A Reversible and Rotational-Invariant Watermarking Scheme Using Polar Harmonic Transforms

Polar harmonic transformation (PHT) is a well-known family of rotation and scaling image-invariant descriptors. The PHT is frequently used in the watermarking applications. The watermark bits of the image are mixed into the image moments. The modified moments are then used to reconstruct the watermarked image. The ability of the image embedded with the watermark bits to resist an attack depends on the strength of the image moments to resist the attack. Therefore, the accuracy of the moments determines the robustness of the watermarking algorithm. An improved method based on the combination of analytical and numerical techniques is used for computing the image moments. The more accurate moments can be then used for the watermarking uses. The experimental results show that the improved moments have the ability to resist rotational attacks and are completely reversible. A comparative study also suggest that the improved moments have better robustness for rotational attacks than some of the existing image moment-based watermarking techniques.