钟亦友(1996-), 男, 硕士生, 主要研究领域为可逆信息隐藏, 数字水印
黄方军(1973-), 男, 博士, 教授, 博士生导师, CCF高级会员, 主要研究领域为AI安全, 多媒体内容安全
图像可逆认证是一项将可逆信息隐藏和脆弱水印相结合的新技术, 其既能实现对图像的脆弱认证, 还能在提取认证信息的同时无失真地恢复出原始载体, 对图像的原始性和完整性认证具有非常重要的意义. 针对现有可逆认证方法认证精度低、对具有复杂纹理的图像或图像中部分纹理复杂区域无法实现有效保护的问题, 提出一种新的图像可逆认证方法. 首先对待认证图像进行分块, 根据每个子块可嵌入容量将其分为差分块和平移块, 并采用不同的可逆嵌入方法对不同类型的块进行认证码嵌入操作. 为了增大嵌入容量以提高对每个子块的认证效果, 还采取了分层嵌入的方式. 在认证方, 可以通过从每个子块中提取认证码实现子块的篡改检测和定位. 此外, 所提方法还可与形态学中的膨胀和腐蚀操作结合以细化篡改检测标记, 进一步提高检测效果. 实验结果表明, 所提方法能够在同样的认证精度下对纹理平滑和纹理复杂的图像进行保护, 同时还能够实现对几乎所有子块的独立认证和恢复, 具有广泛的适用性.
As a new technology that combines reversible data hiding and fragile watermarking, image reversible authentication (RA) can not only realize the fragile authentication of images but also recover the original carrier image without distortion while extracting the authentication code. Thus, it is of great significance to authenticate the originality and integrity of images. Existing reversible authentication methods have low authentication accuracy and cannot effectively protect images with complex textures or some areas with complex textures in the images. To this end, this study proposes a new reversible authentication method. Firstly, images to be authenticated are divided into blocks, and the obtained sub-blocks are classified as differential blocks (DB) and shifting blocks (SB) according to their embedding capacity. Different reversible embedding methods are employed to embed the authentication codes into different types of blocks. It also adopts a hierarchical embedding strategy to increase embedding capacity and improve the authentication effects of each sub-block. On the authentication side, tamper detection and localization can be realized by the authentication code extracted from each sub-block. In addition, this method can be combined with dilation and corrosion in morphology to refine tamper detection marks and further improve the detection accuracy rate. Experimental results show that the proposed method can protect images with smooth texture and complex texture under the same authentication accuracy, and can also realize independent authentication and restoration of almost all sub-blocks, which has widespread applicability.
随着互联网和多媒体技术的快速发展, 数字图像、音视频作品已遍布人们日常生活的每个角落. 同时随着计算机软件技术的迅速发展, 图像、音视频处理软件也越来越多, 不法分子利用这些软件工具能够轻易地对多媒体文件进行篡改, 如何确保海量多媒体文件在存储和传输过程中的原始性和完整性是一个亟待解决的问题[
为了实现对特定图像的保护, 脆弱水印[
可逆信息隐藏是一种新的无损信息嵌入方式, 它可以在提取所嵌入信息的同时, 无失真地恢复原始载体, 因此也被称之为无损信息隐藏[
可逆认证(reversible authentication, RA)[
文献[
本文第1节详细介绍本文所提出的可逆认证方法. 第2节通过对比实验证明了本文方法的有效性. 第3节是本文总结.
本文算法的整体框架如
Algorithm framework of this thesis
本文算法框架
为了避免值为0, 1, 254或255的像素在嵌入过程中产生上溢或者下溢, 在每一层操作前均统一对输入图像进行如下预处理操作. 具体如下: 首先创建一个空的动态数组, 并将其作为预处理定位表
预处理完成后, 将每个4×4子块按照如
Set partition and diamond predictor
集合划分与菱形预测器
根据得到的预测误差值
对块进行分类后, 利用哈希函数
差分块的认证方法主要分为两部分, 即认证码的嵌入和认证码的提取. 在嵌入端, 采用差值扩展方法将1比特认证码
为了尽可能减少差值扩展过程中引入的失真以及可能产生的溢出, 我们采取先对预测误差排序, 然后根据灰度值自适应地选择扩展方向, 最后利用预测误差之间的差值进行扩展的方式. 具体做法如下所示.
