An image watermarking scheme using new wavelet filter banks

The rapid growth of digital multimedia and Internet technologies have brought big challenge to copyright protection. Billions of dollars are lost every year because of illegal use of digital products. Digital watermarking is an effective method of copyright protection. Recently, watermarking in wavelet domain turns into an active research direction due to the multiresolution property of wavelet transform. Although these researches have achieved quite good results, traditional separable wavelet´s inability of capturing singularities in various directions becomes an obstacle in the way of further improving watermarking capacity and robustness. To solve this problem, we use a new non-separable wavelet filter banks for watermarking. The high-frequency subimages of non-separable wavelet transform can reveal more features which are desirable in watermark embedding than that of common used separable wavelet transform. A new watermarking algorithm is proposed. Experimental results shows that our scheme is more robust against several attacks like JPEG compression, Gaussian noising and cropping ect. than the algorithms carried in the separable wavelet domain.