A progressive QIM to cope with SVD-based blind image watermarking in DWT domain

摘要

A novel progressive QIM technique is developed to cooperate with the SVD-based watermarking scheme in the DWT domain. The progressive structure in quantization step sizes not only allows the exploitation of human visual characteristics but provides the clues of the ground level of brightness. A particle swarm optimization method is adopted to identify the ground level prior to performing watermark extraction. The experimental results show that the proposed scheme renders better robustness in comparison with other SVD-based methods. With the capability of the ground level identification, this scheme can also survive the level shift attacks.