Advertisement

Detection of double JPEG compression using modified DenseNet model

  • Ximei Zeng
  • Guorui Feng
  • Xinpeng Zhang
Article
  • 54 Downloads

Abstract

With the increasing tendency of the tempering of JPEG images, development of methods detecting image forgery is of great importance. In many cases, JPEG image forgery is usually accompanied with double JPEG compression, leaving no visual traces. In this paper, a modified version of DenseNet (densely connected convolutional networks) is proposed to accomplish the detection task of primary JPEG compression among double compressed images. A special filtering layer in the front of the network contains typically selected filtering kernels that can help the network following to discriminating the images more easily. As shown in the results, the network has achieved great improvement compared to the-state-of-the-art method especially on the classification accuracy among images with lower quality factors.

Keywords

Double JPEG compression DenseNet Filtering layer F-LDA Residual noises 

Notes

Acknowledgements

This work was supported by the National Natural Science Foundation of China under Grants (U1536109, U1636206, 61525203, 61373151, 61472235).

References

  1. 1.
    Amerini I, Uricchio T, Ballan L, Caldelli R (2017) Localization of JPEG double compression through multi-domain convolutional neural networks. In: IEEE Conference on computer vision and pattern recognition workshops on media forensicsGoogle Scholar
  2. 2.
    Amerini I, Uricchio T, Caldelli R (2017) Tracing images back to their social network of origin: a CNN-based approach. In: IEEE Workshop on information forensics and security, pp 1–6Google Scholar
  3. 3.
    Bayar B, Stamm MC (2016) A deep learning approach to universal image manipulation detection using a new convolutional layer. In: The 4th ACM workshop on information hiding and multimedia security, pp 5–10Google Scholar
  4. 4.
    Chen C, Shi Y-Q, Su W (2008) A machine learning based scheme for double JPEG compression detection. In: IEEE International conference on pattern recognition, pp 1–4Google Scholar
  5. 5.
    Girshick R, Donahue J, Darrell T, Malik J, Berkeley U (2014) Rich feature hierarchies for accurate object detection and semantic segmentation. In: IEEE conference on computer vision and pattern recognition, pp 580–587Google Scholar
  6. 6.
    He K, Sun J (2015) Convolutional neural networks at constrained time cost. In: IEEE conference on computer vision and pattern recognition, pp 5353–5360Google Scholar
  7. 7.
    He K, Zhang X, Ren S, Sun J (2016) Deep residual learning for image recognition. In: IEEE conference on computer vision and pattern recognition, pp 770–778Google Scholar
  8. 8.
    Huang F, Huang J, Shi Y-Q (2010) Detecting double JPEG compression with the same quantization matrix. IEEE Trans Inf Forens Secur 5(4):848–856CrossRefGoogle Scholar
  9. 9.
    Huang G, Sun Y, Liu Z, Sedra D, Weinberger KQ (2016) Deep networks with stochastic depth. In: European conference on computer vision, pp 646–661CrossRefGoogle Scholar
  10. 10.
    Huang G, Liu Z, Maaten LVD, Weiberger KQ (2017) Densely connected convolutional networks. In: IEEE Conference on computer vision and pattern recognitionGoogle Scholar
  11. 11.
    Ioffe S, Szegedy C (2015) Batch normalization: accelerating deep network training by reducing internal covariate shift. In: International conference on machine learning, pp 448–456Google Scholar
  12. 12.
    Krizhevsky A, Sutskever I, Hinton G (2012) ImageNet classification with deep convolutional neural networks. In: International conference on neural information processing systems, pp 1097–1105Google Scholar
  13. 13.
    Larsson G, Maire M, Shakhnarovich G (2016) FractalNet: ultra-deep neural networks without residuals, arXiv:1605.07648
  14. 14.
    LeCun Y, Boser BE, Denker JS, et al. (1990) Handwritten digit recognition with a back-propagation network. Adv Neural Inf Process Syst, 396–404Google Scholar
  15. 15.
    Li B, Luo H, Zhang H, Tan S, Ji Z (2017) A multi-branch convolutional neural network for detecting double JPEG compression, arXiv:1710.05477
  16. 16.
    Liu Q, Sung A-H, Qiao M (2011) A method to detect JPEG-based double compression. Int Symp Neural Netw LNCS 6676:466–476Google Scholar
  17. 17.
    Luo X, Song X, Li X, et al. (2016) Steganalysis of HUGO steganography based on parameter recognition of syndrome-trellis-codes. Multimed Tools Appl 75 (21):13557–13583CrossRefGoogle Scholar
  18. 18.
    Pevny T, Fridrich J (2008) Detection of double-compression in JPEG images for applications in steganography. IEEE Trans Inf Forens Secur 3(2):247–258CrossRefGoogle Scholar
  19. 19.
    Qu Z, Luo W, Huang J (2008) A convolutive mixing model for shifted double JPEG compression with application to passive image authentication. In: IEEE International conference on acousticsGoogle Scholar
  20. 20.
    Srivastava RK, Greff K, Schmidhuber J (2015) Highway networks, arXiv:1505.00387
  21. 21.
    Srivastava RK, Greff K, Schmidhuber J (2015) Training very deep networks. Adv Neural Inf Process Syst, 2377–2385Google Scholar
  22. 22.
    Verma V, Agarmal N, Khanna N (2018) DCT-domain deep convolutional neural networks for multiple JPEG compression classification. Signal Process-Image Commun 67:22–23CrossRefGoogle Scholar
  23. 23.
    Wang Q, Zhang R (2016) Double JPEG compression forensics based on a convolutional neural network. EURASIP J Inf Secur 2016:23CrossRefGoogle Scholar
  24. 24.
    Yang P, Ni R, Zhao Y (2018) Double JPEG compression detection by exploring the correlations in DCT domain, arXiv:1806.01571v1
  25. 25.
    Yao H, Song S, Qin C, Tang Z, Liu X (2017) Detection of double-compressed H.264/AVC video incorporating the features of the string of data bits and skip. Macroblocks 9(12):313Google Scholar
  26. 26.
    Yao H, Cao F, Tang Z, Wang J, Qiao T (2018) Expose noise level inconsistency incorporating the inhomogeneity scoring strategy. Multimed Tools Appl 77(14):18139–18161CrossRefGoogle Scholar
  27. 27.
    Zhang Y, Qin C, Zhang W, Liu F, Luo X (2018) On the fault-tolerant performance for a class of robust image steganography. Signal Process 146:99–111CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC, part of Springer Nature 2018

Authors and Affiliations

  1. 1.Shanghai Institute for Advanced Communication and Data ScienceShanghai UniversityShanghaiChina
  2. 2.School of Communication and Information EngineeringShanghai UniversityShanghaiChina

Personalised recommendations