Automatic Steganographic Distortion Learning Using a Generative Adversarial Network

被引:284
作者
Tang, Weixuan [1 ,2 ]
Tan, Shunquan [2 ,3 ]
Li, Bin [2 ,4 ]
Huang, Jiwu [2 ,4 ]
机构
[1] Sun Yat Sen Univ, Sch Informat Sci & Technol, Guangzhou 510275, Guangdong, Peoples R China
[2] Shenzhen Key Lab Media Secur, Shenzhen 518060, Guangdong, Peoples R China
[3] Shenzhen Univ, Coll Comp Sci & Software Engn, Shenzhen 518060, Peoples R China
[4] Shenzhen Univ, Coll Informat Engn, Shenzhen 518060, Peoples R China
关键词
Embedding change probabilities; generative adversarial network (GAN); steganalysis; steganography;
D O I
10.1109/LSP.2017.2745572
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
Generative adversarial network has shown to effectively generate artificial samples indiscernible fromtheir real counterparts with a united framework of two subnetworks competing against each other. In this letter, we first propose an automatic steganographic distortion learning framework using a generative adversarial network, which is composed of a steganographic generative subnetwork and a steganalytic discriminative subnetwork. Via alternately training these two oppositional subnetworks, our proposed framework can automatically learn embedding change probabilities for every pixel in a given spatial cover image. The learnt embedding change probabilities can then be converted to embedding distortions, which can be adopted in the existing framework of minimal-distortion embedding. Under this framework, the distortion function is directly related to the undetectability against the oppositional evolving steganalyzer. Experimental results show that with adversarial learning, our proposed framework can effectively evolve fromnearly naive random +/- 1 embedding at the beginning to much more advanced content-adaptive embedding which tries to embed secret bits in textural regions. The security performance is also steadily improved with increasing training iterations.
引用
收藏
页码:1547 / 1551
页数:5
相关论文
共 30 条
[1]  
[Anonymous], ARXIV161103233
[2]  
[Anonymous], P SPIE
[3]  
[Anonymous], 2016, Adv. Neural Inf. Process. Syst.
[4]  
[Anonymous], 2014, Signal and Information Processing Association Annual Summit and Conference (APSIPA), 2014 Asia-Pacific, DOI [10.1109/APSIPA.2014, 10.1109/APSIPA.2014.7041565]
[5]  
Bas Patrick, 2011, Information Hiding. 13th International Conference, IH 2011. Revised Selected Papers, P59, DOI 10.1007/978-3-642-24178-9_5
[6]  
Denemark T., 2014, P IS T SPIE EL IM 20
[7]  
Denton E. L., 2015, ADV NEURAL INFORM PR, P1486, DOI DOI 10.5555/2969239.2969405
[8]   Minimizing Additive Distortion in Steganography Using Syndrome-Trellis Codes [J].
Filler, Tomas ;
Judas, Jan ;
Fridrich, Jessica .
IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY, 2011, 6 (03) :920-935
[9]   Rich Models for Steganalysis of Digital Images [J].
Fridrich, Jessica ;
Kodovsky, Jan .
IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY, 2012, 7 (03) :868-882
[10]   Generative Adversarial Networks [J].
Goodfellow, Ian ;
Pouget-Abadie, Jean ;
Mirza, Mehdi ;
Xu, Bing ;
Warde-Farley, David ;
Ozair, Sherjil ;
Courville, Aaron ;
Bengio, Yoshua .
COMMUNICATIONS OF THE ACM, 2020, 63 (11) :139-144