Mismatch Steganalysis Method Based on Hybrid Dictionary Learning

Abstract

Abstract: Steganalysis technology is more and more close to machine learning in recent years. Dictionary learning, as a crucial research domain in machine learning, shows the unique advantages in solving numerous practical problems, but its research results in steganalysis domain are limited. To address the mismatch of steganography and embedding rate which are needed to be solved in steganalysis domain, we propose a steganalysis algorithm based on hybrid dictionary learning. In order to encode the differences between cover images and stego images, our algorithm learns specific subdictionaries for two kinds of images, at the same time, a shared dictionary is learned for representing the common content. Except it, we learn a synthesis dictionary for data reconstruction, and an analysis dictionary for classification. The experimental results demonstrate the steganalysis based on hybrid dictionary learning has high performance.

Key words: steganalysis, dictionary learning, hybrid dictionary, mismatch, image

摘要： 近年来，图像隐写分析技术与机器学习的结合越来越紧密.字典学习作为机器学习中的一个重要研究方向，在解决众多实际问题时展现了其独特的优势，但在隐写分析领域中的研究成果却十分有限.针对隐写方法失配和嵌入率失配2个隐写分析领域亟待解决的问题，提出了基于混合字典学习的隐写分析算法.该算法为实现对载体图与隐密图之间的差异进行编码，分别为2类图像学习各自的子字典，同时为表示载体图与隐密图之间的相同之处，单独学习一个共享字典.此外，该算法同时学习出综合型字典和解析型字典，分别以重构数据和编码分类.实验结果表明，所提出的基于混合字典学习的隐写分析方法具有很好的性能.

关键词: 隐写分析, 字典学习, 混合字典, 失配, 图像

郭艳卿. 基于混合字典学习的失配隐写分析方法[J]. 信息安全研究, 2016, 2(9): 821-826.

References

［1］Provos N, Honeyman P. Hide and seek: An introduction to steganography［J］. IEEE Security & Privacy, 2003, 1(3): 3244［2］Johnson N F, Jajodia S. Exploring steganography: Seeing the unseen［J］. IEEE Computer, 1998, 31(2): 2634［3］郭艳卿. 隐密对抗的理论及方法研究［D］. 大连: 大连理工大学, 2009［4］Holub V, Fridrich J. Lowcomplexity features for JPEG steganalysis using undecimated DCT［J］.IEEE Trans on Information Forensics and Security, 2015, 10( 2): 219228［5］Kodovsky J A N. Steganalysis of digital images using rich image representations and ensemble classifiers［D］.New York: State University of New York, 2012［6］Pevny T, Fridrich J. Merging markov and DCT features for multiclass JPEG steganalysis［G］SPIE 6505: Proc of SPIE, Security, Steganography, and Watermarking of Multimedia Contents Conf. San Francisco, CA: SPIE Press, 2007:03_103_13［7］Ker D A, Bas P R, Bhme R. Moving steganography and steganalysis from the laboratory into the real world［C］ Proc of ACM Workshop on Information Hiding and Multimedia Security. New York: ACM, 2013: 4558［8］Barni M, Cancelli G, Esposito A. Forensics aided steganalysis of heterogeneous images［C］ Proc of IEEE Conf on Acoustics Speech and Signal Processing. Piscataway, NJ: IEEE, 2010: 16901693［9］Ker A D, Pevny T. The steganographer is the outlier: Realistic largescale steganalysis［J］. IEEE Trans on Information Forensics and Security, 2014, 9(9): 14241435［10］黄炜, 赵险峰, 盛任农. 基于KFD指标聚类的高隐蔽性JPEG隐写分析［J］. 计算机学报, 2012, 35(9): 19511958［11］Kodovsky J, Sedighi V, Fridrich J. Study of cover source mismatch in steganalysis and ways to mitigate its impact［G］ SPIE 9208: Media Watermarking,Security, and Forensics. San Francisco, CA: SPIE Press, 2014: 0M010M12［12］Lubenko I, Ker A D. Steganalysis with mismatched covers: Do simple classifiers help［C］ Proc of the 13th ACM Workshop on Multimedia and Security. New York: ACM, 2012: 1118［13］Li Xiaofeng, Kong Xiangwei, Wang Bo, et al. Generalized transfer component analysis for mismatched jpeg steganalysis［C］ Proc of IEEE Int Conf on Image Processing (ICIP2013). Piscataway, NJ: IEEE, 2013: 1518［14］Zeng Likai, Kong Xiangwei, Li Ming, et al. JPEG quantization table mismatched steganalysis via discriminative feature transformation［C］ Proc of IS&TSPIE Electronic Imaging. San Francisco, CA: SPIE Press, 2015［15］Zhang D, Yang M, Feng X. Sparse representation or collaborative representation: Which helps face recognition［C］ Proc of IEEE Int Conf Computer Vision (ICCV). Piscataway, NJ: IEEE, 2011: 471478［16］Guo J, Guo Y, Li L, et al. A universal JPEG image steganalysis method based on collaborative representation［C］ Proc of IEEE Int Conf Security Pattern Analysis Cybern (SPAC). Piscataway, NJ: IEEE, 2014: 285289［17］Wright J, Ma Y, Mairal J, et al. Sparse representation for computer vision and pattern recognition［J］. Proc of IEEE, 2010, 98(6): 10311044［18］Gu S, Zhang L, Zuo W, et al. Projective dictionary pair learning for pattern classification［C］ Advances in Neural Information Processing Systems. Cambridge: MIT Press, 2014: 793801［19］Zhou N, Shen Y, Peng J, et al. Learning interrelated visual dictionary for object recognition［C］ Proc of IEEE Conf on Computer Vision Pattern Recognition (CVPR). Piscataway, NJ: IEEE, 2012: 34903497［20］Kong S, Wang D. A dictionary learning approach for classification: Separating the particularity and the commonality［C］ Proc of European Conf on Computer Vision (ECCV). Berlin: Springer, 2012: 186199［21］Gao S, Tsang I W, Ma Y. Learning categoryspecific dictionary and shared dictionary for finegrained image categorization［J］. IEEE Trans on Image Process, 2014: 23(2): 623634［22］Bas P, Filler T, Pevny T. Break our steganographic system: The ins and outs of organizing BOSS［C］ Proc of IEEE Int Conf Information Hiding (IH). Piscataway, NJ: IEEE, 2011: 5970

