Research On Robust Watermarking Algorithm Based On Transform Domain

Digital watermarking is an important research topic in image processing and patternrecognition. And it is an effective solution of digital medium’s copyright protection and theinformation integrity certification. As a multi-disciplinary technology, the study of digitalwatermarking is important academically, and also has great potential in applications. Thispaper focuses on the transform domain’s algorithm for the aim of achieving a good balancebetween transparency and robustness. To achieve this purpose, the focus of our work is onthe processing of watermark image, the inserting of the watermark and the extracting of thewatermark and research the algorithm of robust watermark in transform domain. The majorworks are as follows:1. The research background and significance of digital watermarking both at home andabroad are introduced firstly in this paper. Then aiming at the characteristics andclassification of the watermark, the advantages and disadvantages of these methods areanalyzed.2. A robust and invisible blind algorithm has been proposed, which is based ondiscrete wavelet transform (DWT) and singular value decomposition (SVD). The proposedalgorithm combines both merits of discrete wavelet transform (DWT) and singular valuedecomposition (SVD) watermark system. Firstly, the original image’s low frequencysub-band in DWT domain has been chose as the inserting domain. Then SVD is applied inblocks of the original image’s low frequency sub-band. At last, the watermark is embeddedin the sub-blocks of the singular values in the original image’s low frequency sub-bandafter applying Arnold transform in it. The experiments show that the proposed algorithmhas high robustness to various types of attacks, especially to the compressing by JPEG. Inaddition, the watermark can extract without the original image, therefore the proposedalgorithm has great practical meanings. 3. An adaptive digital watermarking algorithm based on Differential evolutionalgorithm (DEA) and Contourlet transform is proposed. Firstly, the approximationsub-band of host image is achieved after the Contourlet transform of the original image.Then the approximation sub-band of host image is performed SVD to embed thewatermarking. The embedding process will utilize the additive embedding way and theintensity factor is decided by DEA. As the algorithm uses the characteristics of DE, SVDand Contourlet, such as stochastic optimization, less controlled parameters, robustness,strong stability, multi-directional and multi-resolution features, the watermarking algorithmcan resist normal geometric attacks and it has strong robustness. The experimental resultsshow the proposed algorithm performs more effective and invisibility. It also caneffectively resist JPEG compression, filtering, adding noise, histogram equalization,cropping and other attacks.