基于关系的提升小波变换域自适应水印方案

目的为了提高嵌入水印后图像的不可感知性和抵抗攻击的能力,提出一种新的基于关系的提升小波变换域自适应图像水印算法。方法先对水印信号进行预处理以增强水印信号的安全性,然后对载体图像进行互不重叠的分块并对每一分块进行二级提升小波变换,其次根据选定的阈值、每一分块选定的系数差值的绝对值及每一分块的亮度和纹理掩蔽特性,自适应地调整选定的一对系数的大小关系以嵌入水印信号,最后进行提升小波逆变换得到嵌入水印后的图像。提取水印时,不需要原始载体图像的参与。结果大量的仿真实验表明,该算法具有较好的不可感知性,对常规的信号处理和部分Adobe Photoshop CS6软件处理也具有较强的鲁棒性。与相似的水印技术相比,该算法具有更好的抗攻击性能。结论提出的算法能满足图像版权保护应用的需要。

Adaptive Watermarking Scheme Based on Relationship in LWT Domain

The work aims to propose a new adaptive image watermarking algorithm based on relationship in lifting wavelet transform (LWT) domain in order to improve the imperceptibility and the ability against attacks of watermarked image. Firstly, the watermarking signal was preprocessed to enhance its security, and then the carrier image was divided into non-overlapping blocks and the 2-level LWT was performed on each block. Secondly, according to the selected threshold, the absolute difference value of the selected coefficients, and the brightness and texture masking of each block, the watermarking signal was embedded by adaptively adjust their relationship of the selected coefficients. Finally, the watermarked image was obtained by LWT. In the watermarking extraction, the original carrier image was not required. A lot of simulation experiments showed that the proposed algorithm had better imperceptibility and it also had a strong robustness against conventional signal processing and part of Adobe Photoshop CS6 software processing. Compared with similar watermarking technologies, the proposed algorithm had a better capability against attacks. The proposed algorithm can meet the needs of application of image copyright protection.