Based On Wavelet Transform And Singular Value Decomposition Of Color Image Watermarking Algorithm Research

Digital color image include a wide and complex classification, as the size of the Internet grows the volume of digital color image data floating around also increases day by day as the size of the Internet grows. As many advanced tools are readily available to duplicate and modify those data in the Internet easily, security is the major concern, which requires some mechanisms to protect digital multimedia data. Thus watermarking is a technique which supports with feasible solution. Grayscale image watermark mature technology is not suitable to be directly applied to color images, we also should to take full account of the color space and channel selection which are not involved in the grayscale image watermark, base on this two aspects, we combined the discrete wavelet transform and the singular value decomposition proposed two color image watermarking algorithm.This dissertation presented a watermarking algorithm for color image guided by wavelet domain visual model using block singular value decomposition. Taking the high correlation of the RGB color space into account, Firstly, the host image is converted to YUV color space which defines a color space in terms of the most important one luma (Y) and two chrominance (UV) components. Decomposing the Y channel by one level Haar wavelet transform, then divided it into sub-block decided by the size of watermarking, the watermarking is embedded into host image by the slightly modification of the median frequency’s largest singular value decomposition coefficient guided by HVS model consist of brightness, texture, frequency, and contrast masking. Experimental results show that the algorithm maintain the PSNR at the50.2765db and also has strong robustness again common attacks such as noise, rotation and cut.This dissertation presented the discrete wavelet transform (DWT) and the singular value decomposition (SVD) based color image adaptive watermarking algorithm. Firstly decomposing the color component of the original RGB color image using discrete wavelet transform respectively, we apply the SVD to HL, LL band and the watermarking alone, then embed the same watermarking data into three channel by modifying the two band’s singular values. The additional advantage of the proposed technique is taking advantage of the relationship between HVS and JND which can adaptively regulate the watermark embedding strength, So as to achieve the purpose of enhance the watermark robustness and ensure transparency of watermark. The algorithm is suitable for binary and grayscale watermark, an extraction threshold value set in the binary watermark extraction process, to ensure that the integrity of the extracted watermark. Simulation results show that this algorithm for better coordination, transparency and robustness of the watermarking system is a more robust algorithm. Two algorithms study modification about the largest value of the block’s singular matrix and the global singular value of the entire image. Take full advantage of the wavelet transform’s multi-resolution characteristics and singular value decomposition’s good stability, well coordinated visibility and robustness of the watermark, the dissertation do a lot of experiments, the experimental results show that both have good features.