-based watermarks can be vulnerable to malicious attacks that manipulate the image histogram, such as random swapping or shifting of bins. They suggest that robustness against lossy compression of encrypted images is an area for future research. Additionally, they propose studying the generalization of their CWE scheme to compressed domains and investigating reversible CWE schemes within their framework.

Summary updated

BibTeX
@inproceedings{schmitz2012new,
  author = {Roland Schmitz and Shujun Li and Christos Grecos and Xinpeng Zhang},
  title = {{A New Approach to Commutative Watermarking-Encryption}},
  booktitle = {Communications and Multimedia Security},
  year = {2012},
  doi = {10.1007/978-3-642-32805-3\_10},
}

Loading related content...

Figure