收稿日期: 2016-07-31
修回日期: 2016-08-19
网络出版日期: 2016-09-30
基金资助
国家自然科学基金（No.61373151，No.U1536109）；上海市自然科学基金（No.13ZR1415000）资助
Steganalysis of JPEG Image Based on High-Dimensional Feature Regularization
Received date: 2016-07-31
Revised date: 2016-08-19
Online published: 2016-09-30
隐写分析是信息安全的重要内容。为提高JPEG图像隐写的检测能力，建立了散度矩阵的特征谱，提出了一种对载体和含密图像的训练特征进行变换的新方法。首先根据特征谱的分布规律进行建模，划分为3个区域：特征值下降区、平稳区、特征值为零的区域，然后通过白化处理得到白化特征向量，进而对处于3个不同区域的特征向量使用自适应正则化方法。经这几步处理得到特征转移矩阵，也就是输入特征的变换矩阵，最后取变换后特征的前t个向量完成特征选择。并将这些特征数据Fisher线性判决（Fisher linear discriminant，FLD）集成分类器进行训练。结果表明，通过对图像特征进行排序、正则化和选择，FLD集成分类器对JPEG图像隐写的识别准确率得到了提升。
关键词： 特征谱; 正则化; Fisher线性判决; JPEG隐写分析; 特征选择
孙物一, 冯国瑞 . 基于高维特征正则化的JPEG图像隐写分析[J]. 应用科学学报, 2016 , 34(5) : 555 -563 . DOI: 10.3969/j.issn.0255-8297.2016.05.008
Steganalysis plays a vital role in information security. To improve recognition accuracy of JPEG stego-image, we establish a model of scatter matrix eigenspectrum, and purpose a new method to transfer training features of cover and stego images. The eigenspectrum is first modeled by its distribution and divided into three non-overlapping partitions: areas in which the eigenvalue drops rapidly, areas with stable eigenvalues, and areas of zero eigenvalue. The eigenvector is whitened and regularized in the three partitions with adaptive regularization. The result of these steps is a transfer matrix that transfers the input features. The features are now processed and extracted by picking the first t eigenvectors. The processed feature data are then trained with a Fisher linear discriminant (FLD) ensemble classifier. The result shows that recognition accuracy of JPEG stegoimages with the FLD ensemble is improved after the JPEG stego-image features are sorted, regularized and extracted.
[1] Sallee P. Model-based steganography—Digital watermarking [M]. Berlin Heidelberg: Springer, 2003: 154-167.
[2] Fridrich J, Pevný T, Kodovský J. Statistically undetectable JPEG steganography: dead ends challenges, and opportunities [C]//Proceedings of the 9th workshop on Multimedia & security. ACM, 2007: 3-14.
[3] Westfeld A. F5-a steganographic algorithm[C]//Information hiding. Springer Berlin Heidelberg, 2001: 289-302.
[4] Holub V, Fridrich J, Denemark T. Universal distortion function for steganography in an arbitrary domain [J]. EURASIP Journal on Information Security, 2014, 2014(1): 1-13.
[5] Guo L, Ni J, Su W. Using statistical image model for JPEG steganography: uniform embedding revisited [J]. IEEE Transactions on Information Forensics and Security, 2015, 10(12): 2669-2680.
[6] Guo L, Ni J, Shi Y Q. Uniform embedding for efficient JPEG steganography [J]. IEEE Transactions on Information Forensics and Security, 2014, 9(5): 814-825.
[7] Fridrich J, Goljan M, Hogea D. Steganalysis of JPEG images: breaking the F5 algorithm[C]//Information hiding. Springer Berlin Heidelberg, 2002: 310-323.
[8] Kodovský J, Fridrich J. Quantitative steganalysis of LSB embedding in JPEG domain[C]//Proceedings of the 12th ACM workshop on Multimedia and security. ACM, 2010: 187-198.
[9] Kodovský J, Fridrich J. Steganalysis of JPEG images using rich models [C]//Proceeding of the International Society of Optical Engineering, 2012, 8303: 83030A.
[10] Holub V, Fridrich J. Low-complexity features for JPEG steganalysis using undecimated DCT[J]. IEEE Transactions on Information Forensics and Security, 2015, 10(2): 219-228.
[11] Song X, Liu F, Yang C. Steganalysis of adaptive JPEG steganography using 2D Gabor filters[C]//Proceedings of the 3rd ACM Workshop on Information Hiding and Multimedia Security, 2015: 15-23.
[12] Pevný T, Fridrich J. Novelty detection in blind steganalysis [C]//Proceedings of the 10th ACM workshop on Multimedia and security, 2008: 167-176.
[13] Kodovský J, Fridrich J, Holub V. Ensemble classifiers for steganalysis of digital media [J]. IEEE Transactions on Information Forensics and Security, 2012, 7(2): 432-444.
[14] Kodovský J, Fridrich J. Steganalysis in high dimensions: fusing classifiers built on random subspaces 2011: 78800L-78800L-13. [C]//Proceedings of the International Society for Optical Engineering, 2011, 7880: 7880L.
[15] Zhu M, Martinez A M. Subclass discriminant analysis [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2006, 28(8): 1274-1286.
[16] Jiang X, Mandal B, Kot A. Eigenfeature regularization and extraction in face recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2008, 30(3): 383-394.
[17] Cogranne R, Sedighi V, Fridrich J. Is ensemble classifier needed for steganalysis in highdimensional feature spaces? [C]//2015 IEEE International Workshop on Information Forensics and Security, 2015: 1-6.
/
| 〈 |
|
〉 |
