|
As research purposes to improve the robustness of digital image watermarking , this paper introduces the principle of digital watermarking system , the basic framework , attack and performance evaluation ; digital watermarking algorithm based on image features on the basis of in-depth study and analysis summary of the proposed watermarking scheme based on image texture features and regional similarity . Arnold is set on the original watermark signal as non-periodic scrambling process , the the scrambled watermark signal having a periodicity in the recovery process is no longer this method significantly improves the security of the watermark . Multi-resolution wavelet analysis of multi-scale features and the human visual system theory ( HVS ) , the right image texture analysis method based on wavelet high frequency coefficients of variance characteristics , priority watermark embedded in the image region of interest ( ROI ) texture region corresponding to the low frequency wavelet coefficients , the algorithm can effectively resist such as noise , filtering , JPEG compression , shear and other conventional attacks . Will maintain the characteristics of high similarity based on the local area of the image before and after the attack , while the introduction of polar transformation on the number of scale and rotation invariance , based the regional similarity watermark extraction algorithm . The algorithm in resistance to translation, rotation, scaling and other geometric attacks . Algorithm and experimental mentioned in MATLAB ( R2007a ) environment , all the experimental results show that : the proposed algorithm can resist common image noise , filtering , JPEG compression , shear , translation, rotation, scaling, etc. treatment, with strong robustness and invisibility .
|