Step 1. 对于任意给定4×4差分块, 按从上到下从左到右的次序扫描得到灰色像素序列
Step 2. 考虑到灰度值接近0或255的像素在差值扩展后可能溢出, 在具体嵌入前首先统计该子块内灰度值在闭区间[0, 7]和闭区间[248, 255]中的白色像素个数, 分别记为记为
向上扩展如公式(4)所示. 其中,
Step 3. 经过Step 1和Step 2得到新的预测误差序列为
Step 4. 如果得到的像素值
Step 5. 重复以上操作, 处理所有的差分块. 最后将预处理定位表
认证方根据嵌入方法的逆操作提取认证码并进行认证, 具体做法如下所示.
Step 1. 从平移块中提取数组
Step 2. 在嵌入信息后的子块中找到灰色像素序列
Step 3. 按照与嵌入时同样的规则将序列
对于向上扩展按照公式(9)进行还原, 其中
Step 4. 经过以上操作得到的还原后的预测误差序列为
根据第1.1节的哈希函数重新生成该子块的认证码, 如果其与提取的认证信息
平移块的认证方法同样分为认证码的嵌入和认证码的提取两部分. 在嵌入端, 采用预测误差直方图平移方法嵌入认证码, 并按照子块的容量调整嵌入的认证码位数; 在认证端, 采用相应的逆操作将嵌入的认证码提取并进行认证. 下面我们将以第1层嵌入为例详细介绍平移块的认证方法.
平移块的容量
Step 1. 对于任意给定4×4平移块, 按从上到下从左到右的次序扫描得到灰色像素序列
Step 2. 根据公式(13)得到信息嵌入后像素值
Step 3. 重复以上操作, 处理所有的平移块.
认证方根据嵌入方法的逆操作提取认证码并进行认证, 具体做法如下所示.
Step 1. 在嵌入信息后的子块中找到灰色像素序列
Step 2. 按照公式(15)对所有
Step 3. 从提取出的
Step 4. 重复以上操作, 处理所有的平移块. 当所有平移块都处理完成后, 从动态数组
在上述第1.1–1.3节中, 我们详细介绍了在第1层中, 认证码的嵌入提取及图像恢复具体流程. 在第1层嵌入完成后, 紧跟着以同样的方式进行第2层嵌入操作. 认证码的提取和图像恢复是上述嵌入过程的逆过程.
根据前文描述, 本文认证码的嵌入采取了分层嵌入的方式. 以4×4子块为例, 每一层嵌入过程中可以嵌入1–8比特认证码, 其中差分块固定为1比特, 而平移块至少1比特, 最多为8比特. 因此, 经过两层嵌入后每个子块中可嵌入2–16比特认证码. 在认证端, 同样采取分层提取的方式, 从每个子块中可以提取的认证码最少为2比特, 最多为16比特. 如果该子块中所嵌入的认证码偏少, 对于部分被篡改的区域, 可能出现提取的认证码和重新生成的认证码恰好相等的情况. 以子块中嵌入2比特认证码为例, 这种巧合的概率理论上为1/4.
考虑到实际应用中篡改区域一般不会小于4×4范围, 因此为了进一步提高对篡改区域的检测和定位效果, 本文结合形态学中的膨胀和腐蚀操作对常见的细化处理方式[
3×3 structural elements
3×3的结构元素
Four adjacent relations of block B
块
实验部分我们主要针对
Experimental image
实验图像
Comparison of embedding capacity and auxiliary information length
嵌入容量和边信息长度对比
图像 | Hong等人[ |
Yao等人[ |
王泓等人[ |
本文方法 | |||||||
容量 (104 bit) | 边信息 (bit) | 容量 (104bit) | 边信息 (bit) | 容量 (104bit) | 边信息 (bit) | 容量 (104bit) | 边信息 (bit) | ||||
Plane | 5.2 | 1 794 | 5.2 | 7 742 | 4.7 | 2 384 | |||||
Splash | 5.3 | 739 | 5.3 | 5 744 | 4.6 | 1 109 | |||||
Tank | 2.8 | 2 867 | 2.8 | 2 867 | 2.5 | 3 527 | |||||
House | 4.6 | 2 981 | 4.6 | 10856 | 4.2 | 3 784 | |||||
Lena | 3.8 | 2 237 | 3.8 | 10000 | 3.2 | 3 360 | |||||
Pepper | 3.1 | 2 566 | 3.1 | 2 568 | 2.8 | 3 197 | |||||
Man | 7.2 | 1 506 | 7.1 | 6 106 | 5.9 | 2 112 | |||||
Baboon | 1.3 | 7 705 | 1.3 | 3 818 | 1.2 | 8 346 | |||||
平均值 | 4.2 | 2 799 | 4.2 | 6 213 | 3.6 | 3 477 |
