A Robust Chaotic Image Watermarking Technique based on HVS and Differential Affine Motion Estimation

This paper deals with the insertion of a chaotic signature in an image by exploiting the analogy presented by the wavelets transform and the human visual system (HVS) model, to modulate and adapt the signature according to local characteristic's of the image. The blind detection process consists on computing the correlation between the marked DWT coefficients and the watermarking sequence. In order to face the problem of the geometrical de-synchronization, a differential technique of motion estimation is employed to compensate the possible geometrical deformations undergone by the watermarked image. The method has been proved to be robust to various image processing such as compression, filtering and noise addition and various geometrical attacks such as translation, rotation, scaling, cropping and resizing.