基于单向极值预测误差扩展的可逆信息隐藏算法  

Reversible data hiding algorithm based on unidirectional extremum prediction error expansion

作　　者：肖为恩 张正伟 刘天府 李瑶 孟倩 XIAO Wei’en;ZHANG Zhengwei;LIU Tianfu;LI Yao;MENG Qian(Faculty of Computer and Software Engineering,Huaiyin Institute of Technology,Huai’an 223003,China)

机构地区：[1]淮阴工学院计算机与软件工程学院,江苏淮安223003

出　　处：《兵器装备工程学报》2023年第9期313-322,共10页Journal of Ordnance Equipment Engineering

基　　金：国家统计局科技计划项目资助(2018LY12);广东省信息安全技术重点实验室开放资助项目(2020B1212060078)。

摘　　要：为了能够有效解决现有可逆信息隐藏算法中高失真、低嵌入的问题,提出了一种基于单向极值预测误差扩展的可逆信息隐藏算法。该算法采用像素菱形预测策略,利用目标像素十字邻域上的4个参考像素,计算其预测误差值,并通过相邻参考像素的预测误差值对其进行差值预测,利用二者的冗余性,计算出2个极值预测器,生成相应的非对称直方图;然后计算出各目标像素的局部复杂度值,根据该值的高低,对像素进行排序,以使得低复杂度的像素被优先处理;最后通过对直方图的单向预测误差扩展,将水印信息分为两轮依次进行自适应嵌入。实验结果表明,该算法不仅具备较高的嵌入容量,此外在相同嵌入容量下,其载密图像的PSNR值与对比算法相比提高了2 dB左右,具有更高的视觉质量。In order to solve the problems of high distortion and low embedding in existing reversible data hiding algorithms effectively,a new reversible data hiding(RDH)algorithm based on unidirectional error expansion is proposed.The algorithm adopts a pixel rhombus prediction strategy and uses four reference pixels on the cross neighborhood of the target pixel to calculate its prediction error value.In addition,the algorithm predicts its difference by the prediction error value of the neighboring reference pixels and uses the redundancy of the two to calculate two extreme value predictors and generate the corresponding asymmetric histogram.Moreover,the algorithm calculates the local complexity value of each target pixel according to the value of the prediction error.Then,the local complexity value of each target pixel is calculated,and the prediction error is ranked according to the value so that the pixels with low complexity are prioritized.Finally,the watermark information is divided into two rounds for adaptive embedding in turn through the unidirectional prediction error expansion of the histogram.The experimental results show that the algorithm can effectively reduce image distortion and has a high embedding capacity,which has certain performance advantages compared with similar reversible information hiding algorithms.

关 键 词：可逆信息隐藏 局部复杂度 单向预测误差扩展 像素极值预测 非对称直方图 

分 类 号：TP309.7[自动化与计算机技术—计算机系统结构]