A perception based color image adaptive watermarking scheme in YCbCr space

Copyright protection has now become a challenging domain in real life scenario. Digital watermarking scheme is an important tool for copyright protection technique. A good quality watermarking scheme should have high perceptual transparency, and should also be robust enough against possible attacks. A well-known (Lewis-Barni) Human Visual System (HVS) based watermarking model is fairly successful with respect to the first mentioned criterion, though its effectiveness in color images has not been claimed. Furthermore, it is true that although several watermarking schemes are available in literature for grayscale images, relatively few works have been done in color image watermarking, and the little that have been done, have mostly been tested in RGB, YUV, YIQ color spaces. Thus the question remains that, which is the optimal color space for color image watermarking and whether this HVS model is applicable for that color space.

There are two main contributions of the present work with respect to the above. First, it claims that for color image watermarking, the YCbCr space can be used as the perceptually optimum color space, the Cb component being the optimal color channel here. Second, it also tests the effectiveness of the above-mentioned HVS model in that color space. These have been achieved by using the HVS model to propose a new non-blind (original image and the watermark logo image both are needed for extraction) image adaptive Discrete Wavelet transform and Singular Value Decomposition (DWT-SVD) based color image watermarking scheme in YCbCr color space. The multi-resolution property of DWT and stability of SVD additionally makes the scheme robust against attacks, while the Arnold scrambling, of the watermark, enhances the security in our method. The experimental results support the superiority of our scheme over the existing methods.