A dual-color watermarking algorithm based on elliptical monogenic wavelet transform and singular value decomposition

A dual-color watermarking algorithm using elliptical monogenic wavelet transform and singular value decomposition is proposed. Firstly, the original host image and two color watermark images are decomposed into RGB components. Arnold transform is employed to scramble the RGB components of the second watermark image. Secondly, those RGB components of the color host image and two watermark images go through by gray monogenic wavelet transform and an elliptical monogenic wavelet transform. Thirdly, the amplitude and phase are selected as embedded features from the ellipse parameters generated by the host image. The unitary matrix is obtained by singular value decomposition with the amplitude of the first watermark image and the phase of the second watermark image. According to our proposed embedding rules, the watermark is embedded into the unitary matrix. Finally, the watermarked image can be extracted by inverse singular value decomposition and inverse elliptical monogenic wavelet transform. The experimental results show that the proposed algorithm performs very robustly against non-geometric attacks, such as salt & pepper noise, Gaussian noise, median filtering, mean filtering, JPEG compression, and geometric attacks, including rotation, clipping, and translation. Furthermore, our method demonstrates extraordinary performances when compared to related methods in certain aspects.