| With the advent of Internet,the digital media such as pictures,texts,audio,video through has been used widely over the decade years.Since digital media are easily reproduced and processed,anyone is potentially capable of incurring considerable financial loss to the media producers and providers.Digital watermarking is introduced to ensure the image integrity and protect the ownership authentication Apart from signature and stegography.Information related to copy protection could be embedded to the media and,in case,extracted or detected correctly from the media to prove the ownership of the media.The main study in the paper can be described as follows:1 Wavelet transform theory and the development of image watermarking are introduced and several techniques of warermarking are discussed,including the model, characteristics,type and experiment results of which are presented in detail.We also compared the advantages and disadvantages over the classical algorithms above and evaluated them subjectively.2 On the basis of these analyses,a support vector machine SVM-based digital image-watermarking scheme is developed,which is robust against a variety of common image-processing attacks and desynchronization attacks.The watermark embedding and extraction issues can be treated as a classification problem.A set of training patterns is constructed by employing two descriptors representing image features.The results illustrated the scheme is of higher preciseness and lower complexity.3 A geometrically robust image watermarking algorithm based on significant region of normalized image is proposed.The geometrically invariant space is firstly constructed by utilizing the moment invariant theory and image normalization method.The digital watermark is embedded into the significant region through singular value decomposition.Experimental results show that the proposed scheme is not only invisible and robust against common signal processing,but also robust against the geometric distortion. |
PDF Full Text Request