Two-Layer Image Watermarking Without Resorting to the Original Image

Abstract

A two-layer image watermarking without resorting to original image is proposed in this study. Two layers of watermarking algorithms are employed to hide the same watermarks in the spatial domain of an image simultaneously. More specifically, the watermarking in layer one can resist high-frequency destruction, while that in layer two can resist low-frequency destruction. Although the image is modified through two layers of watermarking, the watermarks are still invisible. In addition to robustness and invisibility, the proposed watermarking has other advantages as listed below. First, the proposed embedding technique is based on intra relative within the original image rather than inter relative between the original and watermarked images such that the original image is not required during detection process. Second, the proposed watermark is composed of fixed and variable parts. The former leads that the location and the existence of watermark can be verified directly without referring to the original watermark. The latter is involved to increase flexibility and variety of watermarks. Third, the watermark is a short serial number such that it can be duplicated. Due to the duplicity, the majority voting strategy can be employed to facilitate watermark detection. Finally, the proposed method is simple and fast. It takes only one to two seconds either in embedding stage or in detection stage. Various experiments have been conducted to prove the advantages of the two-layer watermarking such as robustness, invisibility and practicability.