A Novel Invisible Image Watermarking Based On The Relation Between The Selected DCT Coefficients

Digital image watermarking main goals are the high robustness and imperceptibility of the watermark with minimum degradation in the watermarked image. A novel watermarking method is introduced in this work that relies on the Discrete Cosine Transform (DCT). The transformed blocks of the host image are divided into two groups, the first contains the blocks that have the DCT coefficient (2,3) greater than (3,2) and the second group contains the blocks where the DCT coefficients (2,3) less than or equal to (3,2) sorted in descending order. In embedding the binary watermark image, the pixels with value one are embedded in the first group while the zero pixels are embedded in the second group. If one of the groups is empty the embedding is performed on the other group, and the values of the coefficients (2,3) and (3,2) are exchanged. The difference between the two selected DCT coefficients mustn't be small to be able to resist attacks, so small differences are manipulated. The proposed technique achieved high imperceptibility with an average PSNR equal to 49 for 64*64 watermark size with 512*512 cover image size and infinity PSNR for the watermark of size 32*32. The obtained results confirmed the high imperceptibility and robustness of the suggested technique compared with others.