Comparison of PSNR and SSIM
PSNR和SSIM对比
图像 | Hong等人[ |
Yao等人[ |
王泓等人[ |
本文方法 | |||||||
PSNR (dB) | SSIM | PSNR (dB) | SSIM | PSNR (dB) | SSIM | PSNR (dB) | SSIM | ||||
Plane | 51.04 | 0.9964 | 48.84 | 0.9953 | 50.65 | 0.9963 | 42.40 | 0.9908 | |||
Splash | 51.65 | 0.9957 | 49.55 | 0.9934 | 51.31 | 0.9955 | 46.08 | 0.9887 | |||
Tank | 50.17 | 0.9966 | 50.43 | 0.9968 | 49.86 | 0.9964 | 42.32 | 0.9846 | |||
House | 50.37 | 0.9972 | 47.85 | 0.9960 | 49.96 | 0.9971 | 41.24 | 0.9912 | |||
Lena | 50.60 | 0.9963 | 47.95 | 0.9937 | 50.01 | 0.9960 | 42.64 | 0.9878 | |||
Pepper | 50.34 | 0.9960 | 50.62 | 0.9964 | 50.02 | 0.9959 | 41.39 | 0.9813 | |||
Man | 51.51 | 0.9965 | 49.89 | 0.9965 | 50.97 | 0.9962 | 44.35 | 0.9932 | |||
Baboon | - | - | 49.69 | 0.9986 | - | - | 34.59 | 0.9834 | |||
平均值 | 50.81 | 0.9964 | 49.35 | 0.9958 | 50.40 | 0.9962 | 41.88 | 0.9876 |
实验中对4种方法得到的Lena水印图像进行多种篡改攻击, 并且为了更直观地比较本文方法与其他方法的认证能力以及分块大小对认证性能的影响, 我们首先对不同分块大小下的初始正确检测率进行了比较, 然后结合细化处理进一步展示了本文算法的性能.
Initial tamper detection marker
初始篡改检测标记
Comparison of initial correct detection rates (%)
初始正确检测率对比 (%)
篡改攻击 | Hong等人[ |
Yao等人[ |
王泓等人[ |
本文方法 | |||||||||||
4×4 | 8×8 | 16×16 | 4×4 | 8×8 | 16×16 | 4×4 | 8×8 | 16×16 | 4×4 | 8×8 | 16×16 | ||||
剪贴攻击 | 66.47 | 95.73 | 98.53 | 73.60 | 95.26 | 97.06 | 60.67 | 88.63 | 98.53 | 84.13 | 96.68 | 98.53 | |||
常数攻击 | 63.28 | 96.48 | 100.0 | 74.12 | 99.61 | 100.0 | 62.50 | 87.89 | 96.88 | 75.00 | 99.61 | 100.0 | |||
拼贴攻击 | 62.50 | 95.31 | 100.0 | 70.70 | 100.0 | 100.0 | 69.92 | 95.31 | 100.0 | 76.17 | 100.0 | 100.0 | |||
随机篡改 | 58.64 | 87.50 | 90.00 | 48.47 | 57.50 | 60.00 | 58.02 | 82.50 | 90.00 | 81.48 | 85.00 | 90.00 | |||
平均值 | 61.97 | 93.76 | 97.13 | 66.72 | 88.09 | 89.26 | 62.78 | 88.58 | 96.35 | 78.81 | 95.32 | 97.13 |
使用Lo等人[
Refines the tamper detection mark in Lo’s method
使用Lo等人[
本文提出了一种基于分块的高效图像可逆认证方法. 与现有的可逆认证方法相比, 本文方法可以充分利用图像子块中的信息冗余, 减少了边信息的长度, 几乎能对所有子块进行独立认证. 同时通过采用两层嵌入方式, 进一步提高了嵌入容量. 实验结果表明, 本文方法的整体认证精度高于现有最优的方法, 而且能对具有复杂纹理的图像或图像中部分纹理复杂区域实现有效保护. 此外, 本文还结合形态学中的膨胀和腐蚀操作对已有的细化方案进行了改进, 进一步提高了算法的检测效率. 总体而言, 相比于比现有的可逆认证方法, 本文方法具有更高的篡改检测和定位能力.
