Digital image watermarking is one of the proposed solutions for copyright protection of multimedia data. This technique is better than Digital Signatures and other methods because it does not increase
IntroductionDigital image watermarking is a process of embedding a piece of digital content (image) into the original image and also it protects digital content from illegal manipulations. The hidden watermark should be inseparable from the host image and robust enough to resist any manipulations while preserving the image quality [1]. Watermarking adds the additional requirement of robustness. An ideal watermarking system would embed an amount of information that could not be removed or altered without making the cover object entirely unusable [2]. So, watermarking is mainly to prevent illegal copy or claims the ownership of digital media. It is a measure of immunity of watermark against attempts to image modification and manipulation like compression, filtering, rotation, collision attacks, resizing, cropping, etc [3]. In order to minimize the damage of the original image and enhance the robustness, a kind of algorithm has been proposed based on the image being layered. At present, the wavelet decomposition method is the most familiar method based on this kind of algorithm [4][5][6].Cao [7] proposed a watermarking algorithm based on transform domain. This algorithm has some robustness characteristics, but it is not efficient to resist the 1/4 and 1/2 heavily cropping. Dai [8] proposed a digital image blind watermarking algorithm based on wavelet transform, which can efficiently resist the cropping attack, but the robustness to resist the noise attack is not strong. Zhang [9] also proposed a digital image watermarking technology based on wavelet transform, which is operated easily, but the ability to resist the noise attack is not strong. To improve the robustness of digital image watermarking method, this paper proposes a new algorithm with stronger robustness to resist the attacks of impulse noise and cropping based on robust principal component analysis (RPCA) and discrete cosine transform (DCT).