An enhanced Bspline based Zernike Moment evaluations with watermarking applications

Zernike Moments (ZM), are extensively used in pattern recognition and image classification. However, they suffer from severe computational errors especially as the moment order increases. In this paper, in order to reduce the effects of these errors and make ZM suitable for image processing applications, it is proposed to compute ZM for a decimated image, and use Bspline interpolation to interpolate between image pixels. This significantly improves performance, due to its superior low frequency concentration feature and robust performance. Further improvements are also possible, by basing our computations on least squares Bspline decimated images, as well as optimizing some of ZM coefficients. It is also proposed to use the rotation invariant feature of ZM, in modifying a watermarking schemes to be insensitive to rotations or rotation attacks. In this respect, a watermarking technique based on using Natural Preserving Transform NPT, known for its robustness against cropping, compression and noise attacks is modified to be insensitive to rotation attacks. Several simulation experiments are given to verify these results.