Abstract

As a novel digital video steganography, the motion vector (MV)-based steganographic algorithm leverages the MVs as the information carriers to hide the secret messages. The existing steganalyzers based on the statistical characteristics of the spatial/frequency coefficients of the video frames cannot attack the MV-based steganography. In order to detect the presence of information hidden in the MVs of video streams, we design a novel MV recovery algorithm and propose the calibration distance histogram-based statistical features for steganalysis. The support vector machine (SVM) is trained with the proposed features and used as the steganalyzer. Experimental results demonstrate that the proposed steganalyzer can effectively detect the presence of hidden messages and outperform others by the significant improvements in detection accuracy even with low embedding rates.

© 2012 Optical Society of America

Full Article  |  PDF Article

References

  • View by:
  • |
  • |
  • |

  1. H. A. Aly, “Data hiding in motion vectors of compressed video based on their associated prediction error,” IEEE Trans.Inform. Forensics Security 6, 14–18 (2011).
    [CrossRef]
  2. Y. Cao, X. F. Zhao, and D. G. Feng, “Video steganalysis exploiting motion vector reversion-based features,” IEEE Signal Process. Lett. 19, 35–38 (2012).
    [CrossRef]
  3. C. Xu, X. Ping, and T. Zhang, “Steganography in compressed video stream,” in Proceedings of IEEE First International Conference on Innovative Computing, Information and Control (IEEE, 2006), pp. 269–272.
  4. X. He and Z. Luo, “A novel steganographic algorithm based on the motion vector phase,” in Proceedings of IEEE International Conference on Computer Science and Software Engineering (IEEE, 2008), pp. 822–825.
    [CrossRef]
  5. Y. Cao, X. Zhao, D. Feng, and R. Sheng, “Video steganography with perturbed motion estimation,” Lect. Notes Comput. Sci. 6958, 193–207 (2011).
    [CrossRef]
  6. Y. T. Su, C. Q. Zhang, and C. T. Zhang, “A video steganalytic algorithm against motion-vector-based steganography,” Signal Processing 91, 1901–1909 (2011).
    [CrossRef]
  7. J. Zheng and L. P. Chau, “A motion vector recovery algorithm for digital video using Lagrange interpolation,” IEEE Trans. Broadcasting 49, 383–389 (2003).
    [CrossRef]
  8. T. Wiegand, G. J. Sullivan, G. Bjontegaard, and A. Luthra, “Overview of the H.264/AVC video coding standard,” IEEE Trans. Circuits Syst. Video Technol. 13, 560–576(2003).
    [CrossRef]
  9. J. Harmsen and W. Pearlman, “Steganalysis of additive noise modelable information hiding,” Proc. SPIE 5020, 134–142 (2003).
    [CrossRef]
  10. H. Zheng and L. P. Chau, “Efficient motion vector recovery algorithm for H.264 based on a polynomial model,” IEEE Trans. Multimedia 7, 507–513 (2005).
    [CrossRef]
  11. H. Takeda, S. Farsiu, and P. Milanfar, “Kernel regression for image processing and reconstruction,” IEEE Trans. Image Process. 16, 349–366 (2007).
    [CrossRef]
  12. M. D. Buhmann, Radial Basis Functions: Theory and Implementations (Cambridge University, 2003).
  13. Y. K. Wang and M. M. Hannuksela, “The error concealment feature in the H.26L test model,” in Proceedings of IEEE International Conference on Image Processing (IEEE, 2002), pp. 729–732.
    [CrossRef]
  14. C. Lampert, M. Militzer, and P. Ross, “XviD MPEG4 core library,” http://www.xvid.org .
  15. Y. Wang and P. Moulin, “Optimized feature extraction for learning-based image steganalysis,” IEEE Trans. Inform. Forensics Security 2, 31–45 (2007).
    [CrossRef]
  16. C. J. C. Burges, “A tutorial on support vector machines for pattern recognition,” Data Mining and Knowledge Discovery 2, 121–167 (1998).
    [CrossRef]
  17. C. C. Chang and C. J. Lin, “LIBSVM: a library for support vector machines,” http://www.csie.ntu.edu.tw/~cjlin/libsvm .
  18. Y. Deng, Y. J. Wu, and L. N. Zhou, “Video steganalysis exploiting motion vector calibration-based features,” Adv. Mater. Res. (N.Y.) 482, 168–172 (2012).
    [CrossRef]

