Providing invariance to nonlinear valumetric scaling for quantization basedwatermarking

摘要

A novel framework providing invariance to a class of nonlinear valumetric distortions, such as gamma correction, for QIM-based watermarking techniques is presented. Valumetric distortions are quite common in image and video processings and the sensitivity to valumetric scalings represents the main weakness of the watermarking techniques belonging to the quantization based class. The proposed method amounts to perform a mapping of the host samples in a proper transformed domain where the watermark is subsequently embedded using a gain invariant QIM-based technique. The effectiveness of this approach has been verified by applying the watermarking system using RDM, AQIM and DM in the logarithmic domain as embedding algorithms. Simulation results provide a useful comparison of the performance of these different techniques within the proposed scheme as well as they confirm the invariance to nonlinear distortions.