Refines tamper detection markers in the proposed method
使用本文方法对篡改检测标记细化处理
Comparison of correct detection rate between different refinement methods (%)
不同细化方式正确检测率对比(%)
篡改攻击 | Hong等人[ |
Yao等人[ |
王泓等人[ |
本文方法 | |||||||
Lo等人[ |
本文 | Lo等人[ |
本文 | Lo等人[ |
本文 | Lo等人[ |
本文 | ||||
剪贴攻击 | 92.80 | 96.27 | 94.00 | 97.33 | |||||||
常数攻击 | 96.78 | 97.46 | 96.87 | 98.73 | |||||||
拼贴攻击 | 91.80 | 95.31 | 95.70 | 98.44 | |||||||
随机篡改 | 58.64 | 58.64 | 48.47 | 48.47 | 58.02 | 58.02 | 81.48 | 81.48 | |||
平均值 | 85.00 | 84.38 | 86.15 | 94.00 |
Chang CC, Chen KN, Lee CF, Liu LJ. A secure fragile watermarking scheme based on chaos-and-hamming code. Journal of Systems and Software, 2011, 84(9): 1462–1470. [doi: 10.1016/j.jss.2011.02.029]
Huo YR, He HJ, Chen F. Alterable-capacity fragile watermarking scheme with restoration capability. Optics Communications, 2012, 285(7): 1759–1766. [doi: 10.1016/j.optcom.2011.12.044]
Qin C, Ji P, Zhang XP, Dong J, Wang JW. Fragile image watermarking with pixel-wise recovery based on overlapping embedding strategy. Signal Processing, 2017, 138: 280–293. [doi: 10.1016/j.sigpro.2017.03.033]
Su GD, Chang CC, Chen CC. A hybrid-Sudoku based fragile watermarking scheme for image tampering detection. Multimedia Tools and Applications, 2021, 80(8): 12881–12903. [doi: 10.1007/s11042-020-10451-1]
Kim C, Yang CN. Self-embedding fragile watermarking scheme to detect image tampering using AMBTC and OPAP approaches. Applied Sciences, 2021, 11(3): 1146. [doi: 10.3390/app11031146]
张新鹏, 殷赵霞. 多媒体信息隐藏技术. 自然杂志, 2017, 39(2): 87–95. [doi: 10.3969/j.issn.0253-9608.2017.02.002]
Zhang XP, Yin ZX. Data hiding in multimedia. Chinese Journal of Nature, 2017, 39(2): 87–95. (in Chinese with English abstract). [doi: 10.3969/j.issn.0253-9608.2017.02.002]
李晓龙. 图像可逆隐藏综述. 信息安全研究, 2016, 2(8): 729–734.
Li XL. A review on image reversible data hiding. Journal of Information Security Research, 2016, 2(8): 729–734. (in Chinese with English abstract).
Tian J. Reversible data embedding using a difference expansion. IEEE Transactions on Circuits and Systems for Video Technology, 2003, 13(8): 890–896. [doi: 10.1109/TCSVT.2003.815962]
Ni ZC, Shi YQ, Ansari N, Su W. Reversible data hiding. IEEE Transactions on Circuits and Systems for Video Technology, 2006, 16(3): 354–362. [doi: 10.1109/TCSVT.2006.869964]
Thodi DM, Rodríguez JJ. Expansion embedding techniques for reversible watermarking. IEEE Transactions on Image Processing, 2007, 16(3): 721–730. [doi: 10.1109/TIP.2006.891046]
Sachnev V, Kim HJ, Nam J, Suresh S, Shi YQ. Reversible watermarking algorithm using sorting and prediction. IEEE Transactions on Circuits and Systems for Video Technology, 2009, 19(7): 989–999. [doi: 10.1109/TCSVT.2009.2020257]
Coltuc D. Improved embedding for prediction-based reversible watermarking. IEEE Transactions on Information Forensics and Security, 2011, 6(3): 873–882. [doi: 10.1109/TIFS.2011.2145372]
Hu RW, Xiang SJ. CNN prediction based reversible data hiding. IEEE Signal Processing Letters, 2021, 28: 464–468. [doi: 10.1109/LSP.2021.3059202]
罗剑高, 韩国强, 沃焱. 新颖的差值扩展可逆数据隐藏算法. 通信学报, 2016, 37(2): 53–62. [doi: 10.11959/j.issn.1000-436x.2016030]
Luo JG, Han GQ, Yan W. Novel reversible data hiding based on difference expansion. Journal on Communications, 2016, 37(2): 53–62 (in Chinese with English abstract). [doi: 10.11959/j.issn.1000-436x.2016030]
