搜索

x

留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

基于深度学习压缩感知与复合混沌系统的通用图像加密算法

陈炜 郭媛 敬世伟

引用本文:
Citation:

基于深度学习压缩感知与复合混沌系统的通用图像加密算法

陈炜, 郭媛, 敬世伟

General image encryption algorithm based on deep learning compressed sensing and compound chaotic system

Chen Wei, Guo Yuan, Jing Shi-Wei
PDF
HTML
导出引用
  • 提出一种适用于灰度图像与RGB格式彩色图像的通用图像加密算法. 利用双线性插值Bilinear与卷积神经网络对图像进行压缩, 再使用二维云模型与Logistic组成的复合混沌系统对压缩图像加解密(滑动置乱与矢量分解), 最后对解密图像进行重构. 重构网络中, 由卷积神经网络与双线性插值Bilinear主要负责重构轮廓信息, 全连接层主要负责重构颜色信息. 实验结果表明, 该基于深度学习压缩感知与复合混沌系统的通用图像加密算法在数据处理质量和计算量上有着很大优势. 由于复合混沌系统有着足够大的密钥空间且将明文哈希值与密钥关联, 可实现一图一密的加密效果, 能有效抵抗暴力攻击与选择明文攻击, 与对比文献相比, 相关系数更接近理想值且信息熵与明文敏感性指标也都在临界值范围内, 其加密算法有着更高的安全性.
    Many image compression and encryption algorithms based on traditional compressed sensing and chaotic systems are time-consuming, have low reconstruction quality, and are suitable only for grayscale images. In this paper, we propose a general image compression encryption algorithm based on a deep learning compressed sensing and compound chaotic system, which is suitable for grayscale images and RGB format color images. Color images can be directly compressed and encrypted, but grayscale images need copying from 1 channel to 3 channels. First, the original image is divided into multiple 3 × 33 × 33 non-overlapping image blocks and the bilinear interpolation Bilinear and convolutional neural network are used to compress the image, so that the compression network has no restriction on the sampling rate and can obtain high-quality compression of image. Then a composite chaotic system composed of a two-dimensional cloud model and Logistic is used to encrypt and decrypt the compressed image (sliding scrambling and vector decomposition), and finally the decrypted image is reconstructed. In the reconstruction network, the convolutional neural network and bilinear interpolation Bilinear are mainly responsible for reconstructing the contour structure information, and the fully connected layer is mainly responsible for reconstructing and combining the color information to reconstruct a high-quality image. For grayscale images, we also need to calculate the average value of the corresponding positions of the 3 channels of the reconstructed image, and change the 3 channels into 1 channel. The experimental results show that the general image encryption algorithm based on deep learning compressed sensing and compound chaos system has great advantages in data processing quality and computational complexity. Although in the network the color images are used for training, the quality of grayscale image reconstruction is still better than that of other algorithms. The image encryption algorithm has a large enough key space and associates the plaintext hash value with the key, which realizes the encryption effect of one image corresponding to one key, thus being able to effectively resist brute force attacks and selective plaintext attacks. Compared with it in the comparison literature, the correlation coefficient is close to an ideal value, and the information entropy and the clear text sensitivity index are also within a critical range, which enhances the confidentiality of the image.
      通信作者: 郭媛, guoyuan171@126.com
    • 基金项目: 国家自然科学基金(批准号: 61872204)、黑龙江省自然科学基金(批准号: F2017029)、黑龙江省省属高等学校基本科研业务费(批准号: 135109236)和研究生创新研究项目(批准号: YJSCX2019042)资助的课题
      Corresponding author: Guo Yuan, guoyuan171@126.com
    • Funds: Project supported by the National Natural Science Foundation of China (Grant No. 61872204), the Natural Science Foundation of Heilongjiang Province, China (Grant No. F2017029), the Fundamental Research Funds for the Higher Education Institutions of Heilongjiang Province, China (Grant No. 135109236), and the Graduate Innovation Research Project, China (Grant No. YJSCX2019042)
    [1]

    Donoho D L 2006 IEEE Trans. Inf. Theory 52 1289Google Scholar

    [2]

    Candes E J, Romberg J, Tao T 2006 IEEE Trans. Inf. Theory 52 489Google Scholar

    [3]