2012

Y. Cao, X. F. Zhao, and D. G. Feng, “Video steganalysis exploiting motion vector reversion-based features,” IEEE Signal Process. Lett. 19, 35–38 (2012).
[CrossRef]

Y. Deng, Y. J. Wu, and L. N. Zhou, “Video steganalysis exploiting motion vector calibration-based features,” Adv. Mater. Res. (N.Y.) 482, 168–172 (2012).
[CrossRef]

2011

Y. Cao, X. Zhao, D. Feng, and R. Sheng, “Video steganography with perturbed motion estimation,” Lect. Notes Comput. Sci. 6958, 193–207 (2011).
[CrossRef]

Y. T. Su, C. Q. Zhang, and C. T. Zhang, “A video steganalytic algorithm against motion-vector-based steganography,” Signal Processing 91, 1901–1909 (2011).
[CrossRef]

H. A. Aly, “Data hiding in motion vectors of compressed video based on their associated prediction error,” IEEE Trans.Inform. Forensics Security 6, 14–18 (2011).
[CrossRef]

2007

H. Takeda, S. Farsiu, and P. Milanfar, “Kernel regression for image processing and reconstruction,” IEEE Trans. Image Process. 16, 349–366 (2007).
[CrossRef]

Y. Wang and P. Moulin, “Optimized feature extraction for learning-based image steganalysis,” IEEE Trans. Inform. Forensics Security 2, 31–45 (2007).
[CrossRef]

2005

H. Zheng and L. P. Chau, “Efficient motion vector recovery algorithm for H.264 based on a polynomial model,” IEEE Trans. Multimedia 7, 507–513 (2005).
[CrossRef]

2003

J. Zheng and L. P. Chau, “A motion vector recovery algorithm for digital video using Lagrange interpolation,” IEEE Trans. Broadcasting 49, 383–389 (2003).
[CrossRef]

T. Wiegand, G. J. Sullivan, G. Bjontegaard, and A. Luthra, “Overview of the H.264/AVC video coding standard,” IEEE Trans. Circuits Syst. Video Technol. 13, 560–576(2003).
[CrossRef]

J. Harmsen and W. Pearlman, “Steganalysis of additive noise modelable information hiding,” Proc. SPIE 5020, 134–142 (2003).
[CrossRef]

1998

C. J. C. Burges, “A tutorial on support vector machines for pattern recognition,” Data Mining and Knowledge Discovery 2, 121–167 (1998).
[CrossRef]

Aly, H. A.

H. A. Aly, “Data hiding in motion vectors of compressed video based on their associated prediction error,” IEEE Trans.Inform. Forensics Security 6, 14–18 (2011).
[CrossRef]

Bjontegaard, G.

T. Wiegand, G. J. Sullivan, G. Bjontegaard, and A. Luthra, “Overview of the H.264/AVC video coding standard,” IEEE Trans. Circuits Syst. Video Technol. 13, 560–576(2003).
[CrossRef]

Buhmann, M. D.

M. D. Buhmann, Radial Basis Functions: Theory and Implementations (Cambridge University, 2003).

Burges, C. J. C.

C. J. C. Burges, “A tutorial on support vector machines for pattern recognition,” Data Mining and Knowledge Discovery 2, 121–167 (1998).
[CrossRef]

Cao, Y.

