结合图像加密与深度学习的高容量图像隐写算法

doi:10.11959/j.issn.1000-436x.2021134

摘要/Abstract

摘要：

针对基于深度学习的高容量图像隐写方案存在的载体图像和含密图像的残差图像会暴露秘密图像的问题，提出了结合图像加密和深度学习的高容量图像隐写算法。该算法设计使用了一种图像特征提取方法，使得从载体图像中提取的特征与从含密图像中提取的特征是一致的。发送方在图像隐写前，从载体图像中提取特征作为密钥，用来加密秘密图像。提取方提取加密过的秘密图像后，从含密图像中提取特征作为密钥，用来解密秘密图像。实验结果表明，攻击者无法从残差图像中发现秘密图像的信息，且密钥传递的频率更低，算法安全性得到了提升。

关键词: 深度学习, 高容量, 图像加密, 图像隐写

Abstract:

Aiming at the problem that the residual image of cover image and carrier image in the high-capacity image steganography scheme based on deep learning will expose the secret image, a high-capacity image steganography scheme combining image encryption and deep learning was proposed.An image feature extraction method was used, so that the features extracted from the cover image were consistent with the features extracted from the carrier image.Before the image steganography, the sender extracted features from the cover image as a key to encrypt the secret image.After the extractor extracted the encrypted secret image, the features were extracted from the carrier image as a key to decrypt the secret image.The experimental results show that the attacker cannot find the information of the secret image from the residual image, and the frequency of key transmission is lower, and the security of the algorithm is improved.

Key words: deep learning, high capacity, image encryption, image steganography

中图分类号:

TP309.2

杨晓元, 毕新亮, 刘佳, 黄思远. 结合图像加密与深度学习的高容量图像隐写算法[J]. 通信学报, 2021, 42(9): 96-105.

Xiaoyuan YANG, Xinliang BI, Jia LIU, Siyuan HUANG. High-capacity image steganography algorithm combining image encryption and deep learning[J]. Journal on Communications, 2021, 42(9): 96-105.