    Candes E J, Wakin M B 2008 IEEE Signal Process. Mag. 25 21Google Scholar

    [4]

    Mousavi A, Patel A B, Baraniuk R G 2015 53rd Annual Allerton Conference on Communication, Control, and Computing Monticello, USA, September 29–October 2, 2015 p1336

    [5]

    练秋生, 富利鹏, 陈书贞, 石保顺 2019 自动化学报 45 2082Google Scholar

    Lian Q S, Fu L P, Chen S Z, Shi B S 2019 Acta Autom. Sin. 45 2082Google Scholar

    [6]

    Kulkarni K, Lohit S, Turaga P, Kerviche R, Ashok A 2016 IEEE Conference on Computer Vision and Pattern Recognition Las Vegas, USA, June 26–30, 2016 p449

    [7]

    Yao H T, Dai F, Zhang SL, Zhang Y D, Tian Q, Xu C S 2019 Neurocomputing 359 483Google Scholar

    [8]

    李静, 向菲, 张军朋 2019 电子设计工程 27 84Google Scholar

    Li J, Xian F, Zhang J P 2019 Int. Electr. Elem. 27 84Google Scholar

    [9]

    Hu X C, Wei L S, Chen W, Chen Q Q, Guo Y 2020 IEEE Access 8 12452Google Scholar

    [10]

    庄志本, 李军, 刘静漪, 陈世强 2020 物理学报 69 040502Google Scholar

    Zhuang Z B, Li J, Liu J Y, Chen S Q 2020 Acta Phys. Sin. 69 040502Google Scholar

    [11]

    Zhang D, Liao X F, Yang B, Zhang Y S 2018 Multim. Tools Appl. 77 2191Google Scholar

    [12]

    石航, 王丽丹 2019 物理学报 68 200501Google Scholar

    Shi H, Wang L D 2019 Acta Phys. Sin. 68 200501Google Scholar

    [13]

    Gong L H, Qiu K D, Deng C Z, Zhou N R 2019 Opt. Laser Technol. 115 257Google Scholar

    [14]

    Qin W, Peng X 2010 Opt. Lett. 35 118Google Scholar

    [15]

    Liu Y N, Niu H Q, Li Z L 2019 Chin. Phys. Lett. 36 044302Google Scholar

    [16]

    Li C B, Yin W T, Jiang H, Zhang Y 2013 Comput. Optim. Appl. 56 507Google Scholar

    [17]

    Dong W S, Shi G M, Li X, Ma Y, Huang F 2014 IEEE Trans. Image Process. 23 3618Google Scholar

    [18]

    Metzler C A, Maleki A, Baraniuk R G 2016 IEEE Trans. Inf. Theory 62 5117Google Scholar

    [19]

    Guo Y, Jing S W, Zhou Y Y, Xu X, Wei L S 2020 IEEE Access 8 9896Google Scholar

    [20]

    Belazi A, El-Latif A A, Belghith S 2016 Signal Process. 128 155Google Scholar

    [21]

    Hua Z Y, Zhou Y C, Pun C M, Chen C 2015 Inf. Sci. 297 80Google Scholar

    [22]

    Wu Y, Zhou Y C, Saveriades G, Agaian S S, Noonan J P, Natarajan P 2013 Inf. Sci. 222 323Google Scholar

  • 图 1  滑动置乱流程图

    Fig. 1.  Flow chart for scrambling.

    图 2  彩色图像的压缩重构网络(采样率MR = 0.18)

    Fig. 2.  Color image compression and reconstruction network, sampling rate MR = 0.18.

    图 3  BCNN模型与全连接层分别重构的Lena和Sea图像 (a) Lena (灰度); (b) Lena (彩色); (c) Sea (灰度); (d) Sea (彩色)

    Fig. 3.  Lena and Sea images reconstructed from BCNN model and fully connected layer respectively: (a) Lena (gray); (b) Lena (color); (c) Sea (gray); (d) Sea (color).

    图 4  整个算法流程图

    Fig. 4.  Entire algorithm flow chart.

    图 5  Lena (gray)图像明文与密文在水平、竖直、斜线3个方向的相关分布图 (a) 明文图像的相关分布图; (b)密文图像的相关分布图

    Fig. 5.  Correlation distribution of the plaintext and ciphertext in the horizontal, vertical and oblique directions of Lena (gray) images: (a) Correlation distribution of plaintext; (b) correlation distribution of ciphertext.