Y. Cao, X. F. Zhao, and D. G. Feng, “Video steganalysis exploiting motion vector reversion-based features,” IEEE Signal Process. Lett. 19, 35–38 (2012).
[CrossRef]

Y. Cao, X. Zhao, D. Feng, and R. Sheng, “Video steganography with perturbed motion estimation,” Lect. Notes Comput. Sci. 6958, 193–207 (2011).
[CrossRef]

Chang, C. C.

C. C. Chang and C. J. Lin, “LIBSVM: a library for support vector machines,” http://www.csie.ntu.edu.tw/~cjlin/libsvm .

Chau, L. P.

H. Zheng and L. P. Chau, “Efficient motion vector recovery algorithm for H.264 based on a polynomial model,” IEEE Trans. Multimedia 7, 507–513 (2005).
[CrossRef]

J. Zheng and L. P. Chau, “A motion vector recovery algorithm for digital video using Lagrange interpolation,” IEEE Trans. Broadcasting 49, 383–389 (2003).
[CrossRef]

Deng, Y.

Y. Deng, Y. J. Wu, and L. N. Zhou, “Video steganalysis exploiting motion vector calibration-based features,” Adv. Mater. Res. (N.Y.) 482, 168–172 (2012).
[CrossRef]

Farsiu, S.

H. Takeda, S. Farsiu, and P. Milanfar, “Kernel regression for image processing and reconstruction,” IEEE Trans. Image Process. 16, 349–366 (2007).
[CrossRef]

Feng, D.

Y. Cao, X. Zhao, D. Feng, and R. Sheng, “Video steganography with perturbed motion estimation,” Lect. Notes Comput. Sci. 6958, 193–207 (2011).
[CrossRef]

Feng, D. G.

Y. Cao, X. F. Zhao, and D. G. Feng, “Video steganalysis exploiting motion vector reversion-based features,” IEEE Signal Process. Lett. 19, 35–38 (2012).
[CrossRef]

Hannuksela, M. M.

Y. K. Wang and M. M. Hannuksela, “The error concealment feature in the H.26L test model,” in Proceedings of IEEE International Conference on Image Processing (IEEE, 2002), pp. 729–732.
[CrossRef]

Harmsen, J.

J. Harmsen and W. Pearlman, “Steganalysis of additive noise modelable information hiding,” Proc. SPIE 5020, 134–142 (2003).
[CrossRef]

He, X.

X. He and Z. Luo, “A novel steganographic algorithm based on the motion vector phase,” in Proceedings of IEEE International Conference on Computer Science and Software Engineering (IEEE, 2008), pp. 822–825.
[CrossRef]

Lampert, C.

C. Lampert, M. Militzer, and P. Ross, “XviD MPEG4 core library,” http://www.xvid.org .

Lin, C. J.

C. C. Chang and C. J. Lin, “LIBSVM: a library for support vector machines,” http://www.csie.ntu.edu.tw/~cjlin/libsvm .

Luo, Z.

X. He and Z. Luo, “A novel steganographic algorithm based on the motion vector phase,” in Proceedings of IEEE International Conference on Computer Science and Software Engineering (IEEE, 2008), pp. 822–825.
[CrossRef]

Luthra, A.

T. Wiegand, G. J. Sullivan, G. Bjontegaard, and A. Luthra, “Overview of the H.264/AVC video coding standard,” IEEE Trans. Circuits Syst. Video Technol. 13, 560–576(2003).
[CrossRef]

Milanfar, P.

H. Takeda, S. Farsiu, and P. Milanfar, “Kernel regression for image processing and reconstruction,” IEEE Trans. Image Process. 16, 349–366 (2007).
[CrossRef]

Militzer, M.

C. Lampert, M. Militzer, and P. Ross, “XviD MPEG4 core library,” http://www.xvid.org .

Moulin, P.

Y. Wang and P. Moulin, “Optimized feature extraction for learning-based image steganalysis,” IEEE Trans. Inform. Forensics Security 2, 31–45 (2007).
[CrossRef]

Pearlman, W.

J. Harmsen and W. Pearlman, “Steganalysis of additive noise modelable information hiding,” Proc. SPIE 5020, 134–142 (2003).
[CrossRef]

Ping, X.

