Improved PPVObased highfidelity reversible data hiding u Source

Improved PPVO-based high-fidelity reversible data hiding u. Source : Signal Processing, vol. 167, February 2020 u. Authors : Haorui Wu, Xiaolong Li, Yao Zhao and Rongrong Ni u. Speaker : Chia-Shou Shih u. Date : 2019/11/07 1

Outline �Related work �Proposed method �Experimental results �Conclusions 2

Related work- Pixel-based pixel value ordering(1/3) Pixel-based pixel value ordering(PPVO) Ø RDH method Ø PPVO is an effective improvement of PVO: PPVO: [48]X. C. Qu, H. J. Kim, “Pixel-based pixel value ordering predictor for high-fidelity reversible data”, Signal Process, vol. 111, pp: 249 -260, 2015. 3

Related work- Pixel-based pixel value ordering(2/3) 9 9 10 11 11 9 10 11 10 9 10 11 254 254 253 254 254 10 10 11 10 10 10 X’=min(9, 10, 11) => X’=9 X’=max(9, 10, 11) => X’=11 skip X’=254 Skip X’=10 Skip 4

Related work- Pixel-based pixel value ordering(3/3) 9 9 10 11 11 9 10 11 10 9 10 11 254 254 253 254 254 10 10 11 10 10 10 X’=9 X”=x-b X’=11 X”=x-b skip X’=254 X”=x+b Skip X’=10 X“=x-b Skip 5

Proposed method PPVO: (e) Proposed, n = 18. (f) Proposed, n = 24. 6

Proposed method (t 1, t 2, t 3)=(40, 62, 86) LC: ( |149 -152|+|148 -149|+|148 -148|+|148 -147|+…+|147149| ) + ( |149 -150|+|148 -149|+…+|149 -147| ) =36 Local Complexity(LC) 135=142? 135 = 135 -0 7

Proposed method (t 1, t 2, t 3)=(40, 62, 86) 8

Experimental results Comparison of PSNR (d. B) among the proposed method and the methods of Qu et al. The embedding capacity is 10, 000 bits. Images [34]Qu et al. [11]Sachnev [24]Wang et al. [32]Peng et al. Proposed method PSNR (t 1, t 2, t 3) Lena 60. 36 58. 20 60. 12 60. 49 61. 12 (39, 45, 73) Baboon 54. 11 54. 14 55. 20 53. 58 54. 27 (337, 465, 10 99) Airplane 63. 76 60. 37 61. 94 62. 97 64. 06 (16, 17, 22) Barbara 60. 08 58. 16 60. 66 60. 48 60. 59 (17, 66, 79) Lake 59. 81 56. 68 57. 45 58. 81 60. 50 (54, 83, 112) Boat 58. 37 56. 13 57. 40 58. 26 58. 70 (78, 96, 144) Peppers 58. 73 55. 56 58. 03 58. 97 59. 32 (29, 92, 121) Elaine 58. 30 56. 12 56. 55 57. 37 59. 46 (82, 116, 142) Average 59. 19 56. 92 58. 41 58. 86 59. 76 [34]X. Qu, H. J. Kim “Pixel-based pixel value ordering predictor for high-fidelity reversible data hiding”Signal Process. , 111 (2015), pp. 249 -260 [11] V. Sachnev, H. J. Kim, J. Nam, S. Suresh, Y. Q. Shi “Reversible watermarking algorithm using sorting and prediction” IEEE Trans. Circuits Syst. Video Technol. , 19 (7) (2009), pp. 989 -999 [24 J. Wang, J. Ni, X. Zhang, Y. Q. Shi “Rate and distortion optimization for reversible data hiding using multiple histogram shifting” IEEE Trans. Cybern. , 47 (2) (2017), pp. 315 -326 [32] F. Peng, X. Li, B. Yang “Improved PVO-based reversible data hiding” Digital Signal Process. , 25 (2) (2014), pp. 255 -265 9

Experimental results Comparison of PSNR (d. B) among the proposed method and the methods of Qu et al. The embedding capacity is 20, 000 bits. Images [34]Qu et al. [11]Sachnev [24]Wang et al. [32]Peng et al. Proposed method PSNR (t 1, t 2, t 3) Lena 56. 65 55. 02 56. 58 56. 56 57. 18 (52, 102, 129) Airplane 60. 01 57. 30 59. 51 59. 07 60. 40 (22, 31, 32) Barbara 56. 28 55. 02 57. 11 56. 20 56. 74 (75, 94, 121) Lake 54. 71 52. 71 53. 14 53. 53 55. 34 (156, 193, 22 7) Boat 54. 19 52. 64 53. 59 53. 83 54. 24 (159, 191, 26 6) Peppers 55. 05 52. 29 54. 24 54. 77 55. 39 (96, 186) Elaine 53. 55 52. 00 52. 19 52. 65 54. 21 (165, 334, 45 2) Average 55. 78 53. 85 55. 19 55. 23 56. 22 10

Experimental results 11

Experimental results 12

Conclusions �improved PPVO-based RDH method �extended predictor �threshold 1

-END- 1