Scrambling Image Watermark Algorithm Based on DCT and HVS

Abstract

With DCT and HVS, this paper proposed a frequency domain image watermarking algorithm based on mapping principle in Toral Automorphism. Firstly, we scrambled the original watermark image with Arnold transformation. Secondly, we transformed the original carrier image by discrete cosine transformation and chose some proper intermediate frequency coefficients based on human visual system. Thirdly, the processed watermark was embedded into these coefficients with spread-spectrum mechanism. Finally, the watermark was extracted by the blind detection and recovered by periodic Arnold transformation, which could authenticate digital works and protect the copyrights. The algorithm has the advantages of robustness and security with scrambling transformation and spread-spectrum techniques, invisibility with HVS features.