An Efficient and Robust Hybrid Watermarking Scheme for Text-Images

Addressing fraud and illegal use of multimedia requires the development of more powerful and robust algorithms. One of the solutions that can contribute significantly to solve such problems can be found in the use of image watermarking. In image watermarking, a watermark is introduced within an image in order to protect it against illegal use. This paper proposes a new hybrid text-image watermarking algorithm based on the singular value decomposition (SVD) and the discrete cosine transform (DCT) with linear interpolation in the embedding/extraction process. Despite a considerable number of works found to-date, the robustness in existing watermark approaches remains a major challenge worthy of further improvement. Furthermore, text-images are used as samples in this work to test our scheme under further challenges and constraints imposed by such images on the embedding techniques used. In contrast to many existing approaches, we achieved through the proposed algorithm a high robustness results against the most dangerous attack scenarios. A major contribution in this paper is found in our unique watermark extraction scheme which differs from the existing literature and takes into account three inputs including the attacked watermarked image. Finally, we discuss the results obtained for our approach under various attack scenarios.