A Robust Watermarking Scheme Based on Fast and Highly Accurate Pseudo-Zernike Moments

The accuracy of moments is crucial for robust moment-based watermarking methods, and it is necessary to guarantee the computational cost and efficiency. Current watermarking methods lack sufficient theoretical research on accuracy. For this reason, a robust image watermarking scheme based on Pseudo-Zernike moments is proposed. Pixel values are computed in polar coordinates, and analytic expressions are utilized to eliminate geometric and numerical errors. It avoids watermarking distortion interactions. An improved quantization approach is used for embedding to achieve good concealment. The symmetric and asymmetric properties of the basic functions are utilized to reduce the computational cost when calculating moment values. Experiments show that the scheme is robust to common image processing attacks and geometric attacks and outperforms existing image watermarking algorithms.