A new blind image-adaptive watermarking scheme: Theoretical and experimental results

In this paper a new blind image-adaptive watermarking scheme is proposed. The scheme employs a new spatial mask whose construction is based on the HVS's (Human Visual System) properties. Specifically, the mask is a function of the local variance of the cover image prediction error sequence. An improved blind detection scheme has also been developed, based on a proper pre-whitening process. Due to the above modifications the proposed technique exhibits superior performance as compared to conventional HVS-based blind adaptive watermarking. The proposed detector's performance improvement has been justified theoretically for the cases of no attack, noise attack and linear filtering attack, and is also verified through extensive simulations. The theoretical analysis is independent of the proposed mask and the derived expressions can be used for any watermarking technique based on spatial masking. Moreover, proper expressions have been derived that enable the computation of detection thresholds adaptable to the attacking conditions.