Robust Watermarking through Spatially Disjoint Transformations

Abstract

An image watermarking scheme is introduced with an aim to satisfy the imperceptibility and robustness requirements. The image is divided into spatially disjoint areas. Watermarking is done by interleaving DCT (Discrete Cosine Transform), DWT (Discrete Wavelet Transform) and DFT (Discrete Fourier Transform). Multiple watermarks are simultaneously embedded in a single image in random order. Each area is selected sequentially and watermarked independently. The main advantage of this concurrent watermarking method is that it makes the watermark invariant to all the available frequency based attacks and geometric attacks. The shortcomings of one transform based watermarking scheme are overcome by the other. DFT based watermarking is invariant to geometric attacks and watermark detection attacks but it is not best suited for frequency based attacks like filtering and noise addition. Its shortcomings are overcome by applying the DCT and DWT based method in other areas which are resilient to frequency based attacks but fail against geometric attacks. Hence if any method fails against an attack the watermark can be efficiently extracted by the other implemented methods. Simulation results show that proposed method is able to withstand many image processing attacks.