Cryptanalysis of Two Correlation-Based Digital Watermarking Schemes in DCT Domain

Digital watermarking frameworks are mostly based on correlation between the original watermark and the information retrieved from the watermarked copy. The techniques always try to sustain this correlation even after different attacks. In this article, we will revisit two schemes from the same classes (Lin et al. 2010, as well as the supposedly improved version by Das et al. 2017) and demonstrate how the secret watermark could be damaged to dilute the ownership. Our experiments question the robustness of a broad class of key-based watermarking schemes. Further, considering the more recent scheme (Das et al. 2017), we also demonstrate that some information about the secret key can also be extracted.