Image watermarking using polynomial detectors in the wavelet domain
K.J. McGonigle · 2004
This paper presents a performance evaluation of image watermarking schemes. The extraction process is carried out in the wavelet domain using a spread transform. In addition, the embedding strength is controlled by a psycho-visual model in order to maximize the embedding power while ensuring perceptual transparency. The detection schemes employ nonlinear functions that were recently proposed for Gaussian channels. These are side-informed using a gradient ascent optimization technique in order to reduce self-interference. Simulations are carried out on a large set of natural images.