A robust watermarking scheme against cut-and-paste attacks based on polar harmonic transform and differential grid verification

Abstract

Digital watermarking technology is a reliable method of copyright protection. However, with the continuous advancement of multimedia technology, a new type of cut-and-paste attack increasingly threatens the copyright security of images. To solve this problem, this paper proposes a robust watermark self-restoration scheme based on polar harmonic transform (PHT) and differential grid verification (DGV) to enhance the robustness of watermarking under cut-and-paste attacks. Specifically, standard distortion-compensated quantization index modulation (DC-QIM) is used to embed the watermark into PHT moments, which ensures the robustness of the watermark against common attacks. Two watermark extraction strategies, regular extraction and sliding window extraction, are designed to provide the watermark with strong extraction capability even when it suffers from translation and cut-and-paste attacks. In order to make the watermark with self-restoration ability, the proposed scheme adopts m-sequence to encode the watermark and realizes the self-restoration of the watermark by m-sequence based DGV during the extraction process. The experimental results show that the watermarking scheme maintains good imperceptibility while significantly outperforming similar schemes in terms of robustness against most types of attacks, especially cropping and translation attacks, and also has the ability to resist cut-and-paste attacks.