Primal sketch based adaptive perceptual JND model for digital watermarking

Watermarking algorithms based on the Just Noticeable Distortion (JND) model show great superiority over other methods, in terms of the watermark imperceptibility and robustness. However, the existing wavelet-based JND models are based on global coefficients, without consideration of image content characteristics. Following the pixel-wise masking idea, a primal sketch based adaptive perceptual JND model (PSAPM) is proposed in this paper, in which an improved watermarking algorithm is designed. It better describes the behavior of the watermark embedder. Experiments show that our algorithm is robust against attacks, including cropping, noise addition and JPEG compression.