    图 6  在密文图像上添加高斯噪声或椒盐噪声后重构出的Lena图像 (a), (b)使用的网络在训练时没有添加高斯噪声; (c), (d)使用的网络在训练时添加了强度为0.10的高斯噪声

    Fig. 6.  Lena image reconstructed by adding Gaussian noise or salt and pepper noise to the ciphertext image: The networks used in (a) and (b) did not add Gaussian noise during training; the networks used in (c) and (d) are trained with Gaussian noise with an intensity of 0.10.

    图 7  Lena图像的密文被剪切后的重构结果 (a)剪切不同大小后的密文; (b), (c)使用的网络在训练时没有添加高斯噪声; (d), (e)使用的网络在训练时添加了强度为0.10的高斯噪声

    Fig. 7.  Reconstruction result after ciphertext cut of Lena image: (a) Cut ciphertexts of different sizes; the networks used in (b) and (c) did not add Gaussian noise during training; the networks used in (d) and (e) are trained with Gaussian noise with an intensity of 0.10.

    图 8  选择明文攻击效果图

    Fig. 8.  Effect pictures of chosen plaintext attack.

    图 9  Lena (gray), Peppers (gray), Lena (color), Peppers (color)图像在明文与密文上的直方图 (a)明文直方图; (b)密文直方图

    Fig. 9.  Histograms of Lena (gray), Peppers (gray), Lena (color), Peppers (color) images in plain text and ciphertext: (a) Plain text histogram; (b) ciphertext histogram.

    表 1  重构的灰度图像在不同算法、不同采样率下的PSNR

    Table 1.  PSNR of reconstructed gray images under different algorithms and different sampling rates.

    采样率算法LenaMonarchFlinstones平均PSNR
    0.25TVAL328.6727.7724.0527.84
    NLR-CS29.3925.9122.4328.05
    D-AMP28.0026.3925.0228.17
    ReconNet26.5424.3122.4525.54
    DR2-Net29.4227.9526.1928.66
    MSRNet30.2128.9026.6729.48
    FCLBCNN31.0929.9727.5729.71
    0.10TVAL324.1621.1618.8822.84
    NLR-CS15.3014.5912.1814.19
    D-AMP22.5119.0016.9421.14
    ReconNet23.8321.1018.9222.68
    DR2-Net25.3923.1021.0924.32
    MSRNet26.2823.9821.7225.16
    FCLBCNN26.9324.5822.0825.41
    0.04TVAL319.4616.7314.8818.39
    NLR-CS11.6111.628.9610.58
    D-AMP16.5214.5712.9315.49
    ReconNet21.2818.1916.3019.99
    DR2-Net22.1318.9316.9320.80
    MSRNet22.7619.2617.2821.41
    FCLBCNN23.3319.5917.1721.51
    0.01TVAL311.8711.099.7511.31
    NLR-CS5.956.384.455.30
    D-AMP5.736.204.335.19
    ReconNet17.8715.3913.9617.27
    DR2-Net17.9715.3314.0117.44
    MSRNet18.0615.4113.8317.54
    FCLBCNN18.1215.6313.9017.62
    下载: 导出CSV

    表 2  在BSD500测试集上不同算法、不同采样率下的平均PSNR和平均SSIM

    Table 2.  Mean PSNR and SSIM of different algorithms and different sampling rates on the BSD500 test set.

    算法MR = 0.08MR = 0.10MR = 0.18MR = 0.25MR = 0.53
    PSNRSSIMPSNRSSIMPSNRSSIMPSNRSSIMPSNRSSIM
    ReconNet23.280.612125.480.7241
    DR2-Net24.260.660327.560.7961
    MSRNet24.730.683727.930.8121
    FCLBCNN (gray)24.830.705628.190.840032.870.9392
    FCLBCNN (color)27.940.825232.070.9279
    下载: 导出CSV

    表 3  Lena图像在各阶段的效果(采样率MR = 0.53 (灰度), 0.18 (彩色))

    Table 3.  Lena image effects at various stages, sampling rate MR = 0.53 (gray), 0.18 (color).