C. Xu, X. Ping, and T. Zhang, “Steganography in compressed video stream,” in Proceedings of IEEE First International Conference on Innovative Computing, Information and Control (IEEE, 2006), pp. 269–272.

Ross, P.

C. Lampert, M. Militzer, and P. Ross, “XviD MPEG4 core library,” http://www.xvid.org .

Sheng, R.

Y. Cao, X. Zhao, D. Feng, and R. Sheng, “Video steganography with perturbed motion estimation,” Lect. Notes Comput. Sci. 6958, 193–207 (2011).
[CrossRef]

Su, Y. T.

Y. T. Su, C. Q. Zhang, and C. T. Zhang, “A video steganalytic algorithm against motion-vector-based steganography,” Signal Processing 91, 1901–1909 (2011).
[CrossRef]

Sullivan, G. J.

T. Wiegand, G. J. Sullivan, G. Bjontegaard, and A. Luthra, “Overview of the H.264/AVC video coding standard,” IEEE Trans. Circuits Syst. Video Technol. 13, 560–576(2003).
[CrossRef]

Takeda, H.

H. Takeda, S. Farsiu, and P. Milanfar, “Kernel regression for image processing and reconstruction,” IEEE Trans. Image Process. 16, 349–366 (2007).
[CrossRef]

Wang, Y.

Y. Wang and P. Moulin, “Optimized feature extraction for learning-based image steganalysis,” IEEE Trans. Inform. Forensics Security 2, 31–45 (2007).
[CrossRef]

Wang, Y. K.

Y. K. Wang and M. M. Hannuksela, “The error concealment feature in the H.26L test model,” in Proceedings of IEEE International Conference on Image Processing (IEEE, 2002), pp. 729–732.
[CrossRef]

Wiegand, T.

T. Wiegand, G. J. Sullivan, G. Bjontegaard, and A. Luthra, “Overview of the H.264/AVC video coding standard,” IEEE Trans. Circuits Syst. Video Technol. 13, 560–576(2003).
[CrossRef]

Wu, Y. J.

Y. Deng, Y. J. Wu, and L. N. Zhou, “Video steganalysis exploiting motion vector calibration-based features,” Adv. Mater. Res. (N.Y.) 482, 168–172 (2012).
[CrossRef]

Xu, C.

C. Xu, X. Ping, and T. Zhang, “Steganography in compressed video stream,” in Proceedings of IEEE First International Conference on Innovative Computing, Information and Control (IEEE, 2006), pp. 269–272.

Zhang, C. Q.

Y. T. Su, C. Q. Zhang, and C. T. Zhang, “A video steganalytic algorithm against motion-vector-based steganography,” Signal Processing 91, 1901–1909 (2011).
[CrossRef]

Zhang, C. T.

Y. T. Su, C. Q. Zhang, and C. T. Zhang, “A video steganalytic algorithm against motion-vector-based steganography,” Signal Processing 91, 1901–1909 (2011).
[CrossRef]

Zhang, T.

C. Xu, X. Ping, and T. Zhang, “Steganography in compressed video stream,” in Proceedings of IEEE First International Conference on Innovative Computing, Information and Control (IEEE, 2006), pp. 269–272.

Zhao, X.

Y. Cao, X. Zhao, D. Feng, and R. Sheng, “Video steganography with perturbed motion estimation,” Lect. Notes Comput. Sci. 6958, 193–207 (2011).
[CrossRef]

Zhao, X. F.

Y. Cao, X. F. Zhao, and D. G. Feng, “Video steganalysis exploiting motion vector reversion-based features,” IEEE Signal Process. Lett. 19, 35–38 (2012).
[CrossRef]

Zheng, H.

H. Zheng and L. P. Chau, “Efficient motion vector recovery algorithm for H.264 based on a polynomial model,” IEEE Trans. Multimedia 7, 507–513 (2005).
[CrossRef]

Zheng, J.

