Color image watermarking in the fusion of LWT and tensor-SVD domain in digital image watermarking framework
Saharul Alom Barlaskar, Mohiul Islam, Naseem Ahmad, Anish Monsley Kirupakaran, Rabul Hussain Laskar, Taimoor Khan · 2024
In this paper, we present color image watermarking considering three RGB channels as third order tensor, upon which tensor singular value decomposition (tensor-SVD) is performed. It provides two orthogonal tensors and one diagonal tensors. As the diagonal tensor holds major energies of a color image, hence used for inserting watermark data in order to achieve adequate visual imperceptibility and robustness of watermark. Gray scale image have been used as watermark for embedding in the first diagonal tensors of SVD matrices. For embedding watermark data, 1-level lifted wavelet transformation (LWT) has been carried out in each of the separate RGB channels of the host image. LWT provide its four decomposed sub-bands, and consequently HL (horizontal) sub-band of each channels are recombine further to perform tensor-SVD operation. Grayscale watermark is transformed into principal component by SVD operation which is embedded into the first diagonal matrices. For measuring the performance, various performance matrices such as peak signal-to-noise ratio (PSNR), structural similarity index measures (SSIM), and normalized similarity ratio (NSR) have been evaluated. The experimental results reveal there is adequate imperceptibility and robustness can be achieved.