    原图采样率压缩图像置乱图像密文图像重构图像PSNRSSIM
    0.5336.43870.9715
    0.1832.55160.9456
    下载: 导出CSV

    表 4  不同加密算法的相关系数比较

    Table 4.  Comparison of correlation coefficients of different encryption algorithms.

    测试图像方向明文(gray)密文
    本文(gray)本文(color)文献[20]文献[21]
    Lena水平0.93960.0010–0.0024–0.00480.0011
    竖直0.9639–0.00660.0012–0.01120.0098
    斜线0.9189–0.00390.0035–0.0045–0.0227
    Peppers水平0.9769–0.0004–0.0023–0.00560.0071
    竖直0.97720.00890.0063–0.0162–0.0065
    斜线0.9625–0.00770.0004–0.0113–0.0165
    平均值水平0.0003–0.0024–0.00520.0041
    竖直0.00120.0038–0.01370.0017
    斜线–0.00580.0020–0.0079–0.0196
    下载: 导出CSV

    表 5  不同加密算法得到的信息熵的比较

    Table 5.  Comparison of the entropy obtained by different encryption algorithms.

    测试图像明文(gray)密文
    本文(gary)本文(color)文献[12]
    Lena7.30357.99497.99447.9544
    Peppers7.43447.99567.99527.9633
    下载: 导出CSV

    表 6  不同加密算法得到的局部信息熵比较

    Table 6.  Comparison of the local entropy obtained by different encryption algorithms.

    测试图像局部信息熵(gray/color)临界值
    $\begin{array}{l} u_{0.05}^{* - } = 7.9019 \\ u_{0.05}^{* + } = 7.9030 \end{array}$$\begin{array}{l} u_{0.01}^{* - } = 7.9017 \\ u_{0.01}^{* + } = 7.9032 \end{array}$$\begin{array}{l} u_{0.001}^{* - } = 7.9015 \\ u_{0.001}^{* + } = 7.9034 \end{array}$
    Lena7.9024/7.9027PassPassPass
    Peppers7.9027/7.9023PassPassPass
    下载: 导出CSV

    表 7  不同加密算法得到的NPCR比较

    Table 7.  Comparison of NPCR obtained by different encryption algorithms.

    测试图像NPCR (gray/color)NPCR理论临界值
    $N_{0.05}^* = 99.5693\% $$N_{0.01}^* = 99.5527\% $$N_{0.001}^* = 99.5341\% $
    Lena0.9960/0.9961PassPassPass
    Lena[12]0.9954/—FailFailPass
    Lena[20]0.9962/—PassPassPass
    Peppers0.9959/0.9957PassPassPass
    Peppers[12]0.9944/—FailFailFail
    Peppers[20]0.9963/—PassPassPass
    下载: 导出CSV

    表 8  不同加密算法得到的UACI比较

    Table 8.  Comparison of UACI obtained by different encryption algorithms.

    测试图像UACI (gray/color)UACI理论临界值
    $\begin{array}{l} u_{0.05}^{* - } = 33.2824\% \\ u_{0.05}^{* + } = 33.6447\% \end{array}$$\begin{array}{l} u_{0.01}^{* - } = 33.2255\% \\ u_{0.01}^{* + } = 33.7016\% \end{array}$$\begin{array}{l} u_{0.001}^{* - } = 33.1594\% \\ u_{0.001}^{* + } = 33.7677\% \end{array}$
    Lena0.3352/0.3357PassPassPass
    Lena[12]0.3303/—FailFailFail
    Lena[20]0.3370/—FailPassPass
    Peppers0.3333/0.3331PassPassPass
    Peppers[12]0.3305/—FailFailFail
    Peppers[20]0.3369/—FailPassPass
    下载: 导出CSV

    表 9  本文压缩加密算法与在原图上直接使用本文加密算法的耗时对比

    Table 9.  Time-consuming comparison that the compression encryption algorithm of this article and the encryption algorithm of this article directly used on the original image.

