A multilevel image-adaptive digital watermarking algorithm based on chaotic sequences and ICA

This paper presents a novel multilevel image watermarking method using chaotic sequences and independent component analysis (ICA) for still images. Discrete wavelet transform (DWT) is applied to one subimage extracted from the original image according to one chaotic sequence label; then multiple independent chaotic watermark signals are generated and embedded into the important wavelet coefficients adaptively. ICA is applied to extract the watermark and determine where there exists the original watermark in the original image. Simulation results demonstrate that this algorithm proposed in the paper is a new effective digital watermarking method with very outstanding robustness against many common attacks such as noise attacks, geometric operations and JPEG compression.