Adaptive Watermarking Algorithm Based On The Color Image Of The Hvs And Svr In Wavelet Domain

With the advent of information age and the popularity of the Internet, the release and transport of digital products become more convenient and frequent. The protection of information safty therefore becomes increasingly prominent. Digital products can easily be illegally edited, modified, copied and disseminated. Thus, to protect the original creators'copyright of these digital products and their economic interests is becoming increasingly imminent. Digital copyright protection issue has attracted people's attention. In this context, the digital watermarking technology emerged. At present, the technology has been considered a solution to digital copyright protection and an important means of safe transmission of digital products.There are lots of digital watermarking algorithms. In recent years, self-adapting digital watermarking technology becomes a hot-spot research field of the digital watermarking. Based on this study of digital image watermarking, the main work of this paper is as follows:The paper introduces the basic characteristics, classification, application and model of digital watermarking technology systematically, and focus on analysis of the existing typical watermarking algorithm. Then this paper expounds the theory of wavelet analysis and support vector machine, focus on analyzing the advantages of wavelet domain digital watermarking technology and application of support vector machine theory in digital watermarking technology.Based on the deep analysis and study of existing digital watermarking algorithms, the paper focus on the digital watermarking in DWT, especially is engaged in the design of self-adapting digital watermarking algorithm.Through improving flaws of existing algorithms based on SVM, an adaptive watermarking algorithm for color image in the wavelet domain based on SVR is proposed in this paper. First, the blue color component is cut into sub-blocks and some sub-blocks are selected. SVR training model is adopted to model the HVS to determine the adaptive watermark strength by choosing the image features of sub-blocks such as the luminance and the feature of texture which affect the HVS as the input vector of SVR. Second, all sub-blocks are decomposed with the wavelet decomposition. Finally, watermark is embedded in the low frequency coefficients of every sub-block. This algorithm can achieve the best effect between robust and invisibility by using HVS and the excellent learning ability of SVR. A large amount of experiments show that the proposed algorithm is effective and robust to common image processing operations.Feature of this paper:In association with HVS, the adaptive watermark strength is determined by using the excellent learning ability of SVR.