    图像大小压缩重构(gray/color)加解密(gray/color)总时间(gray/color)编程工具平台
    256 × 2560.21/0.200.66/0.650.87/0.85Pycharm + Pytorchi5-8500 CPU
    0.93/2.810.93/2.81
    512 × 5120.91/0.892.51/2.513.42/3.40
    3.89/11.963.89/11.96
    1024 × 10244.81/4.629.40/9.4214.21/14.04
    15.84/48.5115.84/48.51
    下载: 导出CSV
  • [1]

    Donoho D L 2006 IEEE Trans. Inf. Theory 52 1289Google Scholar

    [2]

    Candes E J, Romberg J, Tao T 2006 IEEE Trans. Inf. Theory 52 489Google Scholar

    [3]

    Candes E J, Wakin M B 2008 IEEE Signal Process. Mag. 25 21Google Scholar

    [4]

    Mousavi A, Patel A B, Baraniuk R G 2015 53rd Annual Allerton Conference on Communication, Control, and Computing Monticello, USA, September 29–October 2, 2015 p1336

    [5]

    练秋生, 富利鹏, 陈书贞, 石保顺 2019 自动化学报 45 2082Google Scholar

    Lian Q S, Fu L P, Chen S Z, Shi B S 2019 Acta Autom. Sin. 45 2082Google Scholar

    [6]

    Kulkarni K, Lohit S, Turaga P, Kerviche R, Ashok A 2016 IEEE Conference on Computer Vision and Pattern Recognition Las Vegas, USA, June 26–30, 2016 p449

    [7]

    Yao H T, Dai F, Zhang SL, Zhang Y D, Tian Q, Xu C S 2019 Neurocomputing 359 483Google Scholar

    [8]

    李静, 向菲, 张军朋 2019 电子设计工程 27 84Google Scholar

    Li J, Xian F, Zhang J P 2019 Int. Electr. Elem. 27 84Google Scholar

    [9]

    Hu X C, Wei L S, Chen W, Chen Q Q, Guo Y 2020 IEEE Access 8 12452Google Scholar

    [10]

    庄志本, 李军, 刘静漪, 陈世强 2020 物理学报 69 040502Google Scholar

    Zhuang Z B, Li J, Liu J Y, Chen S Q 2020 Acta Phys. Sin. 69 040502Google Scholar

    [11]

    Zhang D, Liao X F, Yang B, Zhang Y S 2018 Multim. Tools Appl. 77 2191Google Scholar

    [12]

    石航, 王丽丹 2019 物理学报 68 200501Google Scholar

    Shi H, Wang L D 2019 Acta Phys. Sin. 68 200501Google Scholar

    [13]

    Gong L H, Qiu K D, Deng C Z, Zhou N R 2019 Opt. Laser Technol. 115 257Google Scholar

    [14]

    Qin W, Peng X 2010 Opt. Lett. 35 118Google Scholar

    [15]

    Liu Y N, Niu H Q, Li Z L 2019 Chin. Phys. Lett. 36 044302Google Scholar

    [16]

    Li C B, Yin W T, Jiang H, Zhang Y 2013 Comput. Optim. Appl. 56 507Google Scholar

    [17]

    Dong W S, Shi G M, Li X, Ma Y, Huang F 2014 IEEE Trans. Image Process. 23 3618Google Scholar

    [18]

    Metzler C A, Maleki A, Baraniuk R G 2016 IEEE Trans. Inf. Theory 62 5117Google Scholar

    [19]

    Guo Y, Jing S W, Zhou Y Y, Xu X, Wei L S 2020 IEEE Access 8 9896Google Scholar

    [20]

    Belazi A, El-Latif A A, Belghith S 2016 Signal Process. 128 155Google Scholar

    [21]

    Hua Z Y, Zhou Y C, Pun C M, Chen C 2015 Inf. Sci. 297 80Google Scholar

    [22]

    Wu Y, Zhou Y C, Saveriades G, Agaian S S, Noonan J P, Natarajan P 2013 Inf. Sci. 222 323Google Scholar

