基于无损压缩和差值扩展的可逆图像水印算法

Reversible Image Watermarking Algorithm Based on Lossless Compression and Difference Expansion

针对传统可逆图像水印算法中存在高失真度和低嵌入率的问题,提出了一种基于无损压缩和差值扩展的可逆图像水印算法。在算法设计中首先对原始图像进行多尺度分解,将原始图像划分为同质块和非同质块。然后对同质块中每个像素进行误差预测,将每个像素点的像素值和对应的预测误差值组成像素对通过差值扩展算法来实现水印嵌入。最后,标注图像溢出像素点并压缩溢出图;将压缩溢出图和剩余未嵌入完的水印信息构成辅助信息一起嵌入到非同质块中。实验结果表明,该算法不仅降低了图像失真程度,而且提高了信息嵌入能力。此外,在同等水印嵌入量下,该算法的PNSR和SSIM高于其他一些常见的算法。

A reversible image watermarking algorithm based on lossless compression and difference expansion is put forward for the contradiction between embedding capacity and the visual quality of image.Firstly,the original image was divided into homogenous and non-homogenous blocks through multi-scale decomposition.Then,every pixel in homogenous block was made error predication,and watermark was embedded by composing pixel value of every pixel point and its corresponding predicating error into pairs with difference expansion algorithm.Finally,the pixel points,which exceeded the range of image gray-value value after embedded information by difference expansion,were marked in overflow graph;the auxiliary information composed by compressing overflow graph and surplus information which were not embedded was embedded into non-homogenous blocks.The results showed that this algorithm not only decreased the image distortion,but also improved the information embedding capacity.In addition,the PNSR and SSIM of this algorithm were higher than some common algorithms under the same watermark embedding capacity.