J. Zheng and L. P. Chau, “A motion vector recovery algorithm for digital video using Lagrange interpolation,” IEEE Trans. Broadcasting 49, 383–389 (2003).
[CrossRef]

Zhou, L. N.

Y. Deng, Y. J. Wu, and L. N. Zhou, “Video steganalysis exploiting motion vector calibration-based features,” Adv. Mater. Res. (N.Y.) 482, 168–172 (2012).
[CrossRef]

Adv. Mater. Res. (N.Y.)

Y. Deng, Y. J. Wu, and L. N. Zhou, “Video steganalysis exploiting motion vector calibration-based features,” Adv. Mater. Res. (N.Y.) 482, 168–172 (2012).
[CrossRef]

Data Mining and Knowledge Discovery

C. J. C. Burges, “A tutorial on support vector machines for pattern recognition,” Data Mining and Knowledge Discovery 2, 121–167 (1998).
[CrossRef]

IEEE Signal Process. Lett.

Y. Cao, X. F. Zhao, and D. G. Feng, “Video steganalysis exploiting motion vector reversion-based features,” IEEE Signal Process. Lett. 19, 35–38 (2012).
[CrossRef]

IEEE Trans. Broadcasting

J. Zheng and L. P. Chau, “A motion vector recovery algorithm for digital video using Lagrange interpolation,” IEEE Trans. Broadcasting 49, 383–389 (2003).
[CrossRef]

IEEE Trans. Circuits Syst. Video Technol.

T. Wiegand, G. J. Sullivan, G. Bjontegaard, and A. Luthra, “Overview of the H.264/AVC video coding standard,” IEEE Trans. Circuits Syst. Video Technol. 13, 560–576(2003).
[CrossRef]

IEEE Trans. Image Process.

H. Takeda, S. Farsiu, and P. Milanfar, “Kernel regression for image processing and reconstruction,” IEEE Trans. Image Process. 16, 349–366 (2007).
[CrossRef]

IEEE Trans. Inform. Forensics Security

Y. Wang and P. Moulin, “Optimized feature extraction for learning-based image steganalysis,” IEEE Trans. Inform. Forensics Security 2, 31–45 (2007).
[CrossRef]

IEEE Trans. Multimedia

H. Zheng and L. P. Chau, “Efficient motion vector recovery algorithm for H.264 based on a polynomial model,” IEEE Trans. Multimedia 7, 507–513 (2005).
[CrossRef]

IEEE Trans.Inform. Forensics Security

H. A. Aly, “Data hiding in motion vectors of compressed video based on their associated prediction error,” IEEE Trans.Inform. Forensics Security 6, 14–18 (2011).
[CrossRef]

Lect. Notes Comput. Sci.

Y. Cao, X. Zhao, D. Feng, and R. Sheng, “Video steganography with perturbed motion estimation,” Lect. Notes Comput. Sci. 6958, 193–207 (2011).
[CrossRef]

Proc. SPIE

J. Harmsen and W. Pearlman, “Steganalysis of additive noise modelable information hiding,” Proc. SPIE 5020, 134–142 (2003).
[CrossRef]

Signal Processing

Y. T. Su, C. Q. Zhang, and C. T. Zhang, “A video steganalytic algorithm against motion-vector-based steganography,” Signal Processing 91, 1901–1909 (2011).
[CrossRef]

Other

C. Xu, X. Ping, and T. Zhang, “Steganography in compressed video stream,” in Proceedings of IEEE First International Conference on Innovative Computing, Information and Control (IEEE, 2006), pp. 269–272.

