The Robust Watermarking Scheme For Color Image

Digital media has made our life more colorful because of its advantages like easier to access, copy and distribute. But as what we have seen, series of malice activities like copyright infringement, information distortion, make serious damages to both the producers and the users of the digital products. So we really need a new technology to protect the copyright, authenticity, integrity of the digital products and the intellectual property and property safety of the users. Watermarking technology emerged under this background.After studying the present situation of watermarking technology and analyzing the research results of the others, this thesis proposes two improved scheme based on DWT( Discrete Wavelet Transform).The first one is based on the research on the characteristics of image tricolor, human visual system and discrete wavelet transform. Firstly, we choose the optimal embedding color component according to the proportion color component takes up. Secondly, split the coefficients which are got from the DWT into several sub-blocks of the same size. Then we classify these blocks by more than one parameter which is designed according to HVS(Human Visual System). At last, we embed the binary watermark into wavelet blocks using different embedding strength based on different block categories we have got above and the features of DWT. The experimental results show the embedded watermark is invisible and robust against common attacks like Gaussian noise, pepper salt noise, rotation etc. and especially the compression.In the second method, By combining DWT and DCT(Discrete Cosine Transform) it proposes a blind watermarking algorithm by taking advantages of both DWT and DCT. At first place, same as the first algorithm choose the optimal embedding color component according to the proportion color component takes up. Then apply DWT to it. Thirdly apply sub-sampling and DCT to the low frequency sub-image which is got from the former discrete wavelet transform. At last, embed the information into the DCT coefficients. The comparison between the simulation results of this scheme and the algorithm in reference shows that the improved algorithm has a good performance on resistance to the most common attacks compression and filtering. In addition to its blind extraction scheme, make this watermarking scheme has a very bright application future.