In this paper, mathematical modeling of digital watermarking is proposed to approximate the image based on the generalized Gaussian distribution. Using maximum a posteriori probability based image segmentation and fuzzy c means image segmentation, the cover image is segmented into several homogeneous areas. In EM segmentation, every region in the image is represented by a generalized Gaussian distribution. The rotation invariant features are extracted from the segmented areas and are selected as reference points by DoG filter and principal component analysis. Rotation and scaling invariance is obtained through the process of image normalization. The watermark embedding and extraction schemes are analyzed mathematically based on the established mathematical model. The mathematical relationship between fidelity and robustness is established. A hybrid watermarking technique is proposed to improve the similarity of extracted watermarks. Furthermore, genetic algorithm (GA) is simultaneously performed to find the optimal values such as fitness value, best points and CPU time. This method has been proved its robustness to geometric attacks through experiments. The experimental results show the effectiveness and accuracy of the proposed scheme.
P. Surekha, P, Y., and Sumathi, S., “An efficient optimization technique for digital watermarking in image processing”, in IEEE Int. Conf. Intelligent Control and Information Processing (ICICIP), , Dalian, China, 2010.