Rotation and cropping resilient data hiding with Zernike moments

Most digital data hiding schemes are sensitive to simple geometric attacks namely rotation, cropping, and scaling. Of these, rotational attacks possess the capability to not only change each pixel value but also dislocate the image pixels in a circular fashion, thus creating a synchronization issue for any linear data hiding algorithm. So far, Fourier Mellin transform (FMT) based algorithms have been considered a standard for rotation invariant data hiding. This method causes some difficulties in implementation due to unstable log-polar mapping, iterative inversion of the interpolation and aliasing effects on the FMT magnitude and phase spectra. In this paper, we propose a data hiding algorithm that is based on the Zernike moment transform (ZMT), accompanied with an odd-even quantizer based embedding scheme to defeat the effects of rotation and cropping attacks. We note that this algorithm is robust to a combination of rotation and other popular attacks. The proposed algorithm has good embedding capacity and very low induced distortion. Experimental results over a range of rotational attacks (from 0° to 360°) show a recovery rate (of the embedded bits) of 97% or greater.