全方向梯度预测和自适应选块的可逆图像水印  

Reversible Image Watermarking Based on Full-context Gradient Prediction and Adaptive Block Selection

作　　者：李淑芝[1] 胡琴[1] 邓小鸿[2] 张翔[1] 

机构地区：[1]江西理工大学信息工程学院,江西赣州341000 [2]江西理工大学应用科学学院,江西赣州341000

出　　处：《小型微型计算机系统》2017年第1期174-178,共5页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(41362015)资助;江西省教育厅科学技术研究项目(GJJ151522)资助

摘　　要：针对基于预测差值扩展方法在预测精度和隐秘图像隐蔽性上的不足,提出一种基于全方向梯度预测和自适应选块的可逆图像水印.首先通过全方向梯度预测算子得到预测误差;然后将图像和预测误差分块,利用均方误差衡量子块的纹理复杂度,根据纹理复杂度自适应选择适合隐藏信息的误差块;最后采用基于预测误差对扩展的可逆算法进行水印嵌入.算法能自适应地选择水印嵌入位置,有效减少嵌入带来的失真,嵌入水印后图像的平均PSNR值较现有方法高出3.25%左右,图像熵值平均降低了23%左右.实验结果表明,本文算法具有更好的预测精度和隐蔽性.算法适用于医学、军事、卫星等领域.To solve the inadequate concealment of covert image and to increase prediction accuracy based on prediction expansion method,we present a reversible image watermarking algorithm based on full-context gradient prediction and adaptive block selection. firstly, the proposed algorithm through full-context gradient prediction to get prediction error;and then partition the Image and predic- tion error into range blocks, using mean square error to measure texture complexity of sub block, according to texture complexity to se- lect error block of suitable for hiding information; finally, watermarking are embedded by using a reversible watermarking algorithm based on pairwise prediction error expansion method, proposed algorithm can adaptively select location for watermark insertion and ef- fectively reduce image distortion caused by embedding, compared with existing method, the proposed algorithm improves the PSNR of watermarked image by an average of 3.25 %. furthermore, the average image entropy decreased about 23 %. experimental results show that, compared with previous similar schemes, the presented method not only achieves distribution concentration＇s prediction error his- togram, but also has better concealment, the presented method can be applied in many fields,including the areas of medicine, military, and satellites.

关 键 词：可逆水印 预测误差 差值扩展 自适应性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]