基于可逆整数变换的无损水印新算法

无损(也叫可逆)水印主要用于数据(如图象)的完整性认证,其特点之一是当加水印图象没有被修改时,所嵌入的水印可以完整地从加水印图象中去除,从而可以在接收端无损地恢复出原始图象。本文提出一种新的基于可逆整数变换的无损水印算法,利用整数变换后图象象素对均值和差值之间所存在的冗余,嵌入认证信息。文献中已有的方法只利用了差值冗余,而我们的算法则进一步考虑了均值和差值的相关性。利用算术编码压缩由二者所得的相关位流,可以获得较大的存储空间和较高的加水印图象质量。

A new algorithm for lossless Watermarking in reversible integer transform

Lossless ( reversible) watermarking is mainly used for data (e.g. image) integrity verification. One of its features is that, when the watermarked image is deemed as authentic, we can completely remove the watermark from the watermarked image and losslessly resume the original image content at the decoder end. In this paper, we propose a new algorithm for lossless watermarking based on invertible integer transform. We use the statistical redundancy between the average and the difference of a pair of image pixels for authentication message embedding. The methods available in literature only consider the redundancy of the difference. However, we take advantage of the correlation between the average and the difference of a pair of image pixels. The bit- stream from their correlation is arithmetically encoded, so that we can achieve larger spare space for embedding and higher quality of the watermarked image.