【作者】 高晓明；

【导师】 齐影虹；

【作者基本信息】 南昌大学 ， 计算机软件与理论， 2008， 硕士

【摘要】 本文在第一代小波变换的基础上,讨论了一个基于提升小波的彩色图像的盲水印算法。文中通过该算法对水印图像进行了两次嵌入:首先用改进的基于奇异值的算法把水印图像嵌入到原始图像的低频部分,然后再用本文提出的基于区间映射的算法把水印图像嵌入到原始图像的中高频部分。在对水印进行提取时,先在低频进行提取,如低频提取不到水印或提取的水印效果不理想,再在中高频进行水印提取。通过仿真实验可知:该算法对水印图像进行了双重保护,即使含水印图像的任何一部分受损,都能从中完整的提取出水印,并达到很好的效果。通过与其它算法对比,证明了该算法在保证了良好的水印不可见性的同时,对剪切、JPEG压缩、滤波、加噪和剪切低频等攻击均具有更好的鲁棒性。更多还原

【Abstract】 On the basis of the first generation of wavelet transform, this thesis discusses a blind watermarking algorithm of color image based on lifting scheme wavelet.The algorithm embeds watermark twice: First, the watermark images are embedded in the low-frequency part of the original image based on the improved algorithm of singular value, and then, watermark images are embedded in the non-low-frequency part of the original image based on the algorithm of interval mapping.In the extraction of the watermark, the algorithm firstly extracts watermark in the low-frequency part of the original image. If the watermark is not extracted or not effective, then the algorithm will extract watermark in the non-low-frequency part of the original image.The simulation experiments show that the algorithm for watermarking images have dual protection role, which will extract integrated watermark image from the original image of any part of the damage, and achieve very good results.The comparison with other algorithms, we prove that the algorithm not only ensure a good visibility of the watermark, but also achieve a better robust to against attack, such as cut, JPEG compression, filtering, noise and cut low-frequency.更多还原