A Watermarking Scheme Based on Compressive Sensing and Bregman Iteration

Abstract

A digital watermark is a marker embedded in a signal for the ownership identification of the copyright. In this paper, we propose a half blind digital watermark method based on compressive sensing and singular value decomposition. The proposed method replaces the singular values of the HH band of the cover image in discrete wavelet transform domain with the singular values of the compressive watermark. To guarantee secure recovery from compressed watermarking measurements, specific chaotic matrix can act as an authentication mechanism, which satisfies the restricted isometry property, to implement action of the sensing procedure. With this special authorized matrix, the watermarking signal can be correctly reconstructed. The proposed method also uses Bregman iteration to improve the perceptual quality for their successful use of denoising of images before. Simulation results verify the effectiveness of the proposed approach and the performance is considerable.