Media Hash-Dependent Image Watermarking Resilient Against Both Geometric Attacks and Estimation Attacks Based on False Positive-Oriented Detection
Watermarking methods usually claim a certain degree of robustness against those attacks that aim to destroy the hidden watermark at the expense of degrading the quality of media data. However, there exist watermark-estimation attacks (WEAs), such as the collusion attack and copy attack that are clever at disclosing the hidden watermark for unauthorized purposes while maintaining media’s quality. The aim of this study was to deal with the WEA problem. We begin by gaining insight into WEA, leading to formal definitions of optimal watermark estimation and perfect cover data recovery. Subject to these definitions, the content-dependent watermark (CDW) is proposed to resist watermarkestimation attacks. The key is to introduce a media hash as a constituent component of the CDW. Mathematical analyses and experiment results consistently verify the effectiveness of the content-dependent watermarking scheme. To our knowledge, this anti-disclosure watermarking is the first work that takes resistance to both collusion and copy attacks into consideration.