A Novel Contourlet Domain Watermark Detector for Copyright Protection

TL;DR

This paper introduces a new contourlet domain watermark detection method using a 2D-GARCH statistical model, demonstrating high efficiency and robustness for copyright protection in digital media.

Contribution

It proposes a novel contourlet domain watermark detector based on 2D-GARCH modeling, improving detection accuracy and robustness against attacks.

Findings

High detection efficiency confirmed by experiments

Effective modeling of contourlet coefficients with 2D-GARCH

Robustness demonstrated under various attack scenarios

Abstract

Digital media can be distributed via Internet easily, so, media owners are eagerly seeking methods to protect their rights. A typical solution is digital watermarking for copyright protection. In this paper, we propose a novel contourlet domain image watermarking scheme for copyright protection. In the embedding phase, we insert the watermark into the image using an additive contourlet domain spread spectrum approach. In the detection phase, we design a detector using likelihood ratio test (LRT). Since the performance of the LRT detector is completely dependent on the accuracy of the employed statistical model, we first study the statistical properties of the contourlet coefficients. This study demonstrates the heteroscedasticity and heavy-tailed marginal distribution of these coefficients. Therefore, we propose using two dimensional generalized autoregressive conditional heteroscedasticity (2D-GARCH) model that is compatible with the contourlet coefficients. Motivated by the modeling results, we design a new watermark detector based on 2D-GARCH model. Also, we analyze its performance by computing the receiver operating characteristics. Experimental results confirm the high efficiency of the proposed detector. Since a watermark detector for copyright protection should be robust against attacks, we examine the robustness of the proposed detector under different kinds of attacks.