图/表 19

图1

图2

图3

图4

图5

表1

表2

图6

图7

表3

表4

图8

图9

图10

图11

图12

表5

图13

图14

参考文献 29

[1]	PEVNY T , FILLER T , BAS P . Using high-dimensional image models to perform highly undetectable steganography[C]// Information Hiding. Berlin:Springer, 2010: 161-177.
[2]	HOLUB V , FRIDRICH J . Designing steganographic distortion using directional filters[C]// Proceedings of 2012 IEEE International Workshop onInformation Forensics and Security (WIFS). Piscataway:IEEE Press, 2012: 234-239.
[3]	HOLUB V , FRIDRICH J , DENEMARK T . Universal distortion function for steganography in an arbitrary domain[J]. EURASIP Journal on Information Security, 2014(1): 1-13.
[4]	HICSONMEZ S , SAMET N , AKBAS E ,et al. GANILLA:generative adversarial networks for image to illustration translation[J]. Image and Vision Computing, 2020,95:103886.
[5]	HE K , GKIOXARI G , DOLLáR P ,et al. Mask R-CNN[C]// Proceedings of 2017 IEEE International Conference on Computer Vision (ICCV). Piscataway:IEEE Press, 2017: 2980-2988.
[6]	SZEGEDY C , IOFFE S , VANHOUCKE V ,et al. Inception-v4,inception-ResNet and the impact of residual connections on learning[J]. arXiv Preprint,arXiv:1602.07261, 2016.
[7]	CHOI Y , CHOI M , KIM M ,et al. StarGAN:unified generative adversarial networks for multi-domain image-to-image translation[C]// Proceedings of 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2018: 8789-8797.
[8]	GOODFELLOW I J , POUGET-ABADIE J ,, MIRZA M ,et al. Generative adversarial nets[C]// Proceedings of the 27th International Conference on Neural Information Processing Systems. Cambridge:MIT Press, 2014: 2672-2680.
[9]	TANG W X , TAN S Q , LI B ,et al. Automatic steganographic distortion learning using a generative adversarial network[J]. IEEE Signal Processing Letters, 2017,24(10): 1547-1551.
[10]	BALUJA S , . Hiding images in plain sight:deep steganography[C]// Advances in Neural Information Processing Systems 30 (NIPS 2017).[S.n.:s.l.], 2017: 2069-2079.
[11]	DUAN X T , GUO D D , LIU N ,et al. A new high capacity image steganography method combined with image elliptic curve cryptography and deep neural network[J]. IEEE Access, 2020,8: 25777-25788.
[12]	SHARMA K , AGGARWAL A , SINGHANIA T ,et al. Hiding data in images using cryptography and deep neural network[J]. Journal of Artificial Intelligence and Systems, 2019,1(1): 143-162.
[13]	RAHIM R , NADEEM S . End-to-end trained CNN encoder-decoder networks for image steganography[C]// Proceedings of the European Conference on Computer Vision (ECCV).[S.n.:s.l.], 2018: 723-729.
[14]	ZHANG R , DONG S Q , LIU J Y . Invisible steganography via generative adversarial networks[J]. Multimedia Tools and Applications, 2019,78(7): 8559-8575.
[15]	SZEGEDY C , VANHOUCKE V , IOFFE S ,et al. Rethinking the inception architecture for computer vision[C]// Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway:IEEE Press, 2016: 2818-2826.
[16]	DUAN X T , JIA K , LI B X ,et al. Reversible image steganography scheme based on a U-net structure[J]. IEEE Access, 2019,7: 9314-9323.
[17]	RONNEBERGER O , FISCHER P , BROX T . U-net:convolutional networks for biomedical image segmentation[C]// International Conference on Medical Image Computing and Computer-Assisted Intervention. Berlin:Springer, 2015: 234-241.
[18]	BADRINARAYANAN V , KENDALL A , CIPOLLA R . SegNet:a deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017,39(12): 2481-2495.
[19]	LIU Q , XIANG X Y , QIN J H ,et al. Coverless image steganography based on DenseNet feature mapping[J]. EURASIP Journal on Image and Video Processing, 2020,2020(1): 39.
[20]	ZHOU Z L , CAO Y , WANG M M ,et al. Faster-RCNN based robust coverless information hiding system in cloud environment[J]. IEEE Access, 2019,7: 179891-179897.
[21]	ZHOU Z , SUN H , HARIT R ,et al. Coverless image steganography without embedding[C]// International Conference on Cloud Computing and Security. Berlin:Springer, 2015: 123-132.
[22]	HE K M , ZHANG X Y , REN S Q ,et al. Deep residual learning for image recognition[C]// Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway:IEEE Press, 2016: 770-778.
[23]	SIMONYAN K , ZISSERMAN A . Very deep convolutional networks for large-scale image recognition[J]. arXiv Preprint,arXiv:1409.1556, 2015.
[24]	DUAN X T , NAO L , GOU M X ,et al. High-capacity image steganography based on improved FC-DenseNet[J]. IEEE Access, 2020,8: 170174-170182.
[25]	BOEHM B . StegExpose - a tool for detecting LSB steganography[J]. arXiv Preprint,arXiv:1410.6656, 2014.
[26]	竺乐庆, 郭钰, 莫凌强 ,等. DGANS:基于双重生成式对抗网络的稳健图像隐写模型[J]. 通信学报, 2020,41(1): 125-133.
	ZHU L Q , GUO Y , MO L Q ,et al. DGANS:robustness image steganography model based on double GAN[J]. Journal on Communications, 2020,41(1): 125-133.
[27]	SHI H C , DONG J , WANG W ,et al. SSGAN:secure steganography based on generative adversarial networks[C]// Advances in Multimedia Information Processing – PCM 2017. Berlin:Springer, 2018: 534-544.
[28]	ZHANG K A , CUESTA-INFANTE A ,, XU L ,et al. SteganoGAN:high capacity image steganography with GANs[J]. arXiv Preprint,arXiv:1901.03892, 2019.
[29]	QIN J H , WANG J , TAN Y ,et al. Coverless image steganography based on generative adversarial network[J]. Mathematics, 2020,8(9): 1394.

层结构	输入尺寸	输出尺寸
4×4×64卷积+LeakyReLU	256×256×6	128×128×64
4×4×128卷积+BN+LeakyReLU	128×128×64	64×64×128
4×4×256卷积+BN+LeakyReLU	64×64×128	32×32×256
4×4×512卷积+BN+LeakyReLU	32×32×256	16×16×512
4×4×512卷积+BN+LeakyReLU	16×16×512	8×8×512
4×4×512卷积+BN+LeakyReLU	8×8×512	4×4×512
4×4×512卷积+ReLU	4×4×512	2×2×512
4×4×512反卷积+BN+ReLU	2×2×512	4×4×512
4×4×512反卷积+BN+ReLU	4×4×1024	8×8×512
4×4×512反卷积+BN+ReLU	8×8×1024	16×16×512
4×4×256反卷积+BN+ReLU	16×16×1024	32×32×256
4×4×128反卷积+BN+ReLU	32×32×512	64×64×128
4×4×64反卷积+BN+ReLU	64×64×256	128×128×64
4×4×3反卷积+ Sigmoid	128×128×128	256×256×3

层结构	输入尺寸	输出尺寸
3×3×64卷积+BN+ReLU	256×256×3	256×256×64
3×3×128卷积+BN+ReLU	256×256×64	256×256×128
3×3×256卷积+BN+ReLU	256×256×128	256×256×256
3×3×128卷积+BN+ReLU	256×256×256	256×256×128
3×3×64卷积+BN+ReLU	256×256×128	256×256×64
3×3×3卷积+sigmoid	256×256×64	256×256×3

算法	秘密图像尺寸	载体图像尺寸	容量/bpp
HUGO^[1]	32×32	32×32	0.1~0.4
S-UNIWARD^[3]	64×64	64×64	0.4
SSGAN^[27]	64×64	204×204	0.4
ISGAN^[14]	256×256	256×256	8
Rahim^[13]	300×300	300×300	8
DGANS^[26]	256×256	256×256	8
Duan^[11]	256×256	256×256	24
Sharma^[12]	256×256	256×256	24
本文算法	256×256	256×256	24