A Novel Watermarking Algorithm Based on ICA

Through calculating the high-order statistic information of data, independent component analysis (ICA) algorithm can separate the observation signals mixed by some mutually independent source signals blindly and effectively. So it has been used to extract the image features and separate the mixed signals blindly. Making use of the merit of ICA, we propose a watermarking scheme based on ICA in this paper. We first introduce ICA to obtain a set of statistically independent sources; then we use ICA for watermark blind extraction. The experimental results show that the proposed scheme has better invisibility and good robustness to different kinds attacks.