[1]	. The Review of Information Hiding Technology Based on GAN Image Generation [J]. Journal of Information Security Research, 2019, 5(9): 771-777.
[2]	. The FPGABased Implementation Scheme for Image Shuffling Algorithm —Taking the Arnold Algorithm as an Example [J]. Journal of Information Security Research, 2019, 5(9): 820-827.
[3]	. Multimodal Public Sentiment Analysis Model Based on Local Semantic Information [J]. Journal of Information Security Research, 2019, 5(4): 340-345.
[4]	. Image Steganography Methods from Traditional to Deep Learning [J]. Journal of Information Security Research, 2019, 5(3): 230-235.
[5]	. A Survey of Image Captioning Technology [J]. Journal of Information Security Research, 2019, 5(11): 988-992.
[6]	. The Technical Principles Behind the Toutiao Recommendation System [J]. Journal of Information Security Research, 2019, 5(11): 1027-1032.
[7]	. Research on Reconstruction and Recognition of Human 3D Shapes from Images and Videos [J]. Journal of Information Security Research, 2019, 5(10): 935-943.
[8]	. Research Progress in Single-Pixel Imaging Based on Compressive Sensing [J]. Journal of Information Security Research, 2018, 4(9): 783-791.
[9]	. A Visually Meaningful Image Encryption Algorithm Based on Compressed Sensing [J]. Journal of Information Security Research, 2018, 4(6): 539-547.
[10]	. Operation Platform of Passenger Transportation Safety Based on Vehicle Monitoring Data [J]. Journal of Information Security Research, 2018, 4(5): 473-479.
[11]	. A Static Tagging Method of Malicious Code Family Based on Multi-Feature [J]. Journal of Information Security Research, 2018, 4(4): 322-328.
[12]	. Chaotic Sequence Image Encryption Scheme Based on LFSR State Sequences [J]. Journal of Information Security Research, 2018, 4(4): 336-341.
[13]	. The Research of a Method to Detect Information Hiding in the Private Network [J]. Journal of Information Security Research, 2017, 3(7): 661-667.
[14]	. Steganalysis of the Big Data over Network [J]. Journal of Information Security Research, 2016, 2(6): 490-500.
[15]	. An Overview and Trends on Digital Image Source Forensics [J]. Journal of Information Security Research, 2016, 2(6): 501-511.