A Novel Robust Reversible Watermarking Method Against JPEG Compression

Abstract

Robustness limits the application of reversible watermarking. To overcome this weakness, several robust reversible watermarking (RRW) techniques have been proposed so far. However, most existing RRW methods are unstable in terms of robustness and reversibility. Then, in this paper, to ameliorate these disadvantages, a new RRW algorithm is presented. A cover image is first divided into non-overlapping blocks, and a high pass filter is applied to each block to generate a histogram which is a Laplacien-like distribution. Then, the watermark is embedded into the blocks by shifting the generated histogram. Specifically, the histogram shifting is conducted by modifying each pixel in the block, and the embedding distortion is minimized based on a new modification mechanism. Moreover, for blind data extraction and image recovery, a strategy for determining the parameters used in histogram shifting is also proposed. In this way, more than 4,096 bits can be reversibly embedded into a cover image with a good visual quality and sufficient robustness against JPEG compression. The superiority of the proposed method is verified through extensive experiments.