Anotace:
The main objective for a watermarking technique is to attain imperceptibility, robustness and security against various malicious attacks applied by illicit users. To fulfil these basic requirements for a scheme is a big issue of concern. So, in this paper, a new image watermarking method is proposed which utilizes properties of homomorphic transform, Redundant Discrete Wavelet Transform (RDWT), Arnold Transform (AT) along with Singular Value Decomposition (SVD) to attain these required properties. RDWT is performed on host image to achieve LL subband. This LL subband image is further decomposed into illumination and reflectance components by homomorphic transform. In order to strengthen security of proposed scheme, AT is used to scramble watermark. This scrambled watermark is embedded with Singular Values (SVs) of reflectance component which are obtained by applying SVD to it. Since reflectance component contains important features of image, therefore, embedding of watermark in this part provides excellent imperceptibility. Proposed scheme is comprehensively examined against different attacks like scaling, shearing etc. for its robustness. Comparative study with other prevailing algorithms clearly reveals superiority of proposed scheme in terms of robustness and imperceptibility.