An effective digital image watermarking scheme incorporating DCT, DFT and SVD transformations.

Abstract

Image watermarking prevents copyright infringements by attaching visible/invisible watermark images as authentication identities in the owner's documents. The article made analysis on the advantages of different transformations for choosing better combinations to make the watermark embedding process and observed that watermarking techniques incorporating discrete cosine transform (DCT) provide better resistance towards JPEG based potential attacks, discrete Fourier transform (DFT) has strong energy compaction with geometrical invariance properties to resist geometric attacks while singular value decomposition (SVD) provides stability, proportion invariance and rotation invariance properties and it provides strong resistance against noise based attacks. Considering these advantages of different transformations, the article presents a new non-blind watermarking algorithm by utilizing advantages of DFT, DCT and SVD transforms while attaching secret contents in cover images. The algorithm starts by applying DFT followed by onion peel decomposition (OPD) for decomposing Fourier domain carrier image to four frequency sub images. The scheme then applies DCT on the frequency bands and orders them in zigzag fashion to form four individual frequency arrays. In the final step of embedding process, it embeds four copies of watermark singular value contents in DFT domain with the carrier image singular value contents to produce the watermarked image. The experimental results based on subjective and objective metrics on various test images from standard image databases with different test conditions demarcate the stability of new algorithm in producing high quality watermarked images with fewer distortions even when the watermarked images are extremely distorted by potential attacks.