摘要
针对现有的图像篡改检测方法存在篡改定位不精确以及未受篡改区域不能有效进行无损恢复的问题,提出采用独立可逆认证的图像篡改检测方法。首先,采用图像分块的独立可逆认证方法,结合可逆信息隐藏算法的嵌入原理,将现有认证方法中的认证单元由常规的4×4图像子块缩小为4×3图像子块,解决了认证单元偏大的问题,提高了篡改检测的定位精度;然后,根据子块特点,自适应地选取改进的像素值排序、局部直方图平移、差值扩展这3种可逆信息隐藏算法,保证每个子块至少可嵌入1 bit认证码的同时,将平均可嵌入认证码的数量提高到3.6 bit,实现了图像子块独立认证与提升了篡改检测正确率;最后,考虑了3种可逆信息隐藏算法之间的连贯性与可逆性,以确保嵌入端的无效移位尽可能少,以及认证端认证信息的可靠提取。实验结果表明:对于未受篡改的图像区域,所提方法可以实现原始图像的无损恢复;对于剪贴攻击、常数攻击、拼贴攻击和随机篡改等常见的篡改攻击,所提方法的初始正确检测率较现有最优方法平均可提升5.82%~22.3%,细化处理后的正确检测率可接近100%。
To address the imprecise localization of tampering and ineffective recovery of untampered areas in current image tampering detection methods,a new independent reversible authentication-based approach for detecting image tampering is introduced.Initially,leveraging an image block-based independent reversible authentication technique and incorporating the embedding principles of reversible information hiding algorithms,the authentication units in existing methods are reduced from the conventional 4×4 image sub-blocks to 4×3 image sub-blocks.This modification tackles the issue of oversized authentication units,thereby enhancing the localization accuracy of tampering detection.Subsequently,based on the characteristics of sub-blocks,three enhanced reversible information hiding algorithms,specifically improved pixel value sorting,local histogram shifting,and difference expansion,are adaptively selected to ensure that each sub-block is embedded with at least 1 bit authentication code.This process increases the average embeddable authentication code quantity to 3.6 bits,facilitating independent authentication of image sub-blocks and improving tampering detection accuracy.Finally,the coherence and reversibility of the three reversible information hiding algorithms are considered to minimize irrelevant shifts during the embedding stage and ensure the reliable extraction of authentication information during the authentication stage.The experimental results demonstrate that the proposed method enables lossless recovery of the original image in untampered areas.When confronted with common tampering attacks such as cut-and-paste,constant manipulation,collage alterations,and random tampering,the initial accuracy rate of the proposed method can be improved by an average of 5.82%to 22.3%compared to existing optimal techniques,with the refined accuracy rate potentially approaching 100%.
作者
王雨珂
黄方军
WANG Yuke;HUANG Fangjun(School of Cyber Science and Technology,Shenzhen Campus of Sun Yat-sen University,Shenzhen,Guangdong 518107,China;Guangdong Provincial Key Laboratory of Information Security Technology(Sun Yat-sen University),Guangzhou 510006,China)
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
2024年第12期141-152,共12页
Journal of Xi'an Jiaotong University
基金
国家自然科学基金资助项目(62072481,U2336208,62472454)。
关键词
图像篡改检测
可逆信息隐藏
像素值排序
独立可逆认证
image tampering detection
reversible data hiding
pixel-value-ordering
independent reversible authentication
