Image Digital Watermarking Researches Based On Transform Domain

With the rapid development of Internet, computer and wide application of digital technology that make the dissemination and access of digital multimedia products easier than ever, illegal copying, malicious tampering and copyright protection have become very important issues. Digital watermarking, as a solution to copyright protection and information security, is a technology that embeds information into the original carrier. It has becoming a hot research topic of the current academic and applied fields.In this paper, a series of studies about image digital watermarking have been done. Firstly, this article introduces and summarizes the basic theories of digital watermarking, including the concepts, classification, and applications of digital watermarking technology, also including watermarking algorithm model and some classical algorithms; Secondly, it introduces the theory of wavelet transform, singular value decomposition and the watermark pretreatment methods and so on.In view of the above theories, a new watermark scheme is proposed in the fourth chapter based on discrete wavelet and discrete cosine transform. The algorithm embeds the watermark information into the transform domain of original carrier after discrete wavelet and discrete cosine transform. The simulation results prove that the algorithm has achieved to embed watermark successfully, and has strong robustness against conventional processes and malicious attacks.Due to the widely dissemination and application for color images in network times, a new color image watermarking algorithm has been proposed in the fifth chapter based on subsampling and discrete wavelet transform. The scheme embeds the watermark information into the Y channel obtained through transformation by the original color image. In the embedding process, firstly, obtain the embedding subgraph by subsampling technology. Secondly, based on the good location in both spatial——frequency domain and transformation mechanism consistent with human visual characteristics of wavelet transform, the embedding position is selected in the wavelet low frequency components. Finally, the watermarking information is embedded into the singular values of the low frequency coefficients by intermediate matrix, In addition, to further enhance the security of the algorithm,the watermark is encrypted by chaotic sequence. Based on the reference subgraph, a semi-blind extraction is achieved. Lastly, the simulation results show that the algorithm not only effectively improve the PSNR values,but also enhance the robustness against geometric attacks, JPEG compression and Gaussian noise attack and etc..