  • [1] 刘鸿江, 刘逸飞, 谷付星. 基于深度学习的微纳光纤自动制备系统. 物理学报, 2024, 0(0): 0-0. doi: 10.7498/aps.73.20240171
    [2] 欧秀娟, 肖奕. RNA扭转角预测的深度学习方法. 物理学报, 2023, 72(24): 248703. doi: 10.7498/aps.72.20231069
    [3] 孙涛, 袁健美. 基于深度学习原子特征表示方法的Janus过渡金属硫化物带隙预测. 物理学报, 2023, 72(2): 028901. doi: 10.7498/aps.72.20221374
    [4] 战庆亮, 葛耀君, 白春锦. 基于深度学习的流场时程特征提取模型. 物理学报, 2022, 71(7): 074701. doi: 10.7498/aps.71.20211373
    [5] 曹海燕, 叶震宇. 基于压缩感知理论的大规模MIMO系统下行信道估计中的导频优化理论分析与算法设计. 物理学报, 2022, 71(5): 050101. doi: 10.7498/aps.71.20211504
    [6] 张瑶, 张云波, 陈立. 基于深度学习的光学表面杂质检测. 物理学报, 2021, 70(16): 168702. doi: 10.7498/aps.70.20210403
    [7] 凡洪剑, 李江, 王丽华, 樊春海, 柳华杰. 基于DNA折纸模板的铁原子阵列构建及其信息加密应用. 物理学报, 2021, 70(6): 068702. doi: 10.7498/aps.70.20201438
    [8] 干红平, 张涛, 花燚, 舒君, 何立军. 基于双极性混沌序列的托普利兹块状感知矩阵. 物理学报, 2021, 70(3): 038402. doi: 10.7498/aps.70.20201475
    [9] 曹海燕, 叶震宇. 基于压缩感知理论的大规模MIMO系统下行信道估计中的导频优化理论分析与算法设计. 物理学报, 2021, (): . doi: 10.7498/aps.70.20211504
    [10] 赵智鹏, 周双, 王兴元. 基于深度学习的新混沌信号及其在图像加密中的应用. 物理学报, 2021, 70(23): 230502. doi: 10.7498/aps.70.20210561
    [11] 徐昭, 周昕, 白星, 李聪, 陈洁, 倪洋. 基于深度学习的相位截断傅里叶变换非对称加密系统攻击方法. 物理学报, 2021, 70(14): 144202. doi: 10.7498/aps.70.20202075
    [12] 胡耀华, 刘艳, 穆鸽, 秦齐, 谭中伟, 王目光, 延凤平. 基于多模光纤散斑的压缩感知在光学图像加密中的应用. 物理学报, 2020, 69(3): 034203. doi: 10.7498/aps.69.20191143
    [13] 郎利影, 陆佳磊, 于娜娜, 席思星, 王雪光, 张雷, 焦小雪. 基于深度学习的联合变换相关器光学图像加密系统去噪方法. 物理学报, 2020, 69(24): 244204. doi: 10.7498/aps.69.20200805
    [14] 石航, 王丽丹. 一种基于压缩感知和多维混沌系统的多过程图像加密方案. 物理学报, 2019, 68(20): 200501. doi: 10.7498/aps.68.20190553
    [15] 郭静波, 李佳文. 二进制信号的混沌压缩测量与重构. 物理学报, 2015, 64(19): 198401. doi: 10.7498/aps.64.198401
    [16] 柴水荣, 郭立新. 基于压缩感知的一维海面与二维舰船复合后向电磁散射快速算法研究. 物理学报, 2015, 64(6): 060301. doi: 10.7498/aps.64.060301
    [17] 李广明, 吕善翔. 混沌信号的压缩感知去噪. 物理学报, 2015, 64(16): 160502. doi: 10.7498/aps.64.160502
    [18] 李龙珍, 姚旭日, 刘雪峰, 俞文凯, 翟光杰. 基于压缩感知超分辨鬼成像. 物理学报, 2014, 63(22): 224201. doi: 10.7498/aps.63.224201
    [19] 郭静波, 汪韧. 基于混沌序列和RIPless理论的循环压缩测量矩阵的构造. 物理学报, 2014, 63(19): 198402. doi: 10.7498/aps.63.198402
    [20] 刘 强, 方锦清, 赵耿, 李永. 基于FPGA技术的混沌加密系统研究. 物理学报, 2012, 61(13): 130508. doi: 10.7498/aps.61.130508
计量
  • 文章访问数:  6873
  • PDF下载量:  276
  • 被引次数: 0
出版历程
  • 收稿日期:  2020-06-29
  • 修回日期:  2020-08-10
  • 上网日期:  2020-12-07
  • 刊出日期:  2020-12-20

/

返回文章
返回