The Research Of Self-adaptive Quantization Digital Watermarking Algorithm

With the rapid development of the Internet, various kinds of digital multimedia information have widely spreaded through network such as text, image, audio and video etc. But at the same time, the information on Internet can be easily duplicated and modified so that the copyright infringing problems on internet are becoming more and more serious. Therefore, how to provide copyright protection has drawn extensive attention in recent years. To solve this problem the digital watermarking technique has been investigated.Firstly, this thesis introduces the backgrounds and status about the digital watermark at home and abroad. The concept, basic characteristics, classifications, evaluations about the watermark, the attacks faced, countermeasures and so on are comprehensively summarized and concluded.Secondly, the theories of wavelet analysis and Human Visual System are researched. A method is proposed to analyze all levels coefficients of wavelet while selecting the coefficients of wavelet. The method educes the sequences of each subband's importance in order to find the best position of watermark embedding by the way of comparing with the energies, mean values and variances of the sub-images decomposed by wavelet.Finally, on the basis of the above theories, the material scheme about the self-adaptive quantization digital watermarking algorithm is put forward. In this algorithm, the meaningful binary watermark image which is generated from original image is used according to the characteristics on human visual system, an adaptive selection scheme is constructed. Then using human visual system to confirm the quantization step and embedding the binary watermark signal which is disordered by Arnold transform into the sub-image of low frequency coefficient of the primitive image self-adaptively. At last, the inverse transformations are done and restructured to obtain the images including the watermark information as big as the original image.The watermark algorithm mentioned in the thesis is simulated in Matlab environment. All the simulation experiment results show that the algorithm can stand familiar image processing such as compression, blur, noise, crop and so on with strong robustness and transparence.