JPEGCNN: A Transform Domain Steganalysis Model Based on Convolutional Neural Network

Convolutional Neural Network (CNN) has gained an overwhelming advantage in many fields of pattern recognition. Both excellent data learning ability and automatic feature extraction ability of CNN are urgently needed in image steganalysis. However, the application of CNN in image steganalysis is still in its infancy, especially in the field of JPEG steganalysis. In this paper, a steganalysis model based on CNN in gray image transform domain is proposed, which is called JPEGCNN. At the same time, on the basis of JPEGCNN, JPEGCNN is extended to the transform domain of color image by researching and designing different methods of feature extraction. RGBMERGE-JPEGCNN and RGBADD-JPEGCNN are proposed respectively, which make up for the lack of research on steganalysis model based on convolution neural network in the transform domain of color image. Experiments show that JPEGCNN, RGBMERGE-JPEGCNN and RGBADD-JPEGCNN proposed in this paper have good detection ability for steganography algorithm in transform domain.

[1]  Jessica J. Fridrich,et al.  Random Projections of Residuals for Digital Image Steganalysis , 2013, IEEE Transactions on Information Forensics and Security.

[2]  Jiwu Huang,et al.  Large-Scale JPEG Image Steganalysis Using Hybrid Deep-Learning Framework , 2016, IEEE Transactions on Information Forensics and Security.

[3]  Jessica J. Fridrich,et al.  Rich Models for Steganalysis of Digital Images , 2012, IEEE Transactions on Information Forensics and Security.

[4]  Baoyu Ma,et al.  A Method for Improving CNN-Based Image Recognition Using DCGAN , 2018 .

[5]  Jiangqun Ni,et al.  Deep Learning Hierarchical Representations for Image Steganalysis , 2017, IEEE Transactions on Information Forensics and Security.

[6]  Xiaojun Wang,et al.  Matrix Coding-Based Quantum Image Steganography Algorithm , 2019, IEEE Access.

[7]  Tomás Pevný,et al.  "Break Our Steganographic System": The Ins and Outs of Organizing BOSS , 2011, Information Hiding.

[8]  Jessica J. Fridrich,et al.  Low-Complexity Features for JPEG Steganalysis Using Undecimated DCT , 2015, IEEE Transactions on Information Forensics and Security.

[9]  Jessica J. Fridrich,et al.  Phase-aware projection model for steganalysis of JPEG images , 2015, Electronic Imaging.

[10]  Bin Li,et al.  Large-scale JPEG steganalysis using hybrid deep-learning framework , 2016, ArXiv.

[11]  Wei Lu,et al.  Binary Image Steganalysis Based on Distortion Level Co-occurrence Matrix , 2018 .

[12]  Yoshua Bengio,et al.  Understanding the difficulty of training deep feedforward neural networks , 2010, AISTATS.