Introducing the separable moments for image watermarking in a Tottaly moment-oriented framework

Abstract

The proposed method constitutes the first attempt of designing a totally moment-oriented scheme that manages to accomplish the basic watermarking operations via orthogonal moment functions. A recently introduced transformation called separable moments that combines the discrete orthogonal Tchebichef and Krawtchouk polynomials in order to produce an enhanced moment kernel is firstly applied to image watermarking field regarding to their robust performance evaluation. Besides the watermark information embedding via separable moments, the original Tchebichef and Krawtchouk moment families are further applied for geometric distortion estimation and image quality assessment, respectively. The Qk metric manages to assess locally the quality of the watermarked image based on blockwise Krawtchouk moments in contrast to the traditional Peak Signal to Noise Ratio (PSNR) that takes into account the pixel differences. The proposed moment-based image watermarking scheme stays robust under the majority of common signal processing and geometric attacking conditions. Moreover, the proposed transformation manages to overcome even the original moment families occasionally.