Abstract: To balance the contradiction between the robustness and invisibility of the spatial digital watermarking, a multi-scale contextual discontinuous measure was proposed as new perceptual-mask function to determine the texture and edges of image. Combined with the human visual system the mask function was applied to spatial digital watermarking. In the embedding process, the mask function was employed to control the position and intensity of water mark, resulting in self-adaptive embedding of watermark signal. In the detecting process, the inhomogeneity and image diffusion theory were combined together to evaluate the watermark signal and to make prediction, so that the blind examination is realized. Simulation results show that under the same robustness requirements, a watermarking scheme using a mask via inhomogeneity has a better visual quality. Compared with other methods, the extracted signal of the proposed scheme has higher accuracy under the same signal processes or geometric attacks.

Key words: information processing, inhomogeneous, human visual system, masking function, digital watermarking, blind detection