Jia YJ, Yin ZX, Zhang XP, Luo YL. Reversible data hiding based on reducing invalid shifting of pixels in histogram shifting. Signal Processing, 2019, 163: 238–246. [doi: 10.1016/j.sigpro.2019.05.020]
Kim S, Qu XC, Sachnev V, Kim HJ. Skewed histogram shifting for reversible data hiding using a pair of extreme predictions. IEEE Transactions on Circuits and Systems for Video Technology, 2019, 29(11): 3236–3246. [doi: 10.1109/TCSVT.2018.2878932]
Li XL, Li J, Li B, Yang B. High-fidelity reversible data hiding scheme based on pixel-value-ordering and prediction-error expansion. Signal Processing, 2013, 93(1): 198–205. [doi: 10.1016/j.sigpro.2012.07.025]
Peng F, Li XL, Yang B. Improved PVO-based reversible data hiding. Digital Signal Processing, 2014, 25: 255–265. [doi: 10.1016/j.dsp.2013.11.002]
Wu HR, Li XL, Zhao Y, Ni RR. Improved PPVO-based high-fidelity reversible data hiding. Signal Processing, 2020, 167: 107264. [doi: 10.1016/j.sigpro.2019.107264]
Fan GJ, Pan ZB, Gao ED, Gao XY, Zhang XR. Reversible data hiding method based on combining IPVO with bias-added rhombus predictor by multi-predictor mechanism. Signal Processing, 2021, 180: 107888. [doi: 10.1016/j.sigpro.2020.107888]
Ou B, Li XL, Zhao Y, Ni RR, Shi YQ. Pairwise prediction-error expansion for efficient reversible data hiding. IEEE Transactions on Image Processing, 2013, 22(12): 5010–5021. [doi: 10.1109/TIP.2013.2281422]
Ou B, Li XL, Zhang WM, Zhao Y. Improving pairwise PEE via hybrid-dimensional histogram generation and adaptive mapping selection. IEEE Transactions on Circuits and Systems for Video Technology, 2019, 29(7): 2176–2190. [doi: 10.1109/TCSVT.2018.2859792]
Qin JQ, Huang FJ. Reversible data hiding based on multiple two-dimensional histograms modification. IEEE Signal Processing Letters, 2019, 26(6): 843–847. [doi: 10.1109/LSP.2019.2909080]
Zhang T, Li XL, Qi WF, Guo ZM. Location-based PVO and adaptive pairwise modification for efficient reversible data hiding. IEEE Transactions on Information Forensics and Security, 2020, 15: 2306–2319. [doi: 10.1109/TIFS.2019.2963766]
Lo CC, Hu YC. A novel reversible image authentication scheme for digital images. Signal Processing, 2014, 98: 174–185. [doi: 10.1016/j.sigpro.2013.11.028]
Yin ZX, Niu XJ, Zhou ZL, Tang J, Luo B. Improved reversible image authentication scheme. Cognitive Computation, 2016, 8(5): 890–899. [doi: 10.1007/s12559-016-9408-6]
Hong W, Chen MJ, Chen TS. An efficient reversible image authentication method using improved PVO and LSB substitution techniques. Signal Processing: Image Communication, 2017, 58: 111–122. [doi: 10.1016/j.image.2017.07.001]
王泓, 黄方军. 基于可逆信息隐藏技术的认证方案的攻击与改进. 信息安全学报, 2022, 7(1): 56–65. [doi: 10.19363/J.cnki.cn10-1380/tn.2022.01.04]
Wang H, Huang FJ. Attack and improvement of an authentication scheme based on reversible data hiding. Journal of Cyber Security, 2022, 7(1): 56–65. (in Chinese with English abstract). [doi: 10.19363/J.cnki.cn10-1380/tn.2022.01.04]
Howard PG, Kossentini F, Martins B, Forchhammer S, Rucklidge WJ. The emerging JBIG2 standard. IEEE Transactions on Circuits and Systems for Video Technology, 1998, 8(7): 838–848. [doi: 10.1109/76.735380]