X. He and Z. Luo, “A novel steganographic algorithm based on the motion vector phase,” in Proceedings of IEEE International Conference on Computer Science and Software Engineering (IEEE, 2008), pp. 822–825.
[CrossRef]

C. C. Chang and C. J. Lin, “LIBSVM: a library for support vector machines,” http://www.csie.ntu.edu.tw/~cjlin/libsvm .

M. D. Buhmann, Radial Basis Functions: Theory and Implementations (Cambridge University, 2003).

Y. K. Wang and M. M. Hannuksela, “The error concealment feature in the H.26L test model,” in Proceedings of IEEE International Conference on Image Processing (IEEE, 2002), pp. 729–732.
[CrossRef]

C. Lampert, M. Militzer, and P. Ross, “XviD MPEG4 core library,” http://www.xvid.org .

Cited By

OSA participates in CrossRef's Cited-By Linking service. Citing articles from OSA journals and other participating publishers are listed here.

Alert me when this article is cited.


Figures (8)

Fig. 1.
Fig. 1.

Video encoder block diagram.

Fig. 2.
Fig. 2.

Motion estimation.

Fig. 3.
Fig. 3.

3×3 neighborhood of MB C(i,j).

Fig. 4.
Fig. 4.

Change tendency of PSNR along different bitrates. (a) Forman; (b) Stefan; (c) Tempete, and (d) Mobile.

Fig. 5.
Fig. 5.

CD histograms. (a) Nonstego videos; (b) stego videos.

Fig. 6.
Fig. 6.

Scatter plot of the features for stego videos and the nonstego one.

Fig. 7.
Fig. 7.

ROC curves of steganalyzers using our proposed, Cao’s, and Zhang’s features.

Fig. 8.
Fig. 8.

histograms of car racing video sequence. (a) Nonstego car racing video; (b) the corresponding stego one.

Tables (5)

Tables Icon

Table 1. Corresponding Coordinates of Neighboring MVs

Tables Icon

Table 2. Description of the Video Dataset

Tables Icon

Table 3. Performance Comparison Between Zhang’s (T1), Cao’s (T2), and Our Proposed Features (T3)

Tables Icon

Table 4. Description of the High Temporal Activity Video Sequences

Tables Icon

Table 5. Performance Comparison between Zhang’s (T1), Cao’s (T2), and Our Proposed fFeatures (T3) on High Activity Temporal Videos (the Embedding Strength is 300 bpf)

Equations (20)

Equations on this page are rendered with MathJax. Learn more.

mvSh(i,j)=mvCh(i,j)+ηh(i,j),mvSv(i,j)=mvCv(i,j)+ηv(i,j),i=1,M/q,j=1,,N/q,
E(ηh)=0,E(ηv)=0.
Pm(x)=β0+β1xx0++βmxx0m,
β^=argminβt=1nytβ0β1xtx0β2xtx02βmxtx0m2=argminβt=1nytPm(xt)2,
β^=argminβt=1nK(xtx0h)ytβ0β1xtx0β2xtx02βmxtx0m2,
β˜=argminβt=1nK(xtx0h)ytβ0β1xtx0β2xtx022,
t=1nεt2=t=1nK(xtx0h)ytP2(xt)2=t=1nK(xtx0h)ytβ0β1xtx0β2xtx022=Q(β0,β1,β2)
Q(β0,β1,β2)=WXXXβ˜WXY2,
XX=[1x1x0x1x021x2x0x2x021xnx0xnx02]
Q(β0,β1,β2)=Aβ˜Z2.
Qβ˜=2AT(Aβ˜Z)=0.
β˜=(ATA)1ATZ.
β˜=(β0,β1,β2)T=(XXTWXTWXXX)1XXTWXTWXY=(XXTWX2XX)1XXTWX2Y,
y^0=β0=e1T(XXTWX2XX)1XXTWX2Y,
hCD(k)=|{xi|Dis(mvxi)=k}|N,(k=0,1,2,,μ;i=1,2,,N),
ΦCD(m)=k=0μhCD(k)exp{j2πkmη+1},0mη,
MCDn=m=0η|Φ(m)|sinn(πmη+1).
MCDn=m=0η/2|Φ(m)|sinn(πmη+1).
M^CDn=MCDnMCD0,n1.
fn=M^CDn,n1.

Metrics