Robust digital image-in-image watermarking algorithm using the fast Hadamard transform

Abstract

In this paper, we propose a robust image-in-image watermarking algorithm based on the fast Hadamard transform (FHT) for the copyright protection of digital images. Most current research makes use of a normally distributed random vector as a watermark and where the watermark can only be detected by cross-correlating the received coefficients with the watermark generated by secret key and then comparing an experimental threshold value. However, the FHT image-in-image method involves a "blind" watermarking process that retrieves the watermark without the need for an original image present. In the proposed approach, a number of pseudorandom selected 8×8 sub-blocks of original image and a watermark image are decomposed into Hadamard coefficients. To increase the invisibility of the watermark, a visual model based on original image characteristics, such as edges and textures are incorporated to determine the watermarking strength factor. All the AC Hadamard coefficients of watermark image is scaled by the watermarking strength factor and inserted into several middle and high frequency AC components of the Hadamard coefficients from the sub-blocks of original image. To further increase the reliability of the watermarking against the common geometric distortions, such as rotation and scaling, a post-processing technique is proposed. Understanding the type of distortion provides a mean to apply a reversal of the attack on the watermarked image, enabling the restoration to the synchronization of the embedding positions. The performance of the proposed algorithm is evaluated using Stirmark. The experiment uses container image of size 512×512×8bits and the watermark image of size 64×64×8bits. It survives about 60% of all Stirmark attacks. The simplicity of Hadamard transform offers a significant advantage in shorter processing time and ease of hardware implementation than the commonly